The Index Looks Fine

The breadth signal deserves precision because its author supplied it: 27.4% of S&P 500 stocks at one-month lows, the highest reading since March 27, and the weakness is not a large-cap phenomenon, with roughly 35% of both the S&P 400 and S&P 600 sitting at the same lows. September opened in typical fashion, all four major indices lower on the first session and one-month lows pushing to multi-month highs across every capitalization tier, and by Wednesday the small caps were up more than a percent again, the kind of whipsaw that the chart desks read as a possible failed breakout at the 161.8% extension rather than a resumption. Beneath it, industrials are on pace for their worst three-week stretch since March, with the bellwether machinery maker testing the neckline of a potential head-and-shoulders top at the exact level of its 200-day average, and the utilities sector, left behind by the AI trade, has fallen below two percent of the index for the first time on record. The cumulative advance-decline lines, all four of them, are below their 50-day averages; as the desk that flagged it put it, that is not the end of the world, but it is absolutely a clear sign of weakening breadth. Every advance-decline line rolled over.

SOURCE: Daily Chartbook DC Lite #674, via David Keller (Cumulative Advance-Decline Lines Below Their 50-Day Averages)

The positioning behind the calm is the part that concerns us. Retail investors' allocation to stocks ticked up to a fresh year-to-date high of 71.1% in August, the same month the active managers went all-in and non-dealer futures positioning approached records, and the index's monthly candle, for all that it closed at a record, printed a small body with a long upper wick, the shape of buyers who reached and were refused. The options market is testing the same pivot: the index and the one-month implied-correlation gauge are simultaneously at levels the technicians call decisive, with the short-term trend measures coiled against their volatility bands in the pattern that precedes directional moves rather than more drift. A market where the average stock is quietly breaking down beneath a serene index, with allocations at highs and correlations at lows, is the configuration this letter has been describing since the bifurcation section of #72; the difference this week is that the breadth composite has moved from warning to reading.

The sector table tells the same story from a different angle. Wednesday's bounce was led by basic materials, up nearly 1.7%, with real estate the laggard and the small caps outperforming, which reads as relief rather than repair: the groups that rallied are the ones the rate shock punished first, and a one-day reversal in the most oversold cohort is what the first stage of every breadth deterioration looks like before the second stage tests it. The volatility index fell almost seven percent to 15.2 on the same session, back toward the floor this letter has called the cheapest hedging window of the year, and the combination, oversold bounce plus a collapsing fear gauge into a two-sided week, is the exact configuration in which we prefer to own protection rather than sell it. The market is not broken; it is unevenly priced, and the price of insurance against unevenness is once again low.

Oil and Yields Move as One

The energy thread this letter promoted in #72 and extended twice since got its loudest week. Crude broke out of the triangle that had contained it since spring, clearing the downtrend line from the April highs, and the measured move from that pattern projects $120; the curation desk's own map runs $95 as the next level, $100 back in play on a reclaim, and, if the commodity ever starts to look like its sector, $120 and then the prior all-time highs near $146. Energy is up 46% this year, by far the strongest sector, and the sector fund closed at another all-time high Tuesday as tensions with Iran escalated again, the memorandum-is-dead reporting of last week now expressed in price. The exploration-and-production fund just posted its highest monthly close since 2015, the refiners are squeezing with short interest that hit record highs going into the summer's conflict, and the agriculture complex kept its streak going with the producers' fund at four-year highs and sugar breaking out. Central banks, meanwhile, are buying physical gold in record amounts even as the metal consolidates near $4,434 after its rejection at the cycle top's cost basis, which is exactly the tension our #75 trim assumed: the structural buyer remains, the tactical level failed, and we wait for the reclaim.

The chart that ties the commodity file to the bond file is the one this letter now treats as the regime's signature. The correlation between the ten-year yield and oil is the highest since the financial crisis, the correlation between the ten-year and Brent sits near multi-year highs, and the two have moved as a single trade for weeks: energy shock feeds headline inflation, inflation feeds the front end, the front end drags the long end, and the long end's rise makes the fiscal arithmetic worse, which is bullish for the real assets that hedge it. That loop is why our energy and agriculture longs and our rate-sensitive shorts have been two expressions of one position since June, and this week both expressions were paid on the same days for the same reason. Oil and yields move as one.

SOURCE: Daily Chartbook DC Lite #675 (The 10-Year Yield and Oil, Highest Correlation Since the GFC)

There is a policy dimension to the loop that the chair himself supplied. A central bank that measures underlying inflation by the share of the consumption basket with rising prices is a central bank whose gauge moves with energy and food before it moves with rents and services, and the week gave that gauge a crude breakout, an agriculture producers' fund at four-year highs, and a sugar market at fifty-two-week highs. That is not a forecast of the next inflation print, which the calendar delivers on the 10th and 11th; it is an observation about what the Fed's new analytic will see when it looks, and what it will see is breadth of rising prices widening at the commodity end of the basket just as the September meeting demands a decision. The commodity tape, as we wrote last week, is voting on September, and this week it voted louder.

The Hyperscalers Crowd the Long End

The bond selloff went global, and the numbers are historic in the plain sense. Yields are pushing to new highs around the world, with the exception of low-debt havens like Switzerland; the global bond market has reached a level not seen since 2008; the US bond market has now been in drawdown for more than six years, by far the longest in its history; the Treasury selloff widened the US-China ten-year spread to 312 basis points, near a record; and Japan's ten-year is rising on a combination of aggressive policy-rate expectations and a surging term premium, the world's most important creditor now demanding to be paid for duration at home. Tokyo's term premium is surging.

SOURCE: The Daily Shot Brief September 2 (Japan's 10-Year, Policy Expectations Plus a Surging Term Premium)

The new entrant in this letter's Treasury thread is the AI buildout itself. The five hyperscalers plus the chip leader have issued roughly $320 billion of debt in 2026 so far, including the special-purpose vehicles for which they stand as backstop obligors on data-center leases, and the chart desk's headline said the quiet part: they are issuing so much that they may be causing a supply-demand issue at the long end of the curve. Set that beside the Treasury's $1.4 trillion of placement over the next two quarters from #75, the buyback doubling on the 9th, and data-center construction spending that has now surpassed the historical peak of office construction, and the long end's problem stops being a fiscal story and becomes a national balance-sheet story: the government and the AI complex are competing for the same marginal buyer of duration at the same time, and the marginal buyer is raising the price. The honest counterpoint arrived in the same brief: yields are at multi-year highs, but the drawdowns remain far smaller than 2022's, because coupons are higher and the starting yields cushion the losses. The drawdown is not yet 2022.

SOURCE: The Daily Shot Brief September 2 (Global Bond Selloff, Yields at Multi-Year Highs but Drawdowns Below 2022)

The long bond fund is holding a key support level with monthly volatility extremely compressed, the technicians' way of saying a major move for bonds is close and that the direction has not yet been decided. The Treasury's fiscal engineering, the chair's hawkish-for-free posture, the hyperscalers' issuance, and Japan's term premium all press on the same instrument from different sides, and the compression will resolve in the same fortnight that holds jobs, PPI, CPI, and the meeting. The hyperscalers crowd the long end.

SOURCE: Daily Chartbook DC Lite #675 (Hyperscalers and Nvidia, $320 Billion of Debt Issued in 2026)

The bond market's international echo deserves one more sentence, because the pattern is the same everywhere and the exceptions prove the rule. Yields are rising in every major market except the ones with the least debt, which means the selloff is not a growth story or an inflation story alone but a creditor story: the buyers of sovereign duration are repricing the fiscal risk of the borrowers, and the borrowers with the most to refinance, the United States and Japan chief among them, are paying the widest premiums. A US-China ten-year spread near a record is what it looks like when the world's largest debtor and one of its largest surplus economies are priced as opposite ends of the same trade, and it is the single most bearish chart in the week's reading for the idea that foreign demand will quietly absorb what the Treasury and the hyperscalers need to place.

The AI File Turns Double-Edged

Every fact in this week's AI file cuts both ways, which is why the book holds the trade with both hands rather than one. On the bull side: the long-term earnings growth estimate for the index is at its highest since at least 1995, which is why the growth-adjusted multiple reads cheap even as the forward multiple reads rich; cybersecurity broke out against semiconductors to multi-month highs and above its relative 200-day average, the curation desk nominating it as the next big AI trade as threats scale with the technology; and data-center construction has outrun the last cycle's office boom. On the bear side: 80% of one brief's readers believe AI stocks are in a bubble, with a growing share expecting the bubble to persist for a year or more, which is the definition of a crowd that knows and stays; the LLM token-expenditure index put in a new low rather than a bottom, with the open-model index overtaking the closed one for the year; and the concentration statistic that stopped us cold, 80% of the two frontier labs' enterprise revenue comes from 1% of their customers, a level of customer concentration the analysts who track it have not seen in any other software category, and it is not improving. One percent of customers pay eighty.

SOURCE: Daily Chartbook DC Lite #675 (AI Customer Concentration, 80% of Enterprise Revenue From 1% of Customers)

The last of the mega-cap semiconductor names reported after Wednesday's close; its guidance is the first issued by a leader into the post-keynote rate regime, its reaction lands into today's tape, and #77 will grade it with the print in hand rather than from the wire. What we can say already is that the semis-to-software rotation of #74 has a third participant, and the cybersecurity breakout against chips is the market telling us where the next leg of the AI trade's leadership is being decided: not in the enablers, and increasingly not in the models, but in the layers that sell to everyone the models threaten.

Buy the Mid-Cycle Hike?

The single most useful frame for the coinflip meeting came from the chart desk's history of first hikes: when the curve has not inverted, the index hates the hike for one to three months and makes new highs within six, with the exception of 2022, when the curve had inverted first. The curve today is flatter than it was a month ago but not inverted, and the desk's conclusion, no inversion means buy the mid-cycle hike, is the kind of statistical claim this letter files as a scenario rather than a forecast, because the sample is small and the fiscal backdrop is unlike any in it. What it does supply is a written expectation: if the chair hikes on the 16th, the base case from history is a one-to-three-month drawdown, not a bear market, unless the long end refuses to cooperate, which is precisely the trap from #75. Buy the mid-cycle hike.

SOURCE: Daily Chartbook DC Lite #674 (The S&P 500 Around First Hikes, No Inversion Means Buy)

The labor data walked into the jobs report tight, as it has all summer. July job openings rose to 7.27 million against a downward revision to June, hires, quits, and layoffs all ticked lower, and the ratio of vacancies to unemployed workers rose to its highest since January 2025, a labor market with fewer people moving and more positions unfilled. Consumer credit expanded by more than expected while mortgage lending moderated sharply, the rate-sensitive half of the economy slowing exactly where our structural shorts expect while the borrowing half accelerates. The dollar index continues to hold its series of higher lows dating to the start of the year, the falsifier we called warming in #75 still warming. And the chair's own inflation gauge, the breadth of rising prices, gets its next input from a crude breakout and an agriculture streak. Openings rose into the report.

SOURCE: Daily Chartbook DC Lite #674 (July JOLTS, Vacancies to Unemployed Highest Since January 2025)

A word on the two central bank decisions that landed midweek, because they extend the global thread from #74. Canada and New Zealand both decided policy Wednesday, in the same week Korea and the Philippines had already hiked preemptively, and the pattern across the developed world is the one the chair described without committing to: central banks that would rather sound firm than move, watching term premiums rise regardless. The domestic picture that will matter most in the morning is the split the credit data revealed, borrowing accelerating on the consumer side while mortgage lending slows sharply, because it is exactly the mix that produces a firm jobs number with a weakening rate-sensitive economy underneath, the mix our structural shorts were built to monetize.

Tomorrow at 8:30

The setup for the payroll report is a study in asymmetry, and we write the responses now. Consensus sits at 55 thousand jobs with unemployment at 4.1%, against a July print of minus 23 thousand that single-handedly erased last cycle's hike pricing. A strong print, anything comfortably above consensus with firm wages, pushes the September odds from coinflip toward a hike, sends the two-year through the 4.4% tripwire this letter adopted last week, and tests the long end's year-to-date highs, which pays our rate-sensitive shorts and opens the one-to-three-month window of equity indigestion the first-hike history describes. A weak print, a repeat of July, validates the pretend-hawk thesis, rallies the front end, and asks the question that matters: does the long end rally with it, or does it read a cut into 3.4% inflation as the other side of the trap and sell anyway. Our positioning is built for the second question to have the second answer, and it is the one scenario in which the hard-asset pillar and the rate shorts both earn their keep at once.

The rest of the calendar is dense enough to list plainly: the buyback doubling on the 9th, PPI on the 10th, CPI on the 11th, the stablecoin bill's Senate vote on the 15th, and the meeting on the 15th and 16th. Between them sit the Beige Book already released, the central bank decisions in Canada and New Zealand already taken, and a Fed governor who tends to move markets speaking today. The tape that meets all of it is coiled, correlated at a pivot, and quietly weaker than its index, with allocations at highs. The desk framing from #75 holds: earnings season was one-sided, the macro calendar is two-sided, and the first of its answers arrives tomorrow morning.

Charts Worth Watching

Six curations from The Chart Report's Tuesday and Wednesday desks, each with its source below it: the four advance-decline lines under their averages, the exploration-and-production fund's best month since 2015, the ten-year and Brent moving together, the longest bond drawdown in history, the long bond coiled at support, and the central banks' record gold bid.

SOURCE: The Chart Report September 1, via David Keller (All Four Advance-Decline Lines Below Their 50-Day Averages)

SOURCE: The Chart Report September 1, via Scott Brown (Oil and Gas E&P, Highest Monthly Close Since 2015)

SOURCE: The Chart Report September 1, via Kevin Gordon (The 10-Year Yield and Brent, Correlation Near Multi-Year Highs)

SOURCE: The Chart Report September 1, via Charlie Bilello (The Longest Bond Drawdown in History, Six Years and Counting)

SOURCE: The Chart Report September 1, via Steve Strazza (TLT at Support With Volatility Compressed, a Major Move Ahead)

SOURCE: The Chart Report September 2, via Bravos Research (Central Banks Buying Physical Gold in Record Amounts)

The Verdict

Grade the composite first. The one-month-low count that #70 installed and #72 promoted to formal watch status has fired: 27.4% of the index at one-month lows under a record close, all four advance-decline lines below trend, small caps attempting a failed breakout, industrials at their worst stretch since March. The pre-written response is the one we execute: hedges, already rolled at the year's cheapest levels, stay on through the jobs print and the meeting; the small-cap expression of the broadening longs is trimmed to core size while the equal-weight expressions hold, because the breadth deterioration is concentrated in the cohort that failed at the extension; and no new equity risk is added into a two-sided fortnight with correlations at a pivot.

The commodity and rate books are extended on their own confirmations. Crude's breakout puts $95 on the tripwire list and $120 on the measured-move list; the energy and agriculture legs of the hard-asset pillar are extended, the gold leg stays trimmed until the anchored VWAP is reclaimed, and the central banks' record buying is filed as the reason the trim is a trim and not an exit. The rate-sensitive shorts add a thesis: the hyperscalers' $320 billion of issuance is now a named competitor for the long end's marginal buyer, alongside the Treasury's $1.4 trillion and Japan's own term premium, and the long bond's compressed volatility says the resolution is near. The semis add-back stays at one tranche, with cybersecurity's relative breakout against chips noted as the rotation's next chapter and the mega-cap report's reaction to be graded in #77. Tomorrow at 8:30 the coinflip gets its first weight, and every response above was written before it landed.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

Twenty Minutes, Thirteen Basis Points

The speech itself was a masterpiece of saying everything and promising nothing. The chair reaffirmed two percent as a firm, fixed target, warned that price stability does not execute itself, demanded confidence that inflation is returning clearly and at sufficient speed, and, in the philosophical core of the address, criticized the Fed's habit of forward guidance itself, arguing policymakers should preserve flexibility rather than pre-commit to rate paths. Markets heard it as decidedly hawkish: September went from about a third of a hike priced to sixty percent by one wire count at the close, an intraday sixty-one on the terminal models, and a 51-49 split on the prediction markets, which is to say the most important meeting of the year is now a genuine coinflip. The two-year did the heavy lifting, rising 13 basis points to 4.36% and pulling further away from a funds rate that sits at 3.50 to 3.75, and the curve desk's chart of the week shows what that did to the structure: the same official who read the July data dovishly at his press conference put a hawkish spin on identical data at Jackson Hole, and in doing so unwound an entire month of curve steepening in a single session. The curve heard the message.

SOURCE: Daily Chartbook DC Lite #672, Haver Analytics via Robin Brooks (The Curve Steepening, Unwound in a Speech)

What makes this chair analytically interesting rather than merely loud is that he grades inflation his own way, by watching the shares of the consumption basket with rising versus falling prices rather than any single index, and by that gauge the picture explains the tone: the breadth of rising prices has not narrowed the way a victory would require. The long end, meanwhile, remains conspicuously unimpressed, giving the hawkish speech of the year two basis points, from 5.19 to 5.22, before Monday's session sent the ten-year breaking out to new year-to-date highs. The front end bought the speech; the long bond gave it pocket change, and then asked for more yield anyway. The chair grades his own inflation.

SOURCE: The Daily Shot Brief August 31 (Warsh's Own Gauge, Shares of the PCE Basket)

The record behind the rhetoric is why the bond market demands proof rather than prose. This chair has presided over two meetings and delivered two holds, and he has held through the entire reacceleration, from 2.4% inflation in January to 3.4% now, which is not yet the resume of the inflation fighter the market thought it was hiring. The deeper history cuts the same way: in late 2018, with stocks fifteen percent off their highs, he publicly urged the Fed to stop tightening, and the Fed paused within weeks; in July, a single bad payroll print of minus 23 thousand was enough to erase the market's hike pricing while three members of his own committee argued for a hike he declined to join. Hawkish speeches cost nothing. The three-week window that starts this Friday is where words meet acts, and the desk that coined the phrase built a scorecard for exactly that comparison.

The Economy That Forces the Question

The tape behind the speech is why the hawkishness costs nothing to voice. Second-quarter GDP was revised up to a 3.3% annualized rate, led by exports, consumer spending, and business investment, the exact composition a soft-landing believer would order; industrial production edged up for a fourth consecutive month against expectations of a decline; the Fed's preferred inflation gauge remains above target; and the trimmed-core measures are accelerating everywhere outside the euro area. Our two-clocks framework has scored the growth clock ahead of the policy clock since #67, and this week the gap closed from the other side: the policy clock finally started moving toward the growth clock, exactly as the market's own front end has been insisting it must. The economy reaccelerated into the speech.

SOURCE: The Daily Shot Brief August 31 (Second-Quarter GDP Revised to 3.3%)

The analog the technicians reached for is the one that matters. The two-year's breakout came from a falling wedge and cleared the level that has capped it repeatedly for two years, and the divergence between the two-year and the funds rate now resembles late 2021, the last time the bond market got this far ahead of the Fed, months before the 2022 tightening cycle proved it right. The desk's line is falsifiable in the house style: a two-year breach of 4.4% would say the bond market is once again front-running a policy shift. That is not our forecast to make; it is our tripwire to watch, and it sits two basis points above Friday's close.

The corporate economy the Fed must referee reported its own numbers into the same window, and they explain the growth clock's stubbornness. The chip leader's official quarter: revenue of $96.2 billion up 106% from a year ago, operating income up 124%, data-center revenue of $89 billion with the next platform already in full production, and a management line that will outlive the cycle: compute is revenue now. The hyperscalers' AI cloud spending more than doubled, the enterprise software major crossed four billion dollars of AI-specific revenue with its agent platform's recurring revenue up more than 210%, and the cybersecurity firms are reporting triple-digit growth in model usage across their customer bases. An economy printing 3.3% GDP with that engine running underneath it is not an economy that argues for cuts, whatever the long end thinks of the financing.

Extremes on Both Sides

The positioning picture entering September is the strangest of the year, because the extremes are now on both sides simultaneously. Non-dealer equity futures positioning has risen sharply for weeks and is approaching record territory, the crowd leaning long with size. Positioning approaches record territory.

SOURCE: Daily Chartbook DC Lite #673, Goldman Sachs via Robbie Stankard (Non-Dealer Positioning Near Records)

At the same moment, short interest on the median S&P 500 stock has jumped to decade highs, the crowd leaning short with size, one line below the surface. Add the survey extremes from the last two issues, the active managers all-in at 102.7 while nearly half of the global fund-manager survey expects stagflation over the next twelve months even as boom optimism runs at its highest since 2021, and the market's brain is split cleanly in half: maximum conviction at the index level, maximum skepticism at the median-stock level, stagflation and boom in equal measure. Implied one-month correlations in the single digits, the lowest ever recorded exiting a second-quarter earnings season, are the options market's way of pricing that split, every stock on its own island, and the desk that flagged them attached the warning that matters: a market priced for zero correlation is a market vulnerable to the one kind of event that correlates everything, a macro shock. The calendar supplies candidates weekly through the 16th. The shorts crowd the median stock.

SOURCE: Daily Chartbook DC Lite #673, Goldman Sachs via @mikezaccardi (Median-Stock Short Interest at Decade Highs)

The tape's own posture into that calendar is serene to the point of provocation: the index has now spent one hundred consecutive sessions within five percent of an all-time high, September is the only month with a negative average return across the last forty years, the industrial sector, historically the index's highest-correlation companion, has quietly broken below its former highs, and momentum relative to the index has pulled back to a multi-year polarity level. None of these is a signal alone; together they describe a market that has stopped paying anyone for caution precisely when caution's price is lowest.

The Trap, Written Down

The macro essay of the week came from the Turkish desk in our inbox, and it deserves its own section because it converts this letter's Treasury thread into a scorecard with dates. The thesis names the chair a pretend hawk, and the reasoning is the trap: if the Fed hikes, it loses the long end, because the interest bill already runs near a trillion dollars a year, more than defense, a third of the marketable debt rolls inside a year, and every auction reprices the stock; if the Fed cuts into 3.4% inflation, it loses the long end anyway, because the marginal buyer of the next trillion dollars of duration demands more yield to fund a Treasury that must place $1.4 trillion over the next two quarters. Sounding hawkish is free; doing something, in either direction, sends the bill to the long end. The author's discipline is the part this letter adopts with admiration: he attached a date to his own name-calling, writing that a hike on September 16 makes the word look stupid, and a view should have a date on it. Our own book does not require the hike question to resolve either way, because the shorts live where both answers converge, at the long end.

The chronology thread from #74 extended on schedule. The buyback doubling was confirmed for September 9 with a $950 billion checking account standing behind it; the Treasury Secretary went on the record that the Fed's foreign repo facility is an important backstop that should be upsized in the coming months, and Tokyo has already said it plans to use that facility rather than sell its Treasuries; the chair, for his part, has a framework and balance-sheet review due by year-end and has spoken about updating the 1951 Treasury-Fed Accord, the document that once separated the two institutions this administration is busy reconnecting. And beneath all of it, the United States is once more speeding into the debt ceiling, except this time with a debt-to-GDP ratio twenty-five points higher than the last standoff, which is why the chart desk's warning read less like analysis than prophecy: do not be surprised by increasingly unconventional solutions. Between the CLARITY Act's Senate vote on the 15th, the FOMC on the 15th and 16th, and the buyback doubling on the 9th, the entire manufactured-demand architecture this letter has tracked since #73 has a September calendar. Speeding into the ceiling again.

SOURCE: Daily Chartbook DC Lite #672, Bloomberg via Simon White (US Debt Pressures Are Becoming More Acute)

Hard Assets Meet Their Falsifiers

We put three falsifiers under the hard-asset pillar in #73 and promised to grade them weekly; this was the week they earned their keep. Gold failed the test: down more than three percent on the speech, down to $4,499 by Monday from a $4,709 peak the prior Monday, and, in the chart of the week for our purposes, rejected precisely at the anchored VWAP from its all-time high, the average cost basis of every buyer from the cycle top, the exact level whose reclaim would have confirmed the advance. The dollar falsifier is warming at the worst possible moment: the dollar index sits at the lower boundary of its long-term channel with a bullish crossover forming in monthly momentum, a reversal setup arriving, in the curation desk's words, at a particularly inconvenient time for stocks. And the real-yield falsifier fired at the long end, with the ten-year breaking to new year-to-date highs. Three falsifiers, three warnings, and the discipline we pre-committed to in #73 now executes: the gold-adjacent leg of the pillar gets trimmed, not because the thesis died but because the levels we named in advance said trim.

The rest of the complex, however, refused to confirm the retreat, and honesty requires both halves. Energy closed August at the highest monthly level in its history, completing the promotion this letter gave it in #72 with a giant multi-decade cup pattern behind it. Agriculture posted ten consecutive daily gains, a streak not seen since 2009 and matched only eight times in more than a century: corn broke a multi-year rectangle to its best levels since 2023, the agriculture fund closed at decade highs, palladium broke out, silver sits at its own inflection against gold, and the fertilizer complex is attempting to reverse a four-year downtrend. Ten green days in the grains.

SOURCE: Daily Chartbook DC Lite #672, Turning Point via Dean Christians (Ten Consecutive Gains in Agriculture, First Since 2009)

The agriculture streak matters beyond the commodity book, because food is the part of the price level that walks straight into the chair's own gauge. A ten-day run in the broad agriculture index, corn at three-year highs, wheat at multi-year highs with speculators still short from #74, and fertilizer turning all describe input costs accelerating at the exact moment the Fed's new leadership measures inflation by how many categories are rising rather than how fast the average moves. The harder the grains run, the wider the rising share of the chair's basket, and the more his own analytic argues against the patience his committee has shown. The commodity tape, in other words, is now voting on September alongside the payroll number.

Crypto held its own test open: Bitcoin spent the whole violent week pinned near $79,000, still battling the anchored VWAP from its all-time high, the confirmation level #74 wrote down, while the Coinbase premium turned positive, the signature of US institutional demand, and the Solana funds recorded their strongest inflows on record. And the chart that reconciles the week's contradictions is the twenty-year correlation between commodities and the ten-year yield: hard assets and long yields rising together is not a paradox, it is the signature of the regime this letter has described since June, where the price of money and the price of things rise for the same fiscal reasons. The pillar survives the week smaller, more selective, and better tested: energy and agriculture lead, crypto awaits its level, and the metals wait out the chair.

September's Two-Sided Calendar

The desk framing that organizes the month ahead is simple: earnings season was a one-way catalyst set, an AI tailwind that blew in a single direction, while the macro calendar now presents genuinely two-sided risk, and the options market agrees, pricing meaningful implied moves for jobs Friday, PPI on the 10th, CPI on the 11th, and the Fed on the 15th and 16th. Payrolls consensus sits at 55 thousand with unemployment at 4.1%, against a July print of minus 23 thousand, the number that single-handedly removed last cycle's hike pricing; a repeat validates the pretend-hawk thesis, a rebound arms the hawks. The week also brings the last of the mega-cap reporters Wednesday, the Fed's leading intellectual voice on Thursday, the hawkish vice chair Tuesday, the Beige Book, and two foreign central bank decisions, and the benchmark payroll revision that shared the keynote's hour passed without a single headline in any of our sources' coverage, drowned by the chair, a silence this letter notes rather than interprets.

Two more files stay on the calendar's margin. The Canada rupture from #73 escalated rather than resolved, with the trade war now formally worsening into September and the countertariff date approaching, the tariff-margin experiment running live against America's largest goods supplier. And the hardware conference season previewed the next generation of the buildout, the chip leader's full-stack platform in production, a rival architecture pitching an AGI-class processor, and the second-source vendor unveiling its own rack-scale answer, which together frame Wednesday's report from the last of the mega-cap semiconductor names: the market will read it not as one company's quarter but as the first guidance issued into the post-keynote rate regime.

Two files ride into the month with momentum of their own. The software regime change of #74 got its follow-through signal: the sector fund flashed a bullish momentum thrust of the kind that has marked major bottoms since the start of the decade, and equal-weight technology broke out against cap-weight after nearly a year of quiet outperformance, confirmation that the breakout was breadth rather than a handful of names. And the AI file added its most double-edged datum yet: the hyperscalers' other income, largely paper gains on AI investments, now accounts for 54% of their pre-tax profit growth, at the same time the frontier labs' revenue grows at better than triple per year from a combined nine-figure base and the chip leader guides toward a 2028 revenue number that would have been the fourth-largest company on earth a decade ago. Compute is revenue now, as the chief executive put it, and an increasing share of the profit that justifies the multiple is the ecosystem marking up its own holdings. Both facts are true; the book holds them both. The profits are increasingly paper.

SOURCE: The Daily Shot Brief August 31 (Other Income, 54% of Hyperscaler Profit Growth)

Charts Worth Watching

Six curations from The Chart Report's Friday and Monday desks, each with its source below it: the two-year's breakout and its 2021 analog, the ten-year at new highs for the year, the gold rejection at the cycle top's cost basis, the record August in energy, the software momentum thrust, and the fund managers' split brain.

SOURCE: The Chart Report August 28, via Fusion Point Capital (The Two-Year Breaks Out, a Late-2021 Analog)

SOURCE: The Chart Report August 31, via Coiled Spring Capital (The 10-Year at New Year-to-Date Highs)

SOURCE: The Chart Report August 28, via David Rath (Gold Rejected at the Anchored VWAP From Its High)

SOURCE: The Chart Report August 31, via Alfonso De Pablos (Energy Closes the Best August in Its History)

SOURCE: The Chart Report August 28, via MacroCharts (Software Momentum Thrust, the Signature of Major Bottoms)

SOURCE: The Chart Report August 31, via Hector Chamizo (Half of Fund Managers Expect Stagflation)

The Verdict

Grade the pre-written asymmetry first. #74 said that if the chair declined to give the long end a framework, the move would begin where our shorts live, and that is what printed: he declined to pre-commit, the front end repriced toward hikes, and by Monday the ten-year had broken to new year-to-date highs while the thirty-year sat at 5.22 within reach of its own. The rate-sensitive shorts were paid for their thesis, the yen's break through 160 confirmed that the intervention regime remains reactive rather than in control, and the trap essay gives the position its cleanest statement yet: both of the Fed's exits price our way at the long end. We keep the shorts, we keep the September 9, 15, and 16 dates circled, and we adopt the two-year's 4.4% level as the front-end tripwire that would say the market has moved from pricing a hike to forcing one.

The hard-asset pillar takes its first real loss and its first real discipline: the gold-adjacent leg is trimmed per the pre-written falsifiers, the energy and agriculture legs are extended on their own confirmations, and crypto remains sized for a test not yet passed. The equity book leans into the split: our broadening longs against the median-stock short base, our software exposure with the momentum thrust at its back, and hedges rolled while one hundred days inside five percent of the highs keeps them the cheapest in a year. The one honest anxiety is the one the correlation desk named: a market priced for zero correlation meeting a two-sided macro calendar is a market where a single Friday morning can correlate everything at once, and this Friday morning has a payroll number in it. #76 grades the jobs print against the coinflip, with every response, as always, written down before the answer arrives.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

Software Was Supposed to Die

Start with the corpse that sat up. For most of this year the consensus story called it the SaaSpocalypse: artificial intelligence would eat the software industry, agents would replace seats, and the sector traded accordingly, crushed for a year while the hardware that trains the models made records. Thursday the equal-weight software index closed at an all-time high, taking out a peak that had stood since December 2024, and with a few trading days left in August the index is working on the highest monthly close in its history. The software index rewrote its history.

SOURCE: Daily Chartbook DC Lite #671, TrendLabs via @jc_parets (Equal-Weight Software XSW, Record Monthly Close)

The detail that turns a rally into a thesis is who was positioned for it. The flow data shows retail spent August buying semiconductors and selling software, the exact wrong side of the month's decisive relative move, and the ratio of chips to software has now broken to fresh multi-month lows even after Thursday's semiconductor bounce. The accounting software major that led the S&P with a twenty-two percent single-day gain did the heavy lifting, the enterprise names followed, and the cybersecurity complex had its third-best day in the history of its sector fund. This is the dispersion regime of #72 operating at full symmetry: Monday the index fell while eight sectors rose, Thursday the index rose while ten sectors fell, and in both cases the average stock told you nothing about where the money actually went. Our broadening longs hold the software side of the spread by construction, and the position we spent six issues defending against the concentration trade just collected the repricing. The crowd, meanwhile, was positioned against the emerging breakout.

SOURCE: Daily Chartbook DC Lite #671, JPMorgan (Retail Bought Semis and Sold Software in August)

The after-hours session then demonstrated, inside a single evening, why this sector rewards selection and punishes indexing even on its best day in years. The data-analytics name surged fifteen percent on raised guidance and the enterprise-workflow major added two while the apparel retailer jumped seven on a leadership change; meanwhile the backup-security firm fell ten percent despite beating estimates and raising its own guidance, the design-software maker slid six on disappointing projections, and the endpoint-security competitor shed seven on a weak outlook. Three winners and three losers, all reporting within hours of each other, in the sector that had just made history. The dispersion lens from #72 said consumer discretionary could not move as a bloc; software just proved it belongs on the same list, and our expressions there remain single-name and selective rather than sector-wide for precisely this reason.

Nvidia Clears the Bar

The print itself deserves a clean grading because we pre-wrote the conditions. The company beat expectations that assumed a doubling, guided in a way one desk summarized as proof the AI wave is far from over as demand continues to expand, and rose 8.7% for its best session since the April 2025 rebound, snapping a streak that had reached eight down days in nine. The de-grossing thesis from #73 performed exactly as the mechanics suggested: a tape that had already surrendered its leverage had no sellers left to meet good news, and the weakest hands the prime-brokerage data showed leaving at the fastest pace in two years spent Thursday buying back what they sold. The most substantive new fact in the report, per the research desks, is that the AI cycle is constrained by physical bottlenecks, memory and power, rather than by any shortage of end demand, which pushes the cyclical question further out and, in one analyst's phrase, gives investors permission to extend the earnings-growth runway across the second- and third-order beneficiaries of the buildout. That is the memory thesis of #68 and the power thesis of the summer restated as the bull case's official architecture, and it arrived alongside a report that the company is nearing a thirteen-billion-dollar acquisition of the industry's central model repository, vertical integration moving down the stack.

Honesty files the counterweights in the same paragraph. The core enabler still trails the broader constellation of global AI winners across US, Asian, and European indices, a reminder that the market has spent months paying up for everything around the pick-and-shovel trade; one strategist warned that the smaller AI names are just riding the leader's coattails with high volatility and low correlation producing an index calm that is a false sense of security; and the average S&P stock, in another desk's phrase, looks a little exhausted. Our add-back conditions from #71 now read as follows: the flush arrived in #73, the catalyst passed with a strong print into de-risked positioning, and the intermediate-term reclaim is in progress but not complete, with buyers back in control of the five-day trends while the 50-day averages overhead remain the test. We add back the first tranche of the semis trim into that structure and stage the rest behind the reclaim, exactly as written.

All-In at the Lows of Fear

The tape's most uncomfortable statistic this week is not a price, it is a posture. Implied volatility printed its lowest level of 2026 while the index sits within one percent of an all-time closing high; the VIX at 14.5 sits at the bottom of its entire post-crisis normal range, implying daily moves under a percent for the next month; and the volatility of volatility just made its own 2026 low. On its own, as the desk that flagged it concedes, that is not a signal, because low volatility can persist for months. What makes it a condition worth flagging is the calendar it is compressing into: August jobs, PPI, CPI, and the September FOMC decision land inside the next two and a half weeks, four events with the market carrying, in the desk's phrase, almost no premium for being wrong.

Set the positioning beside the pricing and the asymmetry sharpens. The active-manager exposure index climbed to 102.7, its highest reading in over two years, a level that implies leverage, the same cohort that was record short Nasdaq futures in #70 now, in the survey's terms, all-in. The contrarian ledger this letter audited in #72 has completed its round trip: the record shorts covered, the insiders who bought the summer lows are sitting on the gains, and the fuel that powered the melt-up is now the exposure that has to be defended. The one offsetting flow is the dealers' book, where rising gamma would normally prompt managers to increase exposure and they remain underexposed by that measure, a residual bid under the market if nothing breaks. We do not trade volatility directly, but the book's posture obeys the regime: the hedges we roll are cheap precisely when no one wants them, and this is the cheapest they have been all year. The managers are all-in.

SOURCE: Daily Chartbook DC Lite #671, NAAIM (Exposure Index at 102.7, Highest in Over Two Years)

The mirror image of Thursday's tech show is where the money came from, and it completes the rotation's symmetry. Consumer staples finished at the bottom of the sector table, dragged by the packaged-food maker that fell ten percent after earnings to rank as the day's worst S&P performer; healthcare lagged again as the vaccine maker extended its slide; and the retail cohort stayed heavy, with the electronics chain and the off-price apparel name both down hard on disappointing reactions and the retail sector fund off nearly two percent. The defensive bid that led Monday was the source of funds by Thursday, four sessions later. A market that rotates its leadership twice in one week is not trending, it is redistributing, and the small caps sitting under their 20-day average while the equal-weight index holds just off its highs is the tell that the redistribution remains orderly rather than distributive.

The Chronology and the Auctions

The Treasury thread opened in #73 got two verdicts this week, one from the tape and one from the archive. The tape's verdict came in three auctions: a very strong two-year, a solid five-year, and a seven-year sale graded a C by the auction watchers, a snoozer, which after the month the long end has had counts as the best possible outcome, with demand across the short end and belly holding up fine and only the composition shifting, foreign buyers concentrating in the front end while domestic bidders took the belly. The intervention experiment's first test produced no accident. But the two-year yield tells you what the calm is priced on: it sits roughly 58 basis points above the effective funds rate, which is the bond market openly pricing resumed hikes, and the strategists' consensus into this morning is that the chair must deliver at least a hawkish flavor to keep the long end anchored, with one warning that a keynote without framework guidance risks a much larger move in long rates than any hike path would cause.

The archive's verdict is the more important one, and the curation desk assembled it simply by putting the Treasury Secretary's own statements in chronological order. Before taking the job in late 2024: the next few years will bring a grand economic reordering, something equivalent to a new Bretton Woods. In early 2025: over the next twelve months we are going to monetize the asset side of the US balance sheet. This month: we have a big toolkit, and yields do not reflect fundamentals. Read individually, each is a quotable line; read as a series, in the desk's words, it is a man describing a project, then starting it, then telling you it is further along than the market has priced. Check the toolkit against the arithmetic and the options thin out fast: revenue cannot close a gap this size, the growth required is a factor no one forecasts, gold revaluation has been publicly taken off the table, the chair spent fifteen years arguing against printing, and default is not on the list. What remains is manufacturing new demand for the debt, and the mechanism, as the digital-asset analysts have argued since #73, is already law: a regulator cleared the path with a December no-action letter and an innovation exemption, then a statute pointed the resulting demand at government debt by requiring every blockchain dollar to hold Treasuries as backing. A transition expected to take two decades is running on a three-to-five-year clock because Washington needs it to. We keep the rate-sensitive shorts, we keep the acknowledgment that our opponent is a determined state, and we now keep a written chronology of the opponent telling us his plan. The fiscal backdrop has not improved while the engineering proceeds: on primary balances, ex-interest, the United States runs the widest primary deficit on earth.

SOURCE: Daily Chartbook DC Lite #671, Bloomberg via Simon White (US Has the Largest Primary Budget Deficit)

The near-term mechanics get their own date: the expanded long-maturity buyback program is already running ahead of the doubling that takes effect on September 9, so the experiment's scale test arrives within two weeks of the chair's first framework speech. The buybacks double in two weeks.

SOURCE: Daily Chartbook DC Lite #671, via @steve_donze (Treasury Buybacks Ahead of the September 9 Doubling)

The Falsifiers Stay Silent

We promoted hard assets to a co-equal pillar in #73 and put three falsifiers on record, so the honest discipline is to check them weekly. The dollar falsifier stayed silent: the index failed again at its 200-day average this week. The real-yield falsifier stayed silent: long yields sit just shy of their highs but did not spike, and the auctions absorbed supply without incident. The parabola falsifier half-fired and resolved: gold spent the week consolidating just under $4,700 with two flat sessions rather than an air pocket, digesting a three-hundred-dollar week the way bull markets are supposed to, while Bitcoin crossed $80,000 and the total crypto market capitalization posted its strongest three-day rally since the summer of 2021. The crypto technicians' roadmap is explicit and falsifiable in the house style: first a clean break and hold above the May pivot highs, then a reclaim of the anchored volume-weighted average price from the cycle top, the average cost basis of every buyer who entered at the highs, because once those buyers are back in profit the bear market is, in their framework, officially over. Last cycle the equivalent tell was the reclaim of the 50-week average. The move is impressive; the confirmation list is written; we hold the position and grade the list, not the excitement.

The broader complex keeps adding chapters. Wheat has broken to multi-year highs while speculators remain net short, the grain market's version of the record-shorts-at-records setup this letter has traded twice this year in equities; diesel inventories sit at record seasonal lows with refinery utilization already running near capacity, the distillate market entering demand season with no cushion; crude rose for a second day after reporting that the White House has no interest in returning to the June understanding with Iran, which quietly re-arms the chokepoint thread; and credit is voting with the equity bulls, the high-yield-to-Treasury ratio at its highest since February 2025. The global version of the thesis got its cleanest chart yet: across the OECD, government bonds are having a dismal decade, which is the 60/40-replacement argument of #72 written in every currency at once. A dismal decade for bonds everywhere.

SOURCE: The Daily Shot Brief August 27 (OECD Bonds, a Dismal Decade)

An Hour Before the Chair Speaks

This issue goes to press inside the window it grades, so the section is short and the conditions are written down. At ten this morning the chair delivers his first Jackson Hole keynote, previewing, per the Fed watchers, early findings from his five task forces on communications, the balance sheet, economic data, productivity, and the inflation framework; at the same minute the payroll benchmark revision lands, the annual true-up that erased 911 thousand jobs a year ago. The strategists' split is narrow but sharp: one house says the market actively wants more hawkishness because the growth and inflation numbers keep coming in strong and clarity would perk up the dollar; another warns the risk is not the hikes he signals but the framework he declines to give, because a long end left without guidance reprices itself. The morning's macro desk frame is the one we adopt: this is a clearing event, where the large players adjust hedges and the range of probabilities either collapses or splits. The labor data walks in tight: initial claims at 203 thousand against a five-week-ago print that was the joint lowest since 1969, continuing claims posting their biggest yearly decline since late 2022, a labor market whose weekly pulse looks nothing like the annual ledger about to be revised beneath it.

Two structural items file into the AI thread before the verdict. The leading AI lab is preparing to pitch investors a total addressable market above thirty trillion dollars, a number that would have read as satire in any prior cycle and now reads as a financing document, and the historical companion chart shows that buying stocks on massive addressable markets has not been a sure-fire strategy, both facts true at once, which is exactly how this letter holds them: a thirty trillion dollar market. And the world's central banks have started moving before their data forces them: Korea hiked preemptively to anchor inflation expectations, the Philippines followed in line, and the global policy clock is quietly turning hawkish in the same week our own two-year yield trades 58 basis points above the funds rate. The preemptive hikes have begun.

SOURCE: The Daily Shot Brief August 27 (Anthropic's Thirty Trillion Dollar TAM)

SOURCE: The Daily Shot Brief August 27 (Bank of Korea Hikes Preemptively to 3%)

The global file rounds out the week with the same message in three currencies. Twenty-nine countries are outperforming the S&P 500 since the start of 2025, led by a Korean market up more than 250 percent, and Korea is the economy the macro desks regard as the world's early-warning system, which makes its combination of booming exports and a preemptive rate hike the single most informative data pair of the week: the global cycle is strong enough to tighten into. In Europe the warning runs the other way, with the bond desks openly asking whether French government debt is losing its core status, sovereign stress migrating from the periphery to the center of the eurozone. Strength that forces hikes abroad and fiscal doubt that widens spreads at the core are two halves of the same regime this letter has tracked all year: the price of money is rising everywhere, and only the equity markets that can grow through it are being paid.

Charts Worth Watching

Six curations from The Chart Report's Thursday desk, each with its source below it: the buyers' reclaim in chips and biotech, the semis-software ratio at its lows, the gamma-versus-exposure gap, the twenty-nine countries beating the index, the Bitcoin moving-average test, and the wheat short squeeze in waiting.

SOURCE: The Chart Report August 27, via Brian Shannon (SMH and XBI, Buyers Back in Control)

SOURCE: The Chart Report August 27, via Broc Losing (Semiconductors at Multi-Month Lows Relative to Software)

SOURCE: The Chart Report August 27, via Coiled Spring Capital (Rising Gamma, Underexposed Managers)

SOURCE: The Chart Report August 27, via Peter Mallouk (29 Countries Outperforming the S&P 500 Since 2025)

SOURCE: The Chart Report August 27, via @Washigorira (Bitcoin and the 50-Week Moving Average Reclaim)

SOURCE: The Chart Report August 27, via Ole S. Hansen (Wheat at Multi-Year Highs While Speculators Are Short)

The Verdict

Grade the pre-written window first. The Nvidia condition resolved bullish: a strong print into de-risked positioning, the streak snapped with the best session since April 2025, and the first tranche of the semis add-back goes on with the rest staged behind the 50-day reclaims, per the #71 framework, one configuration from complete. The inflation print passed without repricing September in any of our sources' coverage, which is why the week's argument stayed on Nvidia and the Treasury rather than on prices. The software breakout is the week's structural event and the book was already on the right side of it; we resist the urge to chase what we already own and instead note the discipline the dispersion regime demands: the same week software made history, three of its own names fell six to ten percent on earnings, and the index approach to this sector remains a machine for averaging winners with losers.

The open risks are honest and dated. The volatility floor into a four-event fortnight is the cheapest hedging window of the year and we use it. The exposure index at two-year highs converts the crowd from fuel to kindling, and the next leg of the contrarian ledger now argues for patience rather than aggression. The Treasury chronology upgrades our rate-sensitive shorts from a positioning trade to a thesis about an opponent whose plan is published, whose buybacks double on September 9, and whose success would be the single most bearish development for the position, a falsifier we restate so it cannot be forgotten. And the morning's double release lands within the hour: if the chair provides a framework, the long end holds and the growth clock keeps the lease; if he does not, the move the strategists warn about begins at the long end, where our shorts live, and the book is positioned for exactly that asymmetry. The revision beneath him rewrites last year's labor market either way. #75 grades what the ten o'clock hour did to every thread in this letter, with the responses, as always, written down before the answers arrived.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions

The Index Doesn't Own Leadership

Monday's tape was the regime in miniature, and the curation desk's line names it: the S&P 500 doesn't own leadership. Technology fell 1.6% for its seventh straight decline while eight of eleven sectors rose, four of them by more than a percent; the equal-weight index finished green and outperformed; the Dow gained. An index dragged lower by its largest sector while most of its members advance is not a market selling off, it is a market rotating, and it is precisely the configuration this letter's book was built for: the trimmed semis sleeve absorbing the streak, the broadening longs collecting the other eight sectors, the equal-weight expressions doing exactly what they were sized to do. The bulls' line in the sand, per the technicians, remains simple: above the June peak with the advance-decline line in its uptrend, control has not changed hands, and the index does not own leadership.

SOURCE: Daily Chartbook DC Lite #668, Hostile Charts via Larry Thompson (The S&P 500 Doesn't Own Leadership)

The character of the bid underneath deserves its own sentence, because it was not defensive in the fearful sense. Consumer staples led Monday's sector table, healthcare and utilities firmed, and yet the small caps and the equal-weight index rose with them, which is not what a flight to safety looks like; a flight to safety does not carry two thousand stocks along. What Monday showed is the bifurcation this letter mapped in #72 running in reverse: for seven sessions the periphery has absorbed the capital leaving the core, the exact opposite of the pattern that defined the spring, when the core absorbed everything and the periphery starved. Whether that inversion is a week's fashion or the start of the durable broadening our book is built for is precisely what the equal-weight relative trend, now turning up against the cap-weighted index for the first sustained stretch this year, exists to answer.

The semiconductor chart itself has earned its bearish label, and we treat it with respect rather than denial. The index lost its uptrend from the March lows, failed its rally attempt into a now downward-sloping 50-day average, and has carved what the technicians call an inverted cup and handle, a breakdown from which targets the 200-day, with a loss of that level the thing that would, in their words, significantly shift the tone for the broader market. Nothing in that description is a prediction; all of it is a map, the same map our add-back conditions have referenced since #71, and the pattern, as the desk notes, still needs downside follow-through to mean anything.

The flow evidence around the pattern is filling in from both directions. The fund-flow strategists flagged semiconductor ETF outflows picking up, retail beginning to release the sector it spent the summer accumulating, at the same time the institutional de-grossing detailed below accelerates. Seven consecutive declines is also, historically, the kind of streak that exhausts itself: losing streaks of this length in a major sector have been rare outside genuine regime breaks, and the last one, in February 2020, marked the front edge of a crisis that had nothing to do with the sector itself. The distinction we will be grading Wednesday night is between a streak that ends because the news improves and a streak that ends because the sellers are simply finished; the de-grossing data argues the second condition is close, and the reaction to the print, not the print, will tell us.

The Flush Arrives

What makes this semis decline different from every dip since #62 is that the leverage is finally leaving. Goldman's prime desk reports that Info Tech was by far the most heavily net-sold global sector last week, driven almost entirely by long sales rather than fresh shorts, and the de-grossing was the largest in percentage terms in over two years. This is the capitulation ingredient that #71's add-back framework demanded and that every prior dip lacked: not rejustified dip-buying with leverage intact, but actual position surrender, hedge funds cutting gross exposure in the sector at a pace unseen since the 2023 regime. The distinction matters mechanically: dips bought on leverage produce second legs, dips sold into produce durable lows, and for the first time in six issues the flows are doing the latter. The largest de-grossing in two years.

SOURCE: Daily Chartbook DC Lite #668, Goldman Sachs via Robbie Stankard (Info Tech, Largest De-Grossing in Over Two Years)

The fundamental floor under the memory corner of the trade sharpened at the same time, from an unlikely source: a data-journalism chart pair showing that computer memory prices, which declined for literal decades as a law of the industry, have stopped falling and turned higher. Decades of deflation ending in a single inflection is what a genuine supply-demand regime change looks like, it is the engine under the SanDisk and Micron runs this letter has tracked since #68, and it reframes the memory names' violence in both directions as the market repricing a structurally different industry. Our add-back conditions stand exactly as written, one configuration closer: the flush is now visibly underway, and what remains is the undercut-and-reclaim or the confirmed higher low. Decades of deflation just ended.

SOURCE: The Daily Shot Brief August 24, Datawrapper (Memory Prices Fell for Decades, Until Recently)

The Treasury Enters the Chat

The macro event of the week was not a data print but a policy improvisation. Reports confirmed the Treasury may tap its roughly one-trillion-dollar General Account to fund expanded buybacks of long-dated debt, an operation one strategist called a Treasury-led Operation Twist, and the commentary arrived in two waves this letter files side by side. The sympathetic reading: buybacks do not create reserves, so this is duration management rather than money printing, with some QE-like reflationary effects. The critical reading, from the most respected macro voice on the tape: the intervention is troubling and largely ineffectual because it addresses none of the underlying arithmetic, enormous financing demand from the government and from tech simultaneously, while the traditional suppliers of that financing, China, Japan, and the Gulf, grow less dependable, and because market pricing is now leading the Treasury rather than the other way around. The mechanics underneath got their own warnings: funding buybacks by shifting issuance to bills tethers the government's interest bill ever closer to the Fed's policy rate, the debt ceiling caps the ammunition, and if yields spike anyway, the next stop is the Fed's balance sheet, which would make the new chair's cherished balance-sheet debate, in one economist's phrase, entirely academic. One chart in the week's reading described the whole sequence in advance: when countries run out of fiscal space, the curve steepens, governments issue short, and officials declare the market irrational. It is the oldest playbook in fiscal stress.

SOURCE: Daily Chartbook DC Lite #668, Shadow Price Macro via Robin Brooks (The Fiscal-Space Playbook)

The man at the center of Friday's stage sharpens the irony to a point. The new chair is the same official who objected to quantitative easing from inside the institution in 2011, resigned, and spent the following fifteen years arguing that balance-sheet expansion inflates assets for the wealthy while distorting the price of money. He now chairs the Fed at the precise moment roughly ten trillion dollars of Treasury debt needs refinancing at rates dramatically higher than it was issued at, with the Treasury already improvising interventions on the long end and one strategist writing this week that the selloff could still become a full-blown crisis if the arithmetic is left unattended. The bond market's own verdict on the intervention so far came from the Brandywine desk in one line: the vigilantes still do not believe him. This week's two-year, five-year, and seven-year auctions, running Tuesday through Thursday, are the first witnesses called.

The Japan file completes the picture, and it came through the morning print rather than the wires. When the yen came under renewed pressure this month, the world's largest foreign holder of Treasuries faced its recurring choice, sell the bonds to defend the currency or find dollars elsewhere, and it found them elsewhere: coordinated yen-buying intervention joined by the US, plus a Treasury-backed push to expand a Fed dollar-lending backstop that lets Japan borrow dollars against the debt it already owns rather than sell it. Read as a signal rather than a headline, the desk wrote, and we agree: the Treasury Secretary is actively engineering structures to keep America's largest creditors from becoming forced sellers, at the same time he claims the deficit has already peaked. The digital-asset analysts' reconciliation of those two facts, that a new source of Treasury demand is being built in the form of blockchain dollars whose reserve requirements make every stablecoin issued a buyer of government debt, moves from crypto-thesis to policy-thesis the moment you notice that every official action this month has been aimed at the same problem: who buys the forty trillion. Meanwhile the equity arithmetic of the whole question sits in one Goldman line: every fifty basis points of real yield costs the index roughly three percent of its valuation, and fifty basis points costs three percent.

SOURCE: Daily Chartbook DC Lite #668, Goldman Sachs via @mikezaccardi (50bp of Real Yields, a 3% Valuation Hit)

Hard Assets Take the Baton

While the AI complex de-risked, the oldest stores of value went vertical, and the weekly numbers deserve plain statement: gold rose from roughly $4,390 to $4,710 in a week; Bitcoin rose from the mid-sixty-thousands to nearly $79,000, and is now outperforming both the S&P 500 and gold over the trailing six months for the first time since last October; the metals and mining complex has formed a multi-month bull flag; and the technicians' morning essay argued the broadest measure of all, the Bloomberg Commodity Index, has yet to even break out of a decade-long cup and handle, meaning the commodity bull market, on this reading, has not started. This is the 60/40-replacement thesis of #72 going panoramic: energy was the first chapter, and the bid is now rotating across every real asset simultaneously, in the same month the Treasury began buying its own long bonds and the currency interventions went coordinated. Hard money does not need a narrative when policy is writing one.

The falsifiers for the hard-asset pillar go on record now, while the vertical is steep enough to make them uncomfortable. A durable dollar rally would be the first: every leg of the complex, gold, Bitcoin, the metals, the broad commodity index, is at bottom a short position against the currency, and coordinated intervention cuts both ways. A genuine real-yield spike would be the second, the scenario where the Treasury loses its own experiment and rates rise enough to make a non-yielding asset expensive to hold again. And the parabola itself is the third: gold adding three hundred dollars in a week and Bitcoin fourteen thousand is the kind of tape that ends in air pockets regardless of thesis, and Bitcoin's prior high near eighty-two thousand sits close enough overhead to be tested within days. Co-equal pillar means co-equal scrutiny; the position earns its size only while all three falsifiers stay silent.

Honesty requires the counterpoint that also arrived this week, and it cuts in the bulls' favor on a different front: the supply wave this letter flagged in #68 as the wall's eventual reversal, the issuance flood that ends every scarcity regime, remains absent, with debt-and-equity issuance as a share of market capitalization still subdued by historical standards this year. The corporate paper machine has not yet responded to record prices, the buyback bid still exceeds supply, and until that flips, the scarcity tailwind stands. The supply wave has not arrived.

SOURCE: The Daily Shot Brief August 24, via @WarrenPies (Debt Issuance vs Market Cap, Subdued)

The Growth Clock Never Sleeps

Beneath the rotation and the intervention, the economic engine reported another strong week, and the growth clock this letter has scored since #67 extended its winning streak. The flash services PMI printed its highest reading since the end of 2024, offsetting a manufacturing wobble; the eurozone's flash manufacturing PMI hit a four-year high; the Chicago Fed's national activity index held just under trend; and the global earnings picture stayed broad, with the share of countries seeing forward EPS downgrades near cycle lows. The chart-desk shorthand for the week's data, US business is booming, is also the answer to why the long end will not rally and why the doves hold a lease rather than a deed: an economy reaccelerating through services while wages firm and commodities break out is not an economy the bond market can price charitably, whatever the Treasury buys. The services engine reaccelerated.

The rate-path consequence is already visible at the margin. Into Friday's keynote, market-implied odds of resumed hikes edged up rather than down, a small move but a telling one: a services economy at two-year highs, wages firming, commodities breaking out of decade-long bases, and a Treasury visibly worried about its long end is not a configuration that lets a new chair sound dovish, whatever his instincts about the balance sheet. Our two-clocks framework has scored the growth clock ahead since #67, and this week extends the lead; the policy clock's response, delayed all summer by the inflation lull, now has a date and an hour attached to it.

Wednesday Night, Friday Morning

Everything above is prologue to a forty-eight-hour window as dense as any this cycle. Wednesday brings July PCE, the Fed's preferred inflation gauge, in the morning and Nvidia after the close, with consensus expecting earnings and revenue to double year over year, expectations one desk summarized with the phrase that will define the print: Nvidia is the alpha bank of the entire AI buildout, the circularity of the financing runs through it, and the question is whether even great numbers can move a stock the market has spent seven straight sessions de-risking. We note the setup's asymmetry has inverted from the usual: a tape that sells a name for seven days into its report has pre-paid much of the disappointment scenario, and the de-grossing data says the weakest hands have already left. Friday at ten, the chair who spent fifteen years attacking QE delivers his first Jackson Hole keynote while, at the same minute, the payroll benchmark revision lands, the same annual true-up that erased 911 thousand jobs a year ago, with estimates this time scattered across a six-hundred-thousand-job range. A Fed chair speaking about the price of money at the exact moment the labor market's official history gets rewritten is the kind of scheduling only this year could produce.

The geopolitical file added two fronts without resolving the first. The administration formally launched its economic pressure campaign against Iran, threatening punishment for any country doing business with Tehran, and the oil market's response was the most informative price action of the week: Brent fell the most in three weeks, traders reading the campaign as a substitute for escalation rather than a prelude to it, even as the prediction-market odds of the strait normalizing this year sank to a new low of 23% and the reporting warned that real economic pressure invites retaliation against Gulf energy infrastructure. And the Canada front went from deadline to rupture: talks collapsed, 50% tariffs landed on twenty billion dollars of Canadian imports, and dollar-for-dollar countertariffs arrive in two weeks, a live experiment in what the tariff-margin channel does when applied to America's largest goods supplier. Neither front changed the book; both went onto the calendar where the falsifiers live.

The rest of Wednesday's slate deserves its line as well, because Nvidia will not report into silence. The morning brings the second estimate of second-quarter GDP and July durable goods alongside PCE, a triple reading on the economy hours before the market's most important company reports; the software and security names, with the accounting and CRM majors, report the same evening, which makes Wednesday night a referendum on the entire growth complex rather than one chip company. The setup for the software half is the mirror image of the semis half: the divergence between the two tech trades that #72 measured at twenty-six opposite one-percent days this year means one side of that spread will be repriced by the weekend, and our book, long the broadening and trimmed the concentration, holds the same side of the spread it has held since spring.

SOURCE: The Daily Shot Brief August 24, S&P Global (Services PMI, Highest Since Late 2024)

Charts Worth Watching

The rate regime in a single comparison, the homebuilder pattern our shorts live in, the hard-asset changing of the guard, the metals setup, and the bulls' line in the sand.

SOURCE: The Chart Report August 24, via Phil Rosen (30-Year at 5.3%, Fed Funds Then 5.25%, Now 3.50)

SOURCE: The Chart Report August 24, via @chessNwine (Homebuilders Heavy at the Diamond's Lower Boundary)

SOURCE: The Chart Report August 24, via Luke Kawa (Bitcoin Outperforms Stocks and Gold, First Since October)

SOURCE: The Chart Report August 24, via Jason Perz (Metals and Mining, Multi-Month Bull Flag)

SOURCE: The Chart Report August 24, via Ryan Detrick (Bulls in Control Above the June Peak and the A/D Uptrend)

The Verdict

Grade the closed windows first, briefly, as they deserve. The consumer scare of #69 resolved as squeeze rather than turn: the guide window produced no trigger event, the small-cap add survives, and July's retail print files as noise beside a services economy at two-year highs. The minutes produced no hawk-camp expansion worth a headline, and September's hold survived to meet its real test on Friday. Both resolutions arrived silently, which is itself the lesson this letter keeps relearning: most pre-written triggers do not fire, and the discipline's value is knowing instantly that they did not.

The open positions write their own instructions into the window. The semis sleeve stays trimmed through Nvidia with the add-back conditions one configuration closer, the flush underway but the reclaim unproven; a strong print into de-risked positioning that reclaims the 50-day would satisfy condition (b) faster than we expected, and we will not hesitate if it does. The rate-sensitive shorts hold through the Treasury's intervention experiment, with the honest acknowledgment that we are now positioned against an opponent with a trillion-dollar account, and the counter-acknowledgment that every strategist quote this week says the same thing: interventions that fight arithmetic without changing it have never yet won. The hard-asset complex, energy, gold-adjacent, and now the broad commodity thesis, graduates to co-equal pillar of the book alongside the broadening longs. And the watch list for the next issue writes itself in three entries: the Nvidia reaction against a seven-day de-risk, the PCE print against a reaccelerating services economy, and Friday's double release, where the chair's first doctrine and the labor market's revised history land in the same minute. #74 grades the densest window of the quarter, with every response, as always, written down before the answers arrive.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions

The Widest Spread in the Market

The Chart Report's dispersion essay deserves its own section, because it names something this letter has been circling for weeks without the right frame. Consumer discretionary, they observe, carries the widest dispersion of any sector in the market, the widest gap between best and worst performer, consistently, in almost any tape, and the reason is structural rather than cyclical: consumer spending is roughly fixed in the short run, so a dollar spent at one retailer is a dollar not spent at another. The sector cannot move as a bloc the way semis or energy can; share gains at one name are share losses at another, mechanically. Their conclusion inverts the usual reading: that is not a warning, it is the opportunity, because a sector that cannot rise together throws off enormous single-name winners while the average goes nowhere, and the index approach fails precisely here, since the sector ETF holds the winners and the losers together and hands you the average of a distribution with almost no mass in the middle.

The frame goes straight onto tonight's guide count. The same desk's retail work shows Target and Ross as this year's leaders with Walmart and TJ Maxx lagging, a leadership map that scrambles the intuition behind our #69 trigger: the discounters assumed to be the defensive winners are split down the middle, and the guides landing tonight and tomorrow will not describe one consumer, they will describe several. The dispersion lens says to read them accordingly: a Target cut and a Ross hold, or the reverse, is not a contradiction, it is the sector working as built. Our trigger counts cuts across the majors precisely because no single guide can speak for the cohort. And the same width has infected the growth complex itself: semiconductors and software have posted opposite one-percent moves on twenty-six days this year, an unusually wide divergence between the two halves of the tech trade, which is the index-veiling dispersion this letter's rotation threads have traded since spring, now measured.

The lens matters for how this letter's own book is built, so we make the implication explicit. A dispersion regime rewards the equal-weight expressions and the sector-relative positions this book has favored all year, and it punishes exactly two things: index-level conviction in either direction, and the assumption that a sector's ticker describes its members. It is also, not incidentally, why the record index-level shorts documented this week may be the wrong instrument entirely; the funds are short the average of a distribution whose winners and losers are canceling each other out, paying carry to bet against a number that the underlying dispersion has made almost uninformative. In a market this wide, the index is the least interesting thing on the screen.

Housing Breaks First

The deep housing data released this week is the fundamental file behind our structural shorts, and it reads like sequential confirmation. Housing starts fell 12.4% in July to a 1.24 million annual rate, far below forecasts, with single-family starts at 808 thousand, the weakest since November 2022 and the second-weakest since May 2020; completions dropped to the fewest since May 2020; and the construction pipeline sits just above its four-year low. This is what a 6.66% thirty-year mortgage, a one-year high reached in the same month, does to the most rate-sensitive industry in the economy: the shovels stopped first.

SOURCE: Neil's Newsletter August 18, Census via Neil Sethi (Housing Starts Fall, Permits Rise to a February High)

The forward-looking series is grimmer still. Pending home sales, the contract signings that lead closings by forty-five to sixty days, fell a second straight month to an index level of 71.2, the second-lowest reading in the twenty-five-year history of the series, undercut only by this January, with every region down on the month and the West off 7.1% from a year ago. One hopeful seed sits in the wreckage, and honesty files it prominently: permits rose 5% to their best level since February, and single-family permits hit a four-month high, builders quietly positioning for the demand that a single leg lower in rates would release. That asymmetry is the housing trade in miniature: activity priced for the rates we have, permits priced for the rates builders expect, and our shorts positioned for the gap between the two persisting longer than the consensus believes. The market that leads every easing cycle is writing contracts at the second lowest ever.

SOURCE: Neil's Newsletter August 18, NAR via Neil Sethi (Pending Home Sales, Second Lowest on Record)

The second-order effects belong in the file too, because housing is never just housing. Existing sales normally represent ninety percent of US home transactions, and their collapse has already handed share to homebuilders dangling incentives, which is why builder equities and actual construction have decoupled. A frozen resale market freezes the turnover economy around it, the renovations, the furnishings, the moving trucks, the realtor commissions, which feeds directly into the discretionary softness the retail guides will describe this week. And it freezes labor mobility, workers who cannot sell cannot relocate, a quiet supply-side tax that shows up nowhere in the payroll count. The housing series are the smallest headline numbers of the week and the widest economic footprint, which is precisely why the structural shorts have been this book's most reliable position of the quarter.

Energy Is the New Bond

The technicians spent the week promoting our quietest position to a thesis. The energy sector ETF closed its seventh consecutive day of gains at a new all-time high, drilling through its prior peak, and the desk's framing goes far beyond momentum: the breakout arrives after a nearly eighteen-year accumulation period, the relative trend against the market has barely started, and, the sentence that earns the section title, energy is starting to gain traction as the new replacement for bonds in the typical 60/40 diversified portfolio. Set that beside the affine model from #69 projecting secularly low 60/40 returns and the bond-bear regime work from #70, and three independent desks have converged on the same structural conclusion from three different directions: in a world where duration no longer diversifies, the inflation-linked real asset does the job bonds used to do.

The oil complex's own internals sharpen the case. The national gasoline fund is tracing a bullish cup-and-handle toward the highest weekly close in its history, the retail-price echo of the seasonal records this letter has tracked for three issues. And the single most clarifying macro chart of the week came from the same curation desk: the 30-year Treasury yield bottomed for the year at precisely the moment traffic through the Strait of Hormuz collapsed. One chart, one timestamp, and the entire causal chain this letter has argued since June, chokepoint to crude to headline inflation to term premium, is visible in a single overlay. The long bond and the strait are one trade, our energy leg and our rate-sensitive shorts are its two expressions, and both made money all week.

Discipline requires writing down what breaks the thesis, and the energy file has exactly two exits. The first is diplomatic: a genuine strait reopening, which the prediction markets now price at barely one in three by year end, would unwind the scarcity premium in weeks, and our standing instruction, treat any deal headline as a rebalancing event, exists for that scenario. The second is demand: a real global downturn, the kind China's July data gestures toward, would swamp the supply story eventually. Neither is present, both are datable, and until one arrives the position's math is the simplest in the book: an eighteen-year base, a two-month-old breakout, a structural bid from the 60/40 refugees, and a chokepoint that compounds weekly. Positions like this are why the barbell was built.

The Contrarian Ledger, Audited

The positioning extreme this letter documented in #70 acquired its sharpest formulation from the technicians' chart of the day: Nasdaq futures positioning among asset managers and hedge funds at a record net short while insider buying across technology reaches a fifteen-year high, garnished with the observation that nearly 75% of active managers have underperformed the market over the past twelve months, which makes their collective bearish positioning, in the desk's words, a questionable vote of confidence. The statistical kicker is the kind this letter files with care: historically, when hedge fund exposure turns back up from an extreme while the index holds above its long-term trend, the market has been higher a year later in every instance on record. Flows agree with the insiders rather than the funds: combined equity and bond fund inflows are running at levels now exceeding the 2021 records, sponsorship that shows no sign of exhaustion, flows beyond the 2021 peak.

SOURCE: Topdown Charts via Callum Thomas (Global PE10 Valuation Ratios)

The career mechanics behind the ledger deserve one paragraph, because they are the fuel gauge. Three-quarters of active managers are behind their benchmark with four months left in the year, and a manager who is behind, short, and wrong does not get to stay short into December; the performance chase is the most reliable seasonal flow in the industry, and it compounds the squeeze arithmetic already embedded in a record futures short. The insiders, meanwhile, are playing the opposite game on the opposite timeframe, buying their own companies at a fifteen-year-high rate with no benchmark to chase and no quarter to survive. When the shortest-horizon money in the market is short and the longest-horizon money is buying, this letter has learned, across fourteen issues of grading itself, which side to respect.

But an audit reports both columns, and the internals produced a genuine contradiction this week. While the one-month-low count across the S&P indexes remains scarce, the quality signal #70 leaned on, NYSE 52-week lows now outnumber 52-week highs by more than three to one, a divergence that says the exchange's long tail, the closed-end funds, the rate-sensitive financials, the small illiquid names, is breaking down beneath an index at records. Both breadth facts are true simultaneously, and their coexistence is the bifurcation deepening: the investable core has no sellers, the periphery is quietly in a bear market, and every prior cycle's reckoning began exactly there. The valuation ledger closes the audit with the week's widest spread of all: on ten-year earnings, the United States now trades within reach of the most expensive single country on the planet, converging with the top of a global distribution whose median sits at less than half our multiple. The Sunday desk's phrase for the aggregate condition was unsafe investing, allocations to cash and bonds far below usual, and their prescription was not a sell order but a sentence this letter endorses: it tells you where you are in the cycle, and it prompts you to plan. We are the most expensive country on earth.

SOURCE: Daily Chartbook DC Lite #664, Fidelity via Jurrien Timmer (Fund Flows Exceed 2021, Equity and Bond)

The World Outside the Corridor

The global file cut both ways this week, and the honest read keeps both halves. On the bearish side, China's July data surprised to the downside across the board, activity weakening broadly just as its equity market had begun drawing global flows, and the Turkish macro desk in our own inbox led its morning note with rising global yields before turning to the Unitree robotics IPO, a reminder that the AI trade's next chapter is being written in offshore listings as much as American capex. On the bullish side, the breadth evidence refuses to narrow: more than 60% of stocks across all three market-cap segments remain in long-term uptrends, the cumulative advance-decline line made new highs alongside the index last week, and the defensive bid that flickered this week is not a flight but a broadening, the healthcare sector ETF closing at new all-time highs of its own. A market where healthcare makes records while energy makes records while the equal-weight consumer disperses is not rotating out of risk; it is rotating through it. Even so, the week's veteran caution stands: the next few sessions reveal whether the recent weakness is noise or something more meaningful, and the downside surprises are back.

SOURCE: The Daily Shot Brief August 18 (China July Activity, Broad Downside Surprises)

One more global spread completes the set, because it rhymes with the domestic bifurcation. The same fortnight that produced China's broad downside surprises produced the Kospi's twenty-percent melt-up two doors away, Japanese earnings above their trend channel, and European banks extending their quiet four-year win over the Magnificent 7. The global rally is not a tide, it is an archipelago, rising fastest where AI leverage, bank normalization, or fiscal repair give it a local engine, and stalling where the old engines, Chinese industrial volume above all, are visibly tired. For our book the instruction is the same abroad as at home: own the dispersion through the specific engines, not the aggregate through the world index.

Press Time, Again

The same-day cadence deserves a note of process honesty: the morning issue graded the tape, this one mines the seams, and both go to press before the day's information peak, the minutes at two, the Target and Lowe's and TJX guides tonight, Walmart tomorrow. The practitioners' letters in the inbox have already voted with their books: the trading desk we read has no semiconductors in its long-term book, waiting for cleaner trends, and recommends a mixed portfolio into fall, small confirmation that the unflushed-leverage caution running through this week's issues is not a bearish pose but a positioning fact. The Sunday chart letter's learnings frame the other side: the equal-weight index is turning up against the cap-weighted, the broadening thesis confirmed at the relative level; earnings revisions are surging with the strong macro; and sentiment and allocations are surging too, which is fuel until it is warning. Every thread in this issue, the dispersion, the housing collapse, the energy promotion, the bifurcated internals, the global spread, waits on the same two documents everyone else is waiting on. The difference, as ever, is that our responses are pre-written.

For the record, the pre-written responses standing into the evening: two or more major retail guide cuts flips the consumer read and puts the small-cap add under formal review; a minutes document revealing a hike camp materially larger than the three recorded dissents reopens the September question the market has closed; and a benign double, guides holding plus isolated dissenters, files this whole red week as the routine reset the quiet-strength regime periodically demands, with the covariance monitor deciding whether the pullback has further business. Nothing in today's deep seams changed those triggers; everything in them raised the resolution at which we will read the answers.

Charts Worth Watching

Five curations from The Chart Report's desk this week, each embedded with its source, each carrying one of this issue's threads in primary form: the internals contradiction, the strait-to-long-bond timestamp, the gasoline record in the making, the tech trade's civil war, and the defensive sector quietly making history.

SOURCE: The Chart Report August 18, via @jasongoepfert (NYSE 52-Week Lows Outnumber Highs 3 to 1)

SOURCE: The Chart Report August 17, via Michael McDonough (The 30-Year Bottomed When Strait Traffic Collapsed)

SOURCE: The Chart Report August 18, via @chessNwine (Gasoline Fund, Cup and Handle Toward a Record Weekly Close)

SOURCE: The Chart Report August 18, via @Bluekurtic (Semis vs Software, 26 Days of Opposite 1% Moves)

SOURCE: The Chart Report August 18, via Larry Thompson (Healthcare Sector ETF, New All-Time Highs)

The Verdict

The book positions unchanged into the two-document evening, and this issue's contribution is depth rather than direction. The housing file hardens the structural shorts: starts at post-Covid lows and contracts at series lows are not a trade thesis, they are a sector's fundamentals surrendering to the price of money, and the permit seed tells us exactly what the position's exit looks like, a genuine rates peak, not a headline. The energy leg graduates from barbell component to structural thesis, the bond-replacement frame now shared by three desks, with the strait timestamp as its causal proof. The dispersion lens goes onto the retail trigger tonight: cuts counted across majors, no single guide read as the consumer. And the internals contradiction, scarce lows in the core, multiplying lows at the periphery, joins the composite as a formal watch item, because bifurcation is how this market's warnings have always arrived, quietly and from below.

The ledger closes where the audit left it: record shorts against record insider buying, record flows against record valuations, an index at highs above an exchange sprouting new lows, the world's priciest market inside the world's broadest uptrend. Every one of those pairings is a spread, every spread is a bet someone is on the wrong side of, and the widest spread of all remains the one between the two clocks this letter has tracked since the verdict week. The minutes at two, the guides tonight, Walmart tomorrow, flash PMIs Friday, the chair at Jackson Hole inside a week. #73 grades the documents, and the seams, as always, will have told us first.

And a scoreboard note for the dispersion era this issue names: fourteen consecutive issues published without edits, every falsifier dated in advance, every exit pre-written, and a book whose best positions this quarter, the energy leg, the rate-sensitive shorts, the equal-weight expressions, are all dispersion trades rather than index calls. The frame this letter has been building since the verdict week has a name now, and the name is the title.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

Three Red Days and a Five Percent Semi Slide

The composition of the decline matters more than its size, which so far totals barely 2% from the highs. Tuesday's damage concentrated exactly where the leverage lives: the semiconductor index dropped 5%, led by the memory and optical names whose dip last week was bought through leveraged ETFs rather than flushed, while the equal-weight complex fell far less and the Dow nearly held flat. Defensives caught their first real bid in weeks, health care and staples green against the tape, energy extending its leadership, and our cyclical-defensive composite, which read unambiguously cyclical for six issues, printed its first softening: still no defensive regime, but no longer a shutout. The one-month-low count that #70 celebrated has begun rising off its floor, the NYSE highs-minus-lows line is falling, and the expectations-casualty file added its most dramatic entry yet, a payments company losing a third of its value on a guidance wipeout. This is what the start of a routine pullback looks like; it is also what the start of every less routine event has looked like, which is why the file exists.

The retail audit's first witness testified and complicated nothing. Home Depot beat, and its beat could not calm the tape, because the market's question was never last quarter's comps; it was housing, larger discretionary projects, and the guide. The pre-committed trigger from #69 remains formally at zero cuts among the majors, with today's slate, Target, Lowe's, TJX, carrying the middle of the count and Walmart closing it Thursday. We resist the temptation to score the trigger early: the tape is currently selling growth-adjacent retail regardless of results, and the trigger was designed to read the guides, not the reactions.

The reaction function itself is the tell worth logging. A clean beat sold, a guidance stumble obliterated, and nothing rewarded: that pattern says the marginal buyer in consumer-facing names has stepped back until the aggregate question resolves, which is rational positioning ahead of a three-day verdict window rather than a conviction change. It also means Thursday's final guide carries asymmetric weight: the biggest retailer on earth holding its outlook into this tape would clear the entire count at a stroke, while a cut from it would validate every defensive rotation of the week in one line item. The trigger waits for the document, not the drama.

The Hurdle Rate Is Rising

The long end is now the market's central character, and Tuesday gave it its sharpest line of the cycle. The 30-year traded to 5.33%, its highest yield since June 2007, and the bears did not let the date pass unremarked: the last time the long bond yielded this much, the equity bull market peaked three months later. We quote the ghost to keep it honest, not to obey it; 2007's yield came with a collapsing credit system underneath, and this one comes with credit spreads at cycle lows. But the level itself does real work regardless of analogs. Every pension allocator, every insurance CIO, every corporate treasurer now sees a quarter-century-high risk-free rate on the long end of the curve, and each of them re-runs the same arithmetic on equities that the hurdle-rate quote describes. Ten-year yields at 4.74% sit at the edge of what one prominent bear calls breakout territory, and mortgage rates at one-year highs already collected their casualties this week: housing starts back near post-Covid lows, completions the fewest since 2020, and pending home sales at the second-lowest reading in the history of the series. Our rate-sensitive shorts, promoted to structural last issue, spent the week collecting exactly this, and the long bond remembers 2007.

SOURCE: The Daily Shot Brief August 18 (30-Year Treasury Yield, Highest Since June 2007)

For the tripwire itself, the instruction stands exactly as rewritten in #70: the 30-year line watches for disorder, not levels, and this week has been the orderly kind of repricing, auctions clearing, bid-to-covers adequate, the curve steepening without gapping. What would change the classification is mechanical: a tailed auction, a funding-market headline, or the covariance alignment extending from two sessions to a week with the dollar bid accelerating. Until one of those appears, a quarter-century-high yield is information about the regime, not an instruction to the book, and the plumbing file from #69 tells us precisely where to look for the first crack if one comes: the repo-financed basis complex, not the equity screen.

The global chart of the week shows why this is not an American quirk. Ten-year-ten-year forward yields, the market's cleanest guess at the long-run price of money, are rising across every major economy at once, and rising fastest exactly where debt stocks are highest and politics most dysfunctional. This is the #55 thread, global yields rising in sync, three months on and structurally confirmed: a worldwide repricing of duration in which the US is a participant, not an outlier. The mechanism underneath got its cleanest statement yet from the flows desk this morning: net interest payments relative to GDP now feed themselves, the government pays out more, the coupons are spent back into the economy, and the Treasury issues more to fund it, a fiscal perpetual motion machine confirmed by outlays accelerating into July. Forward rates are not forecasting an event; they are pricing an arithmetic, and it is rising everywhere at once.

SOURCE: Daily Chartbook DC Lite #664, Shadow Price Macro via Robin Brooks (10y10y Forward Yields, Rising Everywhere)

Inflation Risk, Not Recession Risk

The cross-asset configuration answers the question every three-red-day stretch raises, and the answer is unambiguous. Credit spreads sit at cycle lows while long yields sit at cycle highs, and that pairing carries only one macro reading: the market fears inflation, not recession. If contraction were the risk being priced, spreads would be widening as yields fell; instead the system is pricing more nominal growth than the consensus models, exactly as the SOFR curve shows the doves holding the near-term upper hand while the strip refuses to price genuine easing anywhere on the horizon. The desks watching it agree on the tail worth fearing: not 2008 but 2022, the regime where rates and spreads rise together and there is nowhere to hide but energy and cash. The named tripwires for that tail are crude and the core inflation prints, which is to say the tripwires this letter has already double-weighted for six issues, and the doves hold a lease, not a deed.

SOURCE: Daily Chartbook DC Lite #664, Carson Group via Sonu Varghese (SOFR Futures, the Doves Hold Temporarily)

This afternoon's minutes are the lease terms in writing. The strip's math only holds if the committee is content to let the long end do its tightening; the moment the center of the committee migrates toward the dissenters, the front end reprices and the doves' lease expires early. That is the actual stakes of the hawk-count: not September, which is settled, but whether the reaction function that the households already fear, and that a former committee member spent Monday advocating, has quietly become the institution's base case. A document showing three isolated dissenters extends the lease through Jackson Hole; a document showing a debate extends nothing.

The growth clock, meanwhile, keeps stamping its case. The Empire State manufacturing survey, first of the August regional reads, jumped to its highest level since December 2021, torching the forecast sheet, and it lands on top of the 5.8% nowcast, the wage re-acceleration, and the seasonal-record gasoline prices already in the file. An economy this hot into a Fed this constrained is precisely the configuration in which the long end does the tightening on the committee's behalf, which is the single sentence that reconciles everything on the tape this week: the growth clock keeps winning.

SOURCE: The Daily Shot Brief August 18, NY Fed (Empire State Manufacturing, Highest Since December 2021)

The Floor Answers Back

Against all of it, the earnings floor delivered its strongest rebuttal in weeks, and it speaks directly to the crack this letter documented in #68. The revision-breadth worry, only 122 companies with rising forward estimates against past peaks near 150, now has its counterweight: forward earnings estimates have risen to record highs not just for the S&P 500 but for the S&P 400 midcaps and S&P 600 smallcaps simultaneously. Fewer individual engines may be pulling, but the aggregate expectations for every capitalization tier in the American market are at all-time highs at once, which is not what the top of an earnings cycle has ever looked like. Fund flows confirm the sponsorship: both equity and bond funds are absorbing money at rates now exceeding the 2021 records. The veteran strategist's framing stands as the bull case in one line: with earnings soaring and margins at record highs, stocks can still be seen as attractive, even against a 5.3% long bond. Records up and down the cap stack.

SOURCE: Daily Chartbook DC Lite #664, Yardeni Research (Forward Earnings, Record Highs Across Large, Mid, and Small Caps)

We keep the synthesis honest, because this section and the previous one are the same argument seen from opposite banks. The hurdle rate is rising, and the cash flows being discounted are rising too; equities fall only if the first outruns the second. That race, and nothing else, is what the next two months of prints decide, and it is why this letter's book holds both the broadening longs and the rate-sensitive shorts without contradiction: they are the two sides of the race, and the race is the regime.

Three Trillion Off the Books

The week's most important piece of AI reporting came from the accounting desk, not the trading floor. Big Tech has accumulated roughly $3 trillion in mostly AI-related commitments that sit off their balance sheets, the purchase obligations, capacity leases, and compute contracts that fund the buildout without appearing as debt. This is the financialization thread this letter has tracked from Alphabet's first negative FCF quarter through NVIDIA's GPU-collateral structures and the neocloud divide, now aggregated into a single number: the AI economy's true leverage is trillions larger than its reported one, and its funding runs through exactly the vehicles, leases, SPVs, and take-or-pay contracts, that history says get tested when rates rise. Nothing about the number is a warning in itself; committed capital is the bull thesis. But it means the semis complex that fell 5% on Tuesday is not merely an expensive sector taking profit, it is the equity layer of a three-trillion-dollar credit structure, and the flows desk named it precisely this morning: the leverage in semis has not been flushed, everyone rejustified the fundamentals and bought the dip, and the sector ETF is now the decision point between another down leg and a higher low back to the records. Our semis sleeve has been trimmed since #62 for exactly this moment, and the engineering deep-dives in our own inbox this week, benchmarking the chip leader's optical interconnects a full letter grade ahead of its nearest rival, keep the long-run moat argument intact while the near-term leverage clears. Three trillion dollars off the books.

SOURCE: The Daily Shot Brief August 18, WSJ (Big Tech, $3 Trillion of Off-Balance-Sheet Commitments)

For our own sleeve, the resolution conditions are pre-written now rather than improvised later. The semis position, trimmed since #62, adds back on either of two configurations: a genuine flush, meaning capitulation volume, leveraged ETF outflows, and the sector ETF undercutting its August low before reclaiming it; or a confirmed higher low, meaning the decision-point line holding on declining volume while the equal-weight tape makes new highs. What does not qualify is the pattern of the past week, dip-buying on rejustified fundamentals with the leverage intact, because unflushed leverage plus a rising hurdle rate is the configuration that produces the second leg. The moat is long-run; the leverage is now; the sleeve stays trimmed until the tape picks a configuration.

Press Time, Minutes at Two

The verdict calendar splits precisely across our press time. This afternoon at two, the July FOMC minutes reveal the only number that matters, the true size of the hike camp behind the three recorded dissents, into a market that has spent the week rediscovering rate fear from the long end rather than the front. The asymmetry stands as written in #70: September's hold is fully priced, so a benign document confirms what the market already believes, while any hint of a broadening hawkish coalition lands on a tape that is already three days into a pullback with semis unflushed. Behind it, today is also the Canada tariff deadline, the retail guides keep landing through Thursday, Friday brings the flash PMIs, and next week the chair speaks at Jackson Hole with the entire two-clock debate on his desk. Rarely has a quiet August week carried this much scheduled information.

A preview note on Jackson Hole, because the week's tape has quietly rewritten the chair's menu. Two weeks ago the symposium looked like a victory lap for patience: cool CPI, benign PPI, a hold locked in. It now arrives with the long bond at quarter-century highs, a consumer print cracked, wages re-accelerating, and the household survey showing a majority expecting hikes, which means every available speech carries a cost. Leaning dovish against record-easy financial conditions risks feeding the very long-end repricing that is doing the damage; leaning hawkish validates the front end against a tape already three days red. The corner this letter has described since the Warsh minutes, a chair whose credibility instrument is the one market he cannot control, is no longer theoretical, and the speech is now the single most consequential scheduled event between here and the August CPI.

The book enters it positioned rather than predicting. Nothing in three red days touched a trigger: the equal-weight relative line has softened but not broken, the small-cap add survives pending the guide count, the un-broadening alarm reads intact, and the volatility term structure remains inverted, which the flows desk reads as complacency and we read, alongside them, as a pullback that has not yet scared anyone. If the minutes are benign and the guides hold, this week files as the routine reset the quiet-strength regime periodically requires. If not, every exit in this book was written down issues ago, and the covariance monitor, stocks, long bond, dollar, now aligned for two sessions, is the line we watch before any headline.

Charts Worth Watching

One spotlight this week rather than the usual gallery, because the cycle produced a single fresh thread worth embedding: the futures-positioning extreme in Bitcoin, where large speculators sit near record long exposure, a configuration that has historically appeared closer to major lows than major tops, and the cleanest read on how the fast money is leaning across the entire risk complex.

The Verdict

Grade the week against the frame. The covariance alignment fired for the first time since it was installed, and the book behaved exactly as designed: the rate-sensitive shorts and energy leg collected, the trimmed semis sleeve limited the damage from the 5% slide, and the broadening longs gave back price without breaking a single structural line. The composite's first defensive flicker is noted, not acted on; one day of staples leadership is texture, and the trigger for the small-cap add remains the retail guide count, which stands at zero cuts with three reports to come. The pullback is, so far, the flows desk's version: tactical, unflushed, and living inside a bigger picture that is not bearish.

The watch list for the next 48 hours is the shortest of the year: the minutes at two, the guide count through Thursday, the SMH line the flows desk named as the decision point, and the three-market alignment. Beyond them, Jackson Hole and the August CPI remain the scheduled adjudicators, and the strangeness file keeps its entries open with one addition: the BTC positioning extreme, filed with the note that its historical resolution has favored the longs. The two clocks pulled in opposite directions all week, the hurdle rate rose, and the floor rose to meet it. The race is the regime, this page has both runners timed, and #72 grades the minutes, the guides, and whether the first pullback of the melt-up era stayed routine.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The Day the Music Paused

The tape's first stumble was instructive in its composition. The indexes lost half a percent, but underneath, energy was the only sector of eleven to advance while communication services, discretionary, financials, and staples all fell more than a percent, a one-engine day in the wrong direction. What kept the damage contained was the AI complex: semiconductors rose 1.6% against the falling tape, memory names extended on fresh capacity orders, and the leveraged ETF flows into the memory products show the dip being bought, not capitulated. The day's fuel was a private disclosure that the maker of the Claude models grew second-quarter revenue at least fourteen-fold from a year earlier, a number that goes straight into the caboose-and-train file: the demand side of the buildout keeps printing figures that make the capex look rational, which is exactly what the fault-line debate from #67 needs to stay unresolved.

The casualty list grew at the edges, in the pattern we have now tracked for a week. A global sportswear icon fell to a twelve-year low, the consumer discretionary tape's most famous name confirming the retail crack from the spending side; an optical components maker reversed hard after hours on good-but-not-great numbers, the fifth expectations casualty in six sessions. And yet the breadth work that closed the week says the sell-side of this market remains remarkably empty: across the S&P 500, 400, 600, and 1500 alike, the number of stocks setting even one-month lows has been steadily declining as the indexes grind higher. Individual names get executed for imperfection, whole cohorts do not break, and hardly anything is breaking down.

SOURCE: Daily Chartbook DC Lite #663, Thrasher Analytics via Andrew Thrasher (One-Month New Lows, Steadily Declining)

A word on how to read the retail guides as they land, because the mechanics matter more than the headlines. July's comparable sales will be soft; the market knows that, it just watched the aggregate print fall. The information is entirely in the full-year outlook lines: a retailer that holds its guide is telling you July was weather, calendar, and noise; a retailer that cuts is telling you its own transaction data, which extends weeks past the government's, shows the softness persisting into August. That is why #69's trigger was written against guides rather than prints, and why the distinction between a miss and a cut is the difference between noise and regime this week.

Easiest Ever, Shortest Ever

Now the paradox that names this issue. On one side: the Bloomberg US financial conditions index, initiated in 1990, closed Friday at the easiest reading in its entire history, easier than 1999, easier than 2007, easier than 2021. That is the liquidity backdrop this melt-up breathes, the empirical confirmation of the macro desk's thesis that valuations here are a liquidity phenomenon, and the quiet answer to anyone asking how a market absorbs a quarter-century-high bond yield without flinching. On the other side: asset managers and leveraged funds hold a combined negative $20.2 billion net position in Nasdaq-100 futures, the second most negative in available history, just shy of the record set the week before. The most institutionally shorted tech tape in the data is levitating on the loosest conditions ever measured, and both extremes are fuel: conditions this easy feed the advance, and shorts this large feed it too, every new high forcing another round of covering. The strategist watching retail flows called it precisely: the bears are scared, nobody will put money behind the top call, and the sideline is where you get whipsawed. This is what the easiest conditions ever recorded.

SOURCE: Daily Chartbook DC Lite #663, Bloomberg via John Authers (Financial Conditions, Easiest Since 1990 Inception)

Honesty about the short base requires one caveat before we lean on it. A portion of asset-manager futures shorts is overlay and hedge rather than directional conviction, portfolio managers who are fully long cash equities renting downside insurance in the deepest instrument available, so the position is not purely a bet against the market. But the caveat cuts only so far: hedges this large still unwind the same way directional shorts do, mechanically and pro-cyclically, buying futures as the index rises, and the prior episodes near these extremes resolved with the position chasing the tape higher, not the tape meeting the position. Fuel is fuel, whatever the paperwork calls it.

The insider ledger completes the triangle, and it does not side with the shorts. While traders hold record short positions against technology, corporate insiders in those same companies are buying at close to the highest rate in fifteen years, the people with the information leaning against the people with the leverage. And the character of the advance backs the insiders: the breadth analyst's phrase for it, a return of the quiet strength that is a hallmark of bull markets, matches every internal this letter tracks, persistent gains without excitement, declining new lows, records without drama. Strong markets are boring; blowoff tops are not; and whatever this is, it is not exciting, which historically has been the bullish answer. Quiet strength is back.

SOURCE: Daily Chartbook DC Lite #663, Hi Mount Research via Willie Delwiche (Quiet Strength, a Hallmark of Bull Markets)

The Regime That Broke the Playbook

The weekend's bond-market work deserves to be absorbed in full, because it reframes everything this letter has tracked at the long end since #63. The argument in one paragraph: from 1988 to 2020, interest rates only ever went down, and that single forty-year trend hardwired the reflexes of the entire financial industry, the mortgage refi cycle, the 60/40 portfolio, the buy-every-duration-dip instinct. That regime ended in 2020, and every strategy built on falling rates is now, in the desk's phrase, a hope rather than a plan. The evidence that this is structural rather than cyclical is exactly the evidence that confused the consensus all year: inflation retraced essentially its entire spike, and bonds never caught a bid. A cyclical bear ends when its cause reverses; this one's cause reversed and the bear continued, because the drivers are issuance, nominal growth, and an investment boom, not the CPI. Crude holding its levels, corn and wheat bid, copper elevated, and a curve that bear-steepens through every Fed debate all point the same direction, and the futures strip prices only about 42 basis points of additional tightening through 2027, which is the market quietly admitting that nominal GDP is running hotter than anyone's model.

The fiscal layer sits underneath, and one chart summarizes thirty years of it: federal debt has consistently and repeatedly defied every CBO forecast ever drawn, each projection revised toward more, the political economy of deficits beating the arithmetic of consolidation in every decade. The bond strategists now say the quiet part in client notes: every Treasury auction is a referendum on fiscal trajectory, every issuer from the government to the hyperscalers is competing for the same capital, and equities can look past higher-for-longer only for as long as AI momentum keeps meeting expectations. That last clause is the entire regime in one sentence, and the forecasts keep losing to the debt.

SOURCE: The Daily Shot Brief August 17, BCA Research via Martin Barnes (Federal Debt Defies CBO Forecasts)

For the visual record, the chart that should end every bonds-are-cheap conversation: the official inflation series, spike fully retraced, sitting near its lows of the year, printed beside a 30-year yield at 2007 levels. In the old regime that pairing was impossible; in this one it is the defining feature. Nothing about Monday's close at the 2007 highs was a surprise to the framework, and the practical conclusion for our book is unchanged but firmer: the rate-sensitive shorts are a structural position, not a trade, the 30-year tripwire watches for disorder rather than levels, and the money-market crowd earning 3% is not safe, merely slow, because the spike is gone, the bid never came.

SOURCE: Neil's Newsletter, BLS via Neil Sethi (CPI, the Spike Retraced)

For symmetry, the framework also tells us what would end the bond bear, and this letter already owns the sentence: bonds will not bid until the AI capex cycle ends or the credit behind it collapses. Watch the physical pipeline, the permits and the megawatts, and watch the neocloud credit spreads; the day those turn is the day duration becomes a buy and our rate-sensitive shorts come off. Until then the position pays us to wait, and the covariance monitor tells us what kind of day to fear: not a high-yield day, but a day when stocks, the long bond, and the dollar all break the same direction at once. That alignment, in either direction, is the regime's true tripwire.

The Ceasefire Expired

The Iran story crossed from chronic back to acute with a calendar entry. The 60-day ceasefire lapsed on Monday, the president said he was not interested in extending it, and Tehran signaled a more offensive posture if diplomacy fails; crude rose and did its now-familiar work on the long end. This letter's oil thread has argued since June that the strait is a structural feature, not an event, and the prediction markets agree with new conviction: the odds that strait traffic normalizes by year end sit at 34%, a new low, down from 86% in late June. The counterview deserves its space: with midterms approaching, the political pressure to deliver a deal is enormous, and at least one desk argues the risk to oil is now skewed to the downside precisely because everyone has stopped pricing diplomacy. Either way, the energy complex remains the year's quiet winner, the only green sector on Monday and up more than 70% year to date in ETF form, and every week the strait stays shut compounds through the gasoline records into the August CPI that decides September.

The compounding math deserves its own line, because it has been building since #63 without a single dramatic day. The strategic petroleum reserve sits at 1983 levels, the national gasoline average is the highest ever for this point of a year, energy is the best-performing sector of 2026, and energy earnings revisions have quietly followed price the whole way up. None of this required a missile; it required only that a chokepoint stay closed while the world stopped writing about it. Our energy barbell leg was built for exactly this profile, and the position's instruction stays what it has been for ten issues: hold it while the strait is shut, and treat any deal headline as a rebalancing event rather than a thesis change.

Three Risks, One Address

The flows desk consolidated its equity risk map into three named items, and we adopt the list because each maps onto a thread this letter already runs. Risk one: interest rate volatility driven by higher inflation, the #65 coil that never fully unwound, now armed by wage re-acceleration and seasonal-record gasoline. Risk two: a carry trade unwind against the leverage in the system, with the sharpest new detail being who carries it, foreign capital max-leveraged long AI stocks through the yen channel, which fuses our #68 carry framework and the AI trade into a single point of failure. Risk three: volatility complacency at the highs, now measurable in the term structure itself, one-week volatility priced below one-month at record index levels, hedging pressure lightest exactly where protection is cheapest to justify. Three risks, one address: the long end of the curve.

What makes the setup honest rather than alarmist is the loop that connects it back to the paradox section. The market pricing fewer hikes eases financial conditions; the easiest conditions ever recorded stimulate an economy already running a hot nowcast; hot nominal growth forces the long end higher and re-arms the inflation question. The relief is the risk. This is why the split verdict of #68 cannot simply persist indefinitely: either the long end eventually disciplines the equity market, or the equity market's own fuel keeps regenerating the pressure at the long end. The covariance frame from the same desk becomes our monitoring tool: the largest moves of every cycle come when the S&P, the 30-year, and the dollar move in lockstep, so the alignment of those three, not any one of them, is the tell, and it joins the dashboard tonight.

And as with every entry in the signals file, the clearing conditions go on the page with the risks. Risk one clears if the August CPI prints benign and wage growth rolls back over; risk two clears if the yen resumes its decline with the issuance mix unchanged and the cross-yen pairs keep grinding higher; risk three clears if the volatility term structure normalizes through an orderly expiration. Risks that clear are not defeats for the framework, they are the framework working, and the file exists so that the day one of them fails to clear, we notice before the tape makes it obvious.

The Audit Ahead

Everything above is context for a 48-hour window of verdicts. The retail guides begin landing now, Home Depot first, the discounters and Target and Lowe's behind it, Walmart on Thursday, and #69's pre-committed trigger is formally live: two or more major guides cutting full-year outlooks flips our consumer read from squeeze to turn and puts the small-cap add under review; guides holding files July as noise. Nike at a twelve-year low says the market has already convicted one corner of the consumer complex; the guides decide whether that conviction generalizes. Wednesday brings the minutes, and the single question that matters, as one strategist framed it: are more committee members migrating toward the hawkish camp, or was the July dissent isolated? Goldman's chief economist argues the market's hike pricing is still too aggressive and September is very unlikely; a former Fed president spent Monday arguing a hike makes sense; both cannot be describing the same committee. In a regime where the chair keeps his cards close, minutes carry more weight than usual, not less, and the market enters them with dovish confirmation fully priced.

Behind the window, Jackson Hole next week remains the argument-picker, and the strangeness file keeps its entries current: the volatility term structure inverted at the highs, the SVIX relationship still on watch, the Hindenburg-cluster chatter now reaching the mainstream press. None of it changed our disposition; all of it sharpens the same conclusion this letter reached in #68: the quiet is structural, the risks are named and addressed, and the cost of standing aside, as the whipsaw crowd keeps discovering, has been the highest-priced insurance of the year.

Charts Worth Watching

The embeds below carry the paradox in primary form: the flows desk's three-risk thread, the insider-buying extreme against record shorts, and the second-largest Nasdaq futures short position ever recorded.

The Verdict

The book does not move on a one-day pullback, and it does not move ahead of a 48-hour verdict window; the discipline this week is to let the triggers we pre-wrote do their work. The broadening book stays deployed with the small-cap add under conditional review pending the retail guides; the semis sleeve stays trimmed while the AI bid carries the tape; the rate-sensitive shorts graduate, per the regime work above, from December hedge to structural position against a bond bear whose cause is issuance and nominal growth rather than any single print. The energy exposure inside the cyclical barbell, the quietest position in the book, is now its best performer and stays untouched while the strait stays shut.

The dashboard gains one instrument and one promotion. New: the SPX, 30-year, dollar covariance alignment, the flows desk's observation that every major move in either direction arrives when all three trade in lockstep; we will track it weekly beside the equal-weight relative line. Promoted: the volatility term structure, one-week against one-month, as the cleanest live reading of complacency at the highs, joining the SVIX relationship in the signals file with the same clearing conditions. The falsifier calendar is unchanged and dense: retail guides now, minutes Wednesday, flash PMIs Friday, Jackson Hole next week, the August CPI on September 10. #71 grades the guides and the minutes, and if the paradox resolves in either direction, easiest-ever conditions breaking the shorts or the shorts finally being right, this page will say so in the plain terms it was written down in advance.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The Consumer Blinked

Hold the number against the framework before explaining it away. A 0.6% monthly decline in retail sales after eight consecutive gains is exactly the kind of turn our consumer paragraph in #68 said to watch monthly, and it arrived one month after we wrote it. The squeeze mechanics were already on the page: real weekly earnings down in four of the past five months, gasoline at seasonal records, and a paycheck that has been losing to the price level since spring. What was missing was evidence that the squeeze had reached behavior, and this is that evidence. One month is not a trend, July had fewer shopping days than usual, and the drop follows an unusually strong June; every softening in this cycle has been explained away by something. But the pattern this letter cares about is sequence: spending cracked first, while layoffs remain at record lows, which means the consumer is trimming by choice rather than necessity. That is the mildest possible version of a consumer turn, and also how every less mild version begins.

The audit arrives within days, which is the good news. This week brings the heaviest retail earnings batch of the quarter, more than $100 billion of reporting weight: Walmart, Home Depot, TJX, Lowe's, Deere, and the discounters around them, with Alibaba and Baidu carrying the ex-US read. Between the government's aggregate and the retailers' own tills, we will know by Friday whether July was a rounding error or a regime note. The market has pre-decided: soft consumption plus cool inflation equals fewer hikes, and the 27th record close of the year followed. We simply note that there is a level of consumer weakness that stops being a rates argument and starts being an earnings argument, and the entire difference between those two readings is whether the retail guides this week hold.

The sector tape already voted on how to read the week, and its ballot is worth recording. The best-performing ETF categories of the week were cloud computing, up more than 11%, and energy, up more than 6% and now over 70% higher on the year, with biotech close behind. A consumer wobble that coincides with capex strength and an energy bid is the growth clock's version of events in miniature: spending rotates from households to data centers, and the inflation input the Fed cannot control keeps compounding at the pump. If the retail guides confirm the wobble this week, watch whether the market treats it as a rates gift or an earnings warning; the ETF flows suggest it has already chosen the first reading.

Households Price the Hikes

The most quietly remarkable data point of the week was not the retail number; it was what the households said they expect. The University of Michigan survey shows 53% of consumers now expect interest rates to rise over the next twelve months, up from 44% in July and the highest reading since November 2023. The Fed's hawkish turn, the dissents, the Warsh reaction function, the autumn hike debate, has fully penetrated Main Street psychology, at the precise moment the futures curve stopped pricing even one full hike by year end. Somebody has this wrong. Either the households are extrapolating a headline that the bond market has already moved past, or the bond market's short end is underpricing a reaction function the public has intuited correctly. We flag the split because it now maps onto behavior: a household that expects higher rates defers the car, the renovation, the house, and some of July's retail softness may be exactly this expectation channel at work. Hawkish talk has consequences even without hikes, and the message finally landed.

The Plumbing Under the Long End

Two exhibits from the weekend reading turn #68's auction observation into a positioning story, and together they are the most important risk work in this issue. The first quantifies the machine: hedge funds now run roughly $4 trillion of gross Treasury exposure, $2.4 trillion long against $1.6 trillion short, and the book is dominated not by directional views but by repo-financed relative-value trades, basis, swap-spread arbitrage, and curve positions. This is the modern descendant of the exposure this letter first flagged in #59, when the hedge fund share of outstanding Treasuries had doubled since 2022, and it means the long end's clearing price sits on top of a leverage stack that requires calm funding markets to hold. A 5.21% auction in an orderly tape is one thing; the same yield discovered during a repo squeeze or a forced unwind is a different event entirely, and the difference is not macro, because the plumbing is the position.

SOURCE: The Daily Shot Brief August 17, Financial Times (Hedge Fund Treasury Exposure, $4 Trillion Repo-Financed)

The second exhibit is the directional crowd, and it points the opposite way from the arbitrage book. Hedge funds have been loading up on long-term Treasuries outright, presumably reasoning as the bond bulls do, that a 25-year-high yield eventually pays you to be early, and the historical study attached to the flows is blunt: when the fast money has crowded into duration like this, it has usually been early enough to hurt. Set the two exhibits side by side and the long end becomes the most crowded uncrowded trade in the market, arbitraged by four trillion dollars of leverage in one direction and bottom-fished by directional funds in the other, all sitting under the structural issuance and AI capex flows this letter documented in #67. Our rate-sensitive shorts and the 30-year tripwire live in exactly this neighborhood, and we note the study's warning with respect, because history says they regret it.

The Valuation Stack

The valuation conversation acquired its cleanest chart of the year, from the analyst whose whiplash we documented in #64 and whose secular work has always been better than his timing. His mapping of eighty years of valuations, allocations, and cash rates carries the punchline in the allocation series: US household equity allocations are at record highs, and institutional allocations are at their highest since 2000. The principles he attaches deserve their place in the file: everyone wants to own stocks at the top and few at the bottom; the great valuation compression of the 1970s happened because double-digit cash rates made an actual alternative; cycles persist inside every secular shift; and, the one we underline, market drift is an active decision, because letting a rising equity weight ride without rebalancing is a portfolio choice most holders do not know they are making. After a year in which this letter has documented record flows, record buybacks, and a 58% institutional allocation, the chart simply says what all of it sums to: everyone owns the top.

SOURCE: Topdown Charts via Callum Thomas (Valuations, Allocations and Cash Rates)

The counterpoint arrived in the same day's reading, and it complicates the crowding story in a direction the bears will not enjoy. The forward P/E of the ten largest S&P 500 companies has converged with the forward P/E of the remaining four hundred ninety, meaning the concentration premium that defined this cycle's valuation critique has quietly closed, through the rest of the market re-rating and the top tier's earnings catching up to its prices. Read alongside #68's record in the equal-weight indexes and the Value Line Geometric, the message is consistent: the advance did not just broaden in price, the multiple broadened too. A market where everything trades at the same rich multiple is a different risk than a market held up by ten expensive names: less concentration risk, more valuation risk, everywhere at once. Which reading dominates depends entirely on whether earnings keep arriving, and that question this letter has tracked for eleven issues stays open.

SOURCE: The Daily Shot Brief August 17, J.P. Morgan via @MikeZaccardi (Top 10 Forward P/E Converges With the Rest)

File both exhibits against the open question #68 left on the table, because they feed it directly. The liquidity desk's decomposition says this year's returns are mostly multiple expansion; the allocation chart shows who supplied the multiple, households and institutions at record equity weights; and the convergence chart shows where it went, everywhere at once rather than into ten names. None of that settles whether earnings or liquidity is the true engine, but it sharpens the stakes: a market at uniform rich multiples with record allocation and a thin cash buffer has spent its rerating fuel, and from here the arithmetic hands the job back to earnings. Which is, perhaps uncomfortably, exactly what this week's retail guides and next quarter's AI reports are for.

Signals Behaving Badly

The volatility complex spent the week extending the strangeness #68 flagged. The short-vol product that should rise when stocks rise and volatility falls has spent several weeks moving off script relative to the index, and the practitioner who tracks the relationship drew the conclusion we quote for the file: if the variation is significant, the market cannot be trusted and deserves extra vigilance for more serious misbehavior. This is the same family of evidence as the VIX rising into record highs, hedges being cooked, tails compressing while covariance falls, and it now spans two independent instruments. Meanwhile the search-engine data shows retail interest in risk-on terms far outpacing risk-off terms, the crowd leaning into exactly the tape the vol complex distrusts. None of this is a timing signal; all of it is texture, and the texture says the melt-up's machinery is running hot, off script in a quiet way.

SOURCE: Daily Chartbook DC Lite #662, Vixology via Jim Carroll (SPX vs SVIX, Off Script)

Two more entries for the strangeness file, both from the weekend mail. The seasonality argument that August-of-a-midterm-year should be treacherous is being flatly defied: 2026 ranks as the third-best midterm year through this point since 1950, behind only 1954 and 1958, which either means the script fails this year or the script's reckoning is deferred, and midterm-year history says the autumn decides. And one of the most sophisticated trading firms in the world reportedly lost $15 billion in July, its worst month ever, a reminder that the dispersion-and-compression regime this letter keeps describing is genuinely hard even for the machines. When the market makers bleed and the short-vol products misbehave while the index prints records, the surface calm and the underlying mechanics are telling different stories. We trade the tape while listening to the mechanics.

For discipline's sake we also write down what clears the file. The SVIX relationship snapping back into line for two consecutive weeks retires that entry; the VIX resuming its normal inverse dance with new highs retires the second; and a quiet options expiration plus an orderly August would downgrade the whole section from watch to archive. Strangeness that fails to escalate is just noise that got our attention, and half of these entries have historically resolved exactly that way. The other half preceded the interesting weeks, which is why the section exists.

The Week the Minutes Land

The calendar is light on data and heavy on meaning. There are no Fed speakers scheduled at all this week, an unusual silence, which leaves Wednesday's July FOMC minutes as the only official word, the first extended look inside the meeting that produced three hawkish dissents. With September priced back to a hold, the minutes are asymmetric: dovish confirmation is already in the price, while any hint that the hike camp is larger than three reopens everything. Friday brings the August flash PMIs as the first fresh growth read, and the JPMorgan forecast sheet flags one sleeper: import prices, expected sharply negative, which would either confirm the importers-absorbing-tariffs story from the PPI or start unwinding it. Behind it all, next week is the one this letter has circled since #68: Jackson Hole, where the chair picks an argument.

A note on why these particular minutes matter more than most. This chair's first minutes, back in July of last cycle's timeline, were the document this letter called the most hawkish since 2022, the moment the hike scenario entered the market's bloodstream; every repricing since, the 55% September peak, the retreat to 35%, has been the market arguing with that document. Wednesday's edition is the first drafted after a negative payroll print and before a clean CPI, which makes it a natural experiment in reaction-function reading: if the dissenters' language is contained, the hold thesis hardens into Jackson Hole; if the discussion ranged wider than the vote, the households whose expectations just jumped to 53% will look prescient rather than late. Either way we will know the committee's true center of gravity for the first time since the verdict week.

The Iran file, per the weekend roundup, has settled into exactly the shape the oil market feared most: not escalation but permanence. The administration has reportedly decided against military escalation, the diplomacy drags with little change expected through the mid-terms, and the prediction-market odds that strait traffic normalizes by year end have collapsed from 86% in June to a new low of 34%. A closed strait that is no longer news is a closed strait that compounds quietly through every gasoline record and CPI print. And the AI complex added three filings to its dossier: the chip leader disclosed a $21 billion stake in its launch partner, now its second-largest position, the corporate treasury as venture book; the largest neocloud secured another 800 megawatts of capacity, the physical pipeline #67 told us to watch; and the sharpest energy analyst in the feed reframed model economics in a sentence, arguing that cost per task, not cost per token, is the real metric, on which basis the flagship open-source model runs roughly 25% more expensive than the frontier. The falling-cost narrative that fuels every AI-commoditization scare measures the wrong thing, which matters directly for the moat, and the margins, of everything this letter holds.

One global postscript from the same morning's reading, because it rhymes with our whole thesis: European banks, the asset class everyone left for dead a decade ago, have outperformed the Magnificent 7 since 2022. Not European tech, the banks. The broadening this letter has traded all year is not an American anomaly; it is what a world of positive rates and real earnings does to unloved assets everywhere, and the laggards became leaders.

SOURCE: The Daily Shot Brief August 17, Goldman Sachs via @eliant_capital (European Banks Outperform the Mag-7 Since 2022)

Charts Worth Watching

The embeds below carry the week's freshest primary threads: the household rate-expectation jump, the duration crowd's history lesson, the risk-on search surge, and the cost-per-task reframing of AI economics.

The Verdict

Grade the frame, then the book. #68 named the split verdict and promoted the yen carry to first tripwire; nothing this week healed the split, and two things deepened it: the consumer cracked in the direction the growth bears wanted, and household rate expectations jumped in the direction the growth bulls needed. The two clocks are now visibly pulling on the same consumer. Our positioning is unchanged and, for the first time in three issues, genuinely tested: the broadening book's next leg depends on retail earnings confirming that July was noise, and the designated first cut, the small-cap add, is the position most exposed if the consumer turn is real. We pre-commit the response rather than improvise it later: two or more of the big retail guides cutting the full-year outlook flips our consumer read from squeeze to turn, and the small-cap add goes under formal review that day; guides holding, and the July print files as noise beside the June strength.

The watch list this week is short and sharp. Wednesday's minutes for the size of the hike camp; the retail guides for the consumer verdict; Friday's flash PMIs for the growth clock; import prices for the tariff-margin channel; and the SVIX relationship plus the vol signatures for the market's own honesty. The plumbing exhibit goes on permanent file next to the 30-year tripwire: any funding-market stress headline now reads as a long-end event, not a money-market curiosity. Jackson Hole sits one week out, and this letter's framework enters it with its falsifiers current, its exits pre-written, and its two verdicts still unreconciled. #70 grades the minutes, the retailers, and the week the consumer question got asked out loud.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The Range Broke Upward

The producer price report cleared the second inflation hurdle of the week the way the CPI cleared the first: benign at the headline, complicated underneath. Final demand prices were flat against expectations of a modest rise, the annual rate slowed to 4.7% from 5.5%, a third straight month of deceleration off the war-shock peak, and goods prices fell a second consecutive month with gasoline down nearly 6%. But the cleanest read on underlying pressure in the report, core services excluding trade margins, accelerated at its fastest pace since May, driven by a 6.5% jump in portfolio management fees, and analysts raised their estimates for July PCE, the Fed's preferred gauge, in response. There is something almost recursive in the detail: the bull market itself, through the management fees it generates, is now one of the firmer sources of measured inflation. Meanwhile trade margins fell again, meaning importers are still absorbing tariff costs rather than passing them along, a quiet subsidy from corporate margins to the CPI that cannot run forever.

The market took the report at its headline. September hike odds fell to roughly 35%, from 55% only a week ago, the futures curve no longer prices even one full hike by year end, and jobless claims reinforced the no-stress labor read with the four-week average of initial claims at its lowest since October 2022. The S&P rose 0.7%, crossed 7,800 intraday for the first time, and closed at a record; the Nasdaq led at +0.8%; technology, communication services, and real estate carried the day while oil fell 2% to about $81 and helped everything else. The commentary converged on a single reading: nothing in the week's data demands a September hike, the doves have their case, and the hawks have been rescheduled to the autumn, with Jackson Hole at the end of the month as the next chance to hear the chair think out loud.

The breadth of the record matters as much as the record. The confirmation signal one respected technician published Thursday, a modern reading of the oldest trend framework there is, fired on the simultaneous new all-time highs in the equal-weight S&P and the equal-weight Nasdaq 100, and the supporting cast was everywhere: the NYSE Composite, mid caps, small caps, and, the one we weight most, the Value Line Geometric index, which proxies the median stock, all pushed into record territory together. Whatever else Thursday was, it was not narrow. Our composite read cyclical again, the un-broadening alarm read untripped again, and the pennant resolved higher.

SOURCE: Neil's Newsletter August 13, BLS via Neil Sethi (July PPI, Flat Headline, Core-Core Firmest Since May)

The Other Auction

Now the dissent, and it did not come from a survey or a strategist; it came from the deepest auction on the calendar. Hours before the equity close, the Treasury sold 30-year bonds at 5.21%, the highest auction yield in roughly twenty-five years. Read that against #67's mechanics and the message is coherent: the long end is not pricing a September hike or fighting one, it is pricing a world of 5.8% nominal growth, structural issuance, and an AI capex cycle that keeps real yields pinned, and it has concluded that none of this week's inflation data changes any of it. The stock market celebrated the deferral of the next hike; the bond market kept selling the next decade. Both can be right for a while, which is precisely what makes the configuration unstable: every additional point of equity upside is being discounted against a risk-free curve that keeps getting less friendly at the long end.

For this letter the auction is a tripwire update, not just color. The 30-year has been double-weighted on our dashboard since #63, when it first touched its 2007 highs and the issuance channel was named; a quarter-century-high auction yield is that same signal at full volume, and it arrived without any disorder, which is the detail that keeps it a watch rather than a trigger. The line we pre-wrote still governs: it is a disorderly long-end move alongside a yen break, not a high yield in an orderly tape, that flips the book defensive. Thursday was orderly. The bid-to-cover was adequate, the equity market never flinched, and the curve simply moved to a new, higher shelf. We record the shelf and keep the trigger armed.

The volatility market noticed what the equity indexes did not. The VIX rose on a day the index made new highs, the signature, as the sharpest flow desk we read puts it, of short covering and cooked hedges rather than fresh fear, and cross-asset covariance keeps falling while the tails compress. A market that shakes off every shock is also a market where the probability mass migrates to the extremes: the same desk estimates that at current valuations a modest contraction in macro liquidity, one that would barely register in a normal regime, could move the index 10% in either direction. That cuts both ways, and honesty requires the other half: by the same logic a modest liquidity expansion melts the market up another leg. Sensitivity is not direction.

We also file the week's most serious analytical dissent against our own published view. The same desk decomposes the last two years of index returns, and especially this year's, into mostly valuation expansion rather than earnings, driven by the quantity and price of money, which collides directly with the earnings-driven framing this letter adopted from the Fidelity analog work in #66 and heard again from strategists Thursday. Both decompositions cannot be fully right, and the difference matters enormously: an earnings-driven market survives a liquidity contraction and a liquidity-driven market does not. We hold our earnings reading, on the strength of the revenue breadth, margin records, and delivered guidance this letter has documented for ten issues, but the dissent goes on the page at full strength, because it is the best-argued version of the case against us, and the falsifier is observable: if the market breaks on a liquidity event while earnings are still rising, the desk was right and we were wrong.

The Carry Underneath

The mechanism that connects the two verdicts runs through Tokyo. The same flow desk maps the quantity-of-money channel feeding the equity melt-up: a falling yen funding yen-denominated leverage that ends up in US risk assets, with Aussie-yen and peso-yen grinding higher as the visible signature of capital walking out the risk curve. This is why our yen tripwire at 160 was never really a currency trade: if the carry unwinds, the leverage supporting the equity bid unwinds with it, regardless of how strong the capex numbers or the chip earnings are that week. The 2024 rehearsal is the reference case, when a modest yen rally took the Nikkei down 27% peak to trough and clipped 10% off US indexes in days. The devaluation trend is mechanically driven, as #67 documented, by an issuance mix the finance ministry shows no sign of changing, so the carry keeps building until policy changes or something breaks it.

The frontier edge of that same flow is on brilliant display in Seoul. The Kospi has rallied more than 20% in ten trading days, the kind of move that in #61 this letter was citing from the opposite direction, when Korea, half its market capitalization in two chip stocks, was the designated offshore edge of the semis unwind and the "Boom and Bust 101" exhibit. Fourteen issues later the same concentration is working in reverse: the market most leveraged to the AI cycle is now the fastest-melting-up market on earth. We draw the honest lesson rather than the comfortable one: Korea was never a broken market or a confirmed one, it is the highest-beta expression of the global trade this letter is long, and it will lead in both directions. When the unwind comes, the Kospi will be the first tape to say so. For now it is the loudest confirmation on the board, the same coin, flipped.

SOURCE: The Daily Shot Brief August 13 (Kospi Rallies More Than 20% in 10 Days)

Wages Woke Up

The growth clock, having lost the week's headline arguments, quietly won a structural one. The Atlanta Fed's wage growth tracker, the cleanest wage series there is, re-accelerated across every cohort in July: overall wage growth rose to 3.8% from 3.6%, job stayers to 3.6% from 3.4%, and job switchers to 4.4% from 4.1%. This is the official confirmation of the churn premium this letter filed as a seed in #65 from the ADP data: the reopening gap between what companies pay to keep people and what they pay to poach them is the classic early symptom of labor cost pressure returning, and it is now visible in the benchmark series, in the same month that hiring plans printed their best July since 2022. Real weekly earnings have fallen in four of the past five months, so the paycheck squeeze is still on, but the direction of travel just changed, and the churn premium went mainstream.

SOURCE: Daily Chartbook DC Lite #661, Atlanta Fed (Wage Growth Tracker, July Re-Acceleration)

The other inflation input the Fed cannot control is sitting at the pump. The national average gasoline price is now the highest it has ever been this late in the calendar year, in data going back two decades, the retail-level echo of a strait that has been closed for two months and a crude curve that refuses to build a buffer. Put the two series together and the growth clock's transmission mechanism is fully specified: wages re-accelerating into a labor market with record-low layoffs, and energy costs at seasonal records into an August CPI print that arrives before the September meeting. Nothing about Thursday's celebration addressed either input, and the pump meets the paycheck.

SOURCE: The Daily Shot Brief August 13, Bianco Research (Gasoline, Highest Ever for This Point of the Year)

Two Pre-Written Reversals

The earnings floor this letter leans on acquired a crack worth measuring precisely. The number of S&P 500 companies with rising forward earnings estimates is only about 122, against nearly 150 at the 2018 peak and 163 in 2020, meaning the index-level revision strength we have celebrated for three issues is being generated by a narrower set of engines than in prior cycles. This is not a contradiction of the revenue breadth story, all eleven sectors are still growing, but it is a warning about its durability: fewer firms are doing the upgrading even as more sectors participate in the top line, and if the AI cohort's estimates ever stall, there are fewer secondary engines to take over than the index makes it appear. The market feels broad; only 122 firms are doing the pulling.

SOURCE: Daily Chartbook DC Lite #661, Paulsen Perspectives via Jim Paulsen (Narrow Earnings Momentum, 122 Firms vs Prior Peaks)

The second reversal is in the supply of shares itself. The research that quantified it puts numbers on both ends: the equity issuance surge at the height of the dot-com bubble reduced returns by as much as 10% annually, while the shrinking US share count of the past decade, the buyback era this letter's trillion-dollar wall belongs to, added roughly 0.7% a year to returns. The warning is in the pipeline: a coming issuance wave, the IPOs, secondaries, and stock-funded deals that every euphoric tape eventually invites, would flip the scarcity tailwind into the classic late-cycle headwind. Thursday supplied a preview of the appetite: a single buyout report sent traders on a software treasure hunt, and a memory maker's long-term targets supercharged its whole group. The corporate paper machine responds to exactly this kind of demand, and history says it always shows up eventually. The wall is a buyer of last resort on one side, and the scarcity tailwind can reverse.

SOURCE: The Daily Shot Brief August 13, Ninety One (Share Supply, the Coming Issuance Wave)

Add the week's third small lesson, from the after-hours tape: a semiconductor equipment leader delivered a clean beat and sold off anyway, the third such casualty in two days after the chip challenger and the networking giant, and the pattern now has a name in every desk note we read: at the 95th percentile of positioning, expectations are the only competitor that matters. None of these three is a signal on its own. Together they are the market quietly describing its own vulnerability: narrow engines, incoming supply, and no patience for merely good news.

The Rest of the Tape

Credit spent the week confirming #67's reading rather than complicating it. Leveraged loan defaults have eased and high-yield defaults remain low, from the most-watched credit desk on the street, which sits consistently beside the private credit breakout and the frontier spread compression already on file. Institutions, meanwhile, put numbers on their conversion: equity allocations rose 142 basis points in July to 58%, with cash at 16%, the portfolio-level confirmation of the survey bullishness #66 documented. The global broadening keeps adding chapters, with the UK outperforming its G7 peers in the first half and Greece, of all markets, described by the data desk that tracks it as a spectacular turnaround; the advance is now broad enough that its laggards are the story. And the rotation threads this letter has run since spring got their sparks: the software complex woke on deal speculation and the memory chip group on a maker's long-term targets, the #55 and #58 rotations respectively, both still alive under the index surface.

The tape now waits on this morning's retail sales, the first read on whether the consumer that inflation has been squeezing for five months is still spending anyway, and the last major datum before the Jackson Hole gathering at the end of the month. The Fed speakers between now and then will be auditioning arguments; the chair's speech will pick one. Our falsifier calendar is unchanged behind it: the August CPI on September 10 remains the print that decides whether September's hold survives, and the growth clock's exhibits, wages, gasoline, a 5.8% nowcast, are all inputs into that number.

Charts Worth Watching

The embeds below carry the week's rates and breadth story in primary sources: the futures curve that no longer prices a full hike, the everything-at-records tape, the global breadth backdrop the Kospi is leading, and the flow desk's bond exhibit from the week's sharpest thread.

The Verdict

Grade the frame first, as always. #67 said the direction of the range break would name the winning clock; the honest answer is that each market named a different one, and both with conviction: equities broke to records on the disinflation clock while the 30-year auctioned at a quarter-century high on the growth clock. The book was built for exactly this split and stays put: the full broadening book is long the equity verdict, the rate-sensitive shorts are short the bond verdict, and the two positions are no longer a hedge and a core but parallel expressions of the same regime read. The semis sleeve stays trimmed at the 95th percentile, and no new risk goes on ahead of retail sales and Jackson Hole.

The watch list sharpens rather than grows. The yen carry is now the named transmission channel between the two verdicts, which promotes the 160 line from tripwire to first-among-equals: a yen rally is the one event that hits both sides of our book at once, and we treat any sharp move there as actionable before explanation. The VIX-rising-with-highs signature joins the dashboard alongside the #54 vol-regime read; the Paulsen revision-breadth count and the issuance calendar join as the two pre-written reversal exhibits for the floor and the wall respectively; and the Atlanta Fed tracker joins the August CPI as the growth clock's official scoreboard. The liquidity-versus-earnings decomposition stands as the deepest open question against our thesis, with its falsifier written above.

Cleared, then split. The market that chose and the market that refused are both on this page now, and the next witnesses are already scheduled: retail sales this morning, Jackson Hole in two weeks, the August CPI on September 10. Somewhere in that sequence, one of the two verdicts starts converging toward the other, and this letter's entire architecture, the tripwires, the composite, the pre-written exits, exists for the day the convergence begins. #69 grades from there.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The Coil After the Clear

The tape itself is doing something worth recording while we wait. The S&P 500 has gone essentially nowhere since the surge a week ago, and one chart maker measured the stall precisely: the index just printed its smallest five-day range as a percentage of spot since December of last year, a coil that resembles a pennant more than a top. Pennants after advances resolve in the direction of the trend more often than not, which is the constructive reading, but the same compression is what a market looks like when it has run out of imminent reasons to move and is living from data point to data point. Yesterday's issue called the melt-up corridor open; this pattern is the corridor waiting for traffic.

The details inside the quiet are instructive. After Wednesday's close, two strong earnings reports, from an AI chip challenger and a networking giant, sold off anyway, a reminder that in the ninety-fifth percentile of positioning, good numbers can lose a fight with better expectations. Earlier in the week, on a day when only two index components had earnings reactions at all, the technicians noted that a healthcare distributor is trading like one of the strongest stocks in its sector while a flagship mall REIT is being rejected at major long-term resistance, defensives behaving well and rate-sensitives struggling at the ceilings, exactly the split our composite watches. Even the small business optimism jump that led the week's survey data arrives with an asterisk the compilers themselves flagged, a response rate near 9% that makes the signal as thin as the range. A market this quiet, this crowded, and this close to its highs is not resting; it is waiting for instructions.

Growth Is the Inflation Risk

Start with the reframe that organizes everything else. The consensus treats inflation as a shelter story, an oil story, a tariff story; the sharpest macro work of the week argues it is now primarily a growth story. The nowcast is running near 5.8% real GDP with fixed investment accelerating, roughly 40% of the S&P 500's returns over the past year trace back to the AI buildout, and the warning signs are stacking in an unusual configuration: inventory-to-sales ratios are falling while government spending rises, which is what an economy running ahead of its supply looks like. Real interest rates sit at new highs for the cycle and are still not restrictive enough to slow any of it. That is not a policy failure in the ordinary sense; it is what happens when the private sector decides to build a new industrial base in eighteen months.

The market is already trading this thesis even while the indexes coil. Inflation-sensitive equity factors have quietly become the year's leaders: steel and metals names are up 43% year to date and agricultural names 16%, moves that did not wait for any CPI print. Short-end inflation swaps from Japan to the United Kingdom show the same pressure building outside the dollar bloc. Set against this, the disinflation case remains real: core CPI has decelerated on a glide path that, at its current speed, lands near 2% by year end, and yesterday's issue documented the real-time shelter case that policy may already be tight. Both things can be true, a decelerating price index and an accelerating growth pulse, and the tension between them is the single most important spread in macro right now, because growth itself is the inflation risk.

The independent evidence leans toward the growth side. A leading indicator built by a quantitative research shop, with no narrative stake in the AI debate, ticked up again this month and points to a steady expansion, not a stall. Pair it with the July data run this letter has documented, services activity at a nine-month high, revenue growth the broadest since 2021, productivity beating, and the picture is an economy that keeps taking the over. Every falsifier this letter maintains on the inflation side, the August CPI, the oil chokepoint, the wage churn premium, draws its power from this growth engine underneath.

SOURCE: The Daily Shot Brief August 11, Variant Perception (US Growth Leading Indicator, Steady Outlook)

What the Long End Knows

Now the mechanics. The post-payrolls drop in two-year yields, the move the equity market read as relief, decomposes into an uncomfortable message: the decline came almost entirely from the inflation swap side, while real rates stayed pinned at restrictive levels. The long end barely moved at all. Translated from the rates desk dialect: the bond market did not price a friendlier Fed; it priced slightly cooler inflation against an unchanged real economy, and the long end's refusal to rally says it believes the Fed can be significantly more restrictive without generating credit stress. A market that assigns its central bank a high pain threshold is not a market expecting rescue. It is a market that expects the pressure to continue until something in the real economy finally bends.

The same desk supplied the sentence this letter has been circling for three issues, and we quote it directly: "bonds won't bid until the AI capex ends or the credit behind it collapses." The logic follows from the growth section above: an investment boom absorbing capital at this scale keeps real yields elevated mechanically, so the bond bull case and the AI bear case have become the same trade expressed in two markets. The practical instruction is to watch the value chain rather than the announcements, because electricity permits, zoning approvals, and grid buildouts get secured months before any data center press release; the physical pipeline is the leading indicator for the financial one. And the decade-scale version of the same message arrived independently this week: a model that jointly fits equity and bond risk premia now projects secularly low expected returns for the classic 60/40 portfolio, which is what a world of structurally higher real rates and fully priced equities looks like from ten thousand feet. The cyclical desk and the secular model agree, and the long end already knows.

SOURCE: The Daily Shot Brief August 11, Augur Infinity (Affine Stock-Bond Model, Low Expected 60/40 Returns)

We fold the currency piece in because it completes the machine. Rates and foreign exchange are one trade in two markets: the yen's persistent slide, which our tripwire at 160 monitors, is mechanically driven by a finance ministry funding deficits with bills while the central bank pins the short end, a quantity-of-money story that no rate differential model captures. This is why the August intervention was improvised in euros, why it half-unwound, and why the devaluation trend keeps reasserting. In a trending devaluation, the desk notes, buying breakouts beats buying dips; the same discipline that has paid in equities all year is now the stated playbook in currencies.

One more structural note belongs in the file, because it explains why these pressures cannot simply be legislated away. Every central bank operates inside the same constraint map: a country can hold at most two of a fixed exchange rate, free capital movement, and independent monetary policy. China manages its currency behind a closed capital account; the United States runs open capital markets with independent policy and lets the dollar float; Japan is currently discovering what happens when a country tries to blur the edges of all three at once. The asymmetry in those choices is what routes global savings into dollar assets, holds the yen on its devaluation path, and makes the American long end the clearing price for the world's capital. When our tripwires watch the 30-year and the yen together, this is why: they are two gauges on the same machine.

Credit Is Voting With the Bulls

The quietest big development of the week came from the sector this letter has watched since #49 flagged private credit as the cycle's designated stress point. Private credit stocks are staging a comeback: the sector ETF, crushed earlier this year on funding concerns, has broken out to its highest level since mid-February, completing a double-bottom formation above the 62 to 63 dollar zone. This is the first genuinely bullish reading our private-credit trigger has produced since we adopted it, and honesty requires recording it with the same prominence we gave the warnings. The funding-stress scenario, the one the Wall Street Journal was circling two weeks ago, is being priced out, not in.

The global credit tape says the same thing more loudly. Frontier high-yield sovereign spreads, the riskiest lane of the world's bond markets, have tightened roughly 100 basis points over the past year, extending a post-pandemic compression that shows no sign of stress. Set this beside yesterday's flow evidence, the trillion-dollar buyback window and the record ETF inflows, and the picture is coherent: every risk-pricing channel we track, equity flows, corporate credit, sovereign credit, private credit, is voting the same direction at once. We note the cost of that coherence honestly: when every channel agrees, the information value of each falls, and the system's capacity to absorb a surprise falls with it. But the reading itself is not ambiguous, and this letter's job is to report the tape it sees: credit is voting with the bulls.

SOURCE: The Daily Shot Brief August 12, Alpine Macro (Frontier High Yield Spreads, 100bp Tighter in a Year)

So we restate the trigger's terms rather than retire them. The private credit line flips back to warning if the breakout fails and the ETF loses its base, if new fund flows into the sector reverse, or if the coverage cycle turns again from comeback stories to funding questions; the frontier spread line flips if the year-long compression starts unwinding faster than 25 basis points in a month. None of that is present today. But we learned from the RSPD window that a signal is only useful if its failure conditions are written down while it is behaving, and this is us writing them down.

Hyperscalers and Neoclouds

The week also produced the cleanest naming yet of the fault line inside the AI trade. One research house split the buildout into two businesses that the market has been pricing as one: hyperscalers, whose returns on invested capital exceed their cost of capital, making the capex value-creating almost regardless of the cycle's length; and neoclouds, the newer pure-play infrastructure providers whose weaker returns and higher financing costs put them at risk of destroying value with every incremental data center. This maps precisely onto the caboose-and-train framework this letter adopted in #62: the question was never whether the train is real, it is which cars are earning their fuel. Now the fault line has a name, and the fault line runs between them.

SOURCE: The Daily Shot Brief August 12, BCA Research via @WeisbergerNoah (Hyperscalers vs Neoclouds)

Wednesday complicated the bear half of that story beautifully, and we keep the complication on the page. The largest neocloud just doubled its revenue year over year and beat its operating margin guidance by five points, which is what value creation looks like while it lasts; the market paid it with an 18% single-day gain. The chip supplier at the center of everything is structuring its balance sheet with more debt against GPU collateral that keeps appreciating, and the macro desk's read is worth recording against the doomer consensus: rising debt is not the warning sign, asset-liability mismatch is, and lending against appreciating collateral you understand is a company betting on itself. The bear case for the neoclouds is real and structural; the evidence for it, so far, keeps failing to arrive on schedule. That is what the after-hours action punctuated: strong numbers from the challengers still lost to stronger expectations, which is how a market prices a fault line it has finally noticed.

The Rest of the Economy

The labor paradox this letter has tracked since #64 extended again, and the new data deepen it rather than resolve it. July announced job cuts fell to 33,400, the lowest monthly total in two years, with year-to-date cuts down 41% from last year; yet AI led all stated reasons for cuts for the fifth consecutive month, accounting for a third of July's total, and the tracking firm just introduced a new category to catch announcements that allude to automation without naming it. Meanwhile hiring plans jumped 47% on the month to the best July since 2022. Companies are simultaneously not firing, not hiring at the payroll line, planning to hire, and blaming AI when they cut, and the labor paradox extends again.

SOURCE: Neil's Newsletter August 6, Challenger via Neil Sethi (July Job Cuts, Lowest in Two Years)

Housing, the most rate-sensitive real economy sector and the shorts' oldest argument in our book, produced its most interesting report in months. July existing home sales eased a second month to a three-month low, but prices printed the highest July on record, inventories turned negative year over year for the first time since October 2023, and, the detail we flag, the affordability index improved for the first time since January. A housing market with falling turnover, record prices, shrinking supply, and marginally improving affordability is a market frozen rather than broken, and affordability finally caught a break.

SOURCE: Neil's Newsletter August 11, NAR via Neil Sethi (Existing Home Sales and Affordability, July 2026)

The consumer completes the picture with the same mixed signature. Non-mortgage consumer borrowing rebounded in June after May's decline, so the household is neither retrenching nor bingeing, while real weekly earnings have now fallen in four of the past five months, the quiet cost of the disinflation everyone is celebrating. An economy where borrowing grows modestly, paychecks lag prices, and spending holds anyway is an economy running on employment security rather than wage growth, which loops the whole section back to the labor paradox above: the moment companies stop merely not firing and start actually cutting, the consumer loses the only pillar under it. Nothing in the July data says that moment is near; everything in the AI-led cut reasons says it is worth watching monthly.

Abroad, the earnings floor keeps confirming the breadth story from yesterday's issue: Nikkei 225 earnings rose above their trend channel in the second quarter, meaning Japan's record-high stock market, like ours, is being carried by profits rather than multiple alone. The world's advance, in equities and now visibly in earnings, remains broader than the AI corridor that gets the headlines.

Charts Worth Watching

The embeds below are the rates desk's feed in its own words: the bond and AI trades as one position, the balance-sheet answer to the debt doomers, the chart its author calls the most bearish for bonds, and the real-time shelter recalculation that yesterday's issue built its bull case on.

The Verdict

This issue changes the depth of field, not the position. The book stays exactly where #66 left it: full broadening book deployed, semis sleeve trimmed, rate-sensitive shorts repurposed as December and oil-tail insurance, no additions at 95th percentile crowding. What this issue adds is the machinery behind the tripwires. The 30-year and 10y10y, double-weighted since #63, now have their mechanism spelled out: real rates pinned, the long end assigning the Fed a high pain threshold, and the bond bid structurally postponed until the AI capex cycle ends or its credit cracks. The yen 160 line now has its quantity-of-money explanation, which means intervention headlines alone should not clear it; only a change in issuance mix would. And the #49 private-credit trigger produced its first bullish reading, which we accept at face value while noting that a trigger designed to warn is most dangerous in the season everyone stops watching it.

Two clocks are now running against each other, and naming them is this issue's contribution. The disinflation clock says core CPI glides toward 2% by year end and the hiking cycle dies of natural causes. The growth clock says a 5.8% nowcast with accelerating fixed investment eventually forces the inflation question back open, through wages, through metals, through the swap curves already moving abroad. The pennant on the index chart is the market refusing to choose between them, five days of the tightest trading since December while the crowd, the flows, and the credit market all lean bullish underneath. Our falsifiers are unchanged and dated: the August CPI on September 10, the yen at 160, the equal-weight relative line, and now this morning's PPI as the first checkpoint. When the range breaks, the direction of the break will tell us which clock the market believes, and this page, as always, will grade the answer against what we wrote while it was still unknown.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The short version of the grade: the wall won before the floor was needed, the valuation coil sprang upward, the hedge kept its own counsel, and the crowd got bigger. Small caps, which led the declines on Monday as the designated first casualty of a hot print, led the advance on Wednesday and closed at a record high. That sequence, first out the door in the rehearsal, first through the tape on the day, is the cleanest single validation the broadening book has produced since we began building it in #59. The rest of this issue is the detail, plus the risks that did not retire with the print, because cleared is not the same as clear.

The Number Came In Clean

Composition before celebration. Shelter did roughly two thirds of the month's lifting, and owners' equivalent rent re-accelerated inside it, so the sticky part of the index is still sticky. Core goods rose 0.2%, and inside that line computers and smart-home equipment jumped 3.52% on the month, the second-largest move in the history of the series and the clearest tariff fingerprint yet to surface in the data. Real weekly earnings were negative for the fourth month in five, which is disinflation of the uncomfortable kind, the kind that arrives partly because the consumer is tiring. This was not a perfect print. It was a sufficient one, and in a market priced for a coin flip, sufficient is everything.

The repricing was immediate and orderly. September hike odds fell from 50% to about 42%, full-year hike odds slipped from 79% to 74%, and the New York Fed president spent the afternoon acknowledging that markets are drifting toward December as the live meeting. Read those two numbers together and the message is precise: the market did not cancel the hike, it postponed it. The hawks lost their September case to a 0.2% core print, but three dissents are still on the record, four years of above-target inflation still sits in the memo line, and one more CPI report lands before the September decision. The chair who promised data dependence got a data point, singular.

The strongest bull exhibit of the week argues the official number overstates what inflation is even left. Swap the CPI's famously lagged shelter measure for real-time rent data and the recalculation lands headline inflation near 2.5% and core near 1.5%, effectively at target. The lag argument has been made all cycle, but it matters most precisely now, in a month when shelter was two thirds of the increase: if the official series is simply late, then policy is already restrictive against the economy that exists, the hiking debate is an argument about an artifact, and Wednesday's print was cooler than it even looked.

SOURCE: Daily Chartbook DC Lite #660, WisdomTree via @jeremydschwartz (Real-Time Shelter Data Puts CPI at 2.5% and Core Near 1.5%)

The Tape Voted With Breadth

The reaction told us more than the number did. The Russell 2000 rose 0.6%, led every major index, and closed at a record high; the equal-weight S&P printed another all-time high of its own; the headline index added 0.3%, the Nasdaq half a percent, and eight of eleven sectors advanced. Semiconductors rose 2.5% on top of it, powered by an AI infrastructure name that doubled its revenue and beat its margin guide by five points, and a server maker up 17% in sympathy. Monday's rehearsal had shown which trades a hot print would sell first; Wednesday showed the same trades leading the relief. The most rate-sensitive corner of the book, and our freshest addition, was first through the tape in both directions, which is exactly what a well-specified position should do.

Now the housekeeping this letter owed its readers. The cyclical-versus-defensive spread has gone five issues without a fresh direct reading, and we said its resolution would be the first task of the verdict issue. Here is the honest resolution: the direct series remains absent from our free sources, so we are retiring the apology and replacing the instrument. From this issue forward the cyclical-defensive read is a defined composite we can actually observe weekly: the small-cap relative line, the equal-weight relative line, and sector breadth. On Wednesday that composite read unambiguously cyclical: small caps leading at a record, equal weight at a record, eight of eleven sectors green with rate-sensitives participating. A signal we can see beats a purer signal we cannot, and the debt is closed.

The breadth is not an American story, which is the strongest part of it. More than 80% of country ETFs are now trading above their 50-day moving averages, and the observation attached to the chart is the one that matters: the boom is not confined to the economies capturing the chip and AI buildout. Stocks are rising in places with no hyperscaler to their name. Layer on the survey of nearly 300 institutional managers running more than three and a half trillion dollars, which rose in August to its most bullish reading since January, and the picture is a global advance with professional sponsorship rather than a narrow domestic squeeze. Our composite says cyclical at home, and the whole world is participating.

SOURCE: Daily Chartbook DC Lite #659, @thestalwart (Over 80% of Country ETFs Above the 50-Day Average)

A Trillion Dollars of Standing Bid

Grade the wall first, because #65 leaned on it hardest. The eve-of-print framing was that record flows were the bull's standing defense against a hot number, and the verdict made the question moot: the wall never had to defend, because the number came in cool and the wall simply advanced. But the exhibit that matters for the next month arrived the same morning. The corporate buyback window reopens this week with more than $1 trillion of announced authorizations behind it, the largest amount on record at this point in the calendar, and the flow desk publishing the figure expects buybacks to exceed equity issuance from here, absorbing net supply outright.

Stack the wall's layers as they now stand: cumulative 2026 equity ETF inflows running beyond any historical comparison, a corporate bid moving from 85% eligible to fully open with record authorizations, retail adding a third consecutive month, and institutions at their most bullish since January. Every layer is price-insensitive or trend-following, and every layer got a green light from the same print. With the September binary cleared, the melt-up corridor this letter opened in #63 has its fuel line attached: the same flows that were a defense on Tuesday are a runway on Thursday. The two caveats from #65 stay filed where we left them, the most patient allocator in the market still building cash and the private credit coverage still darkening, but neither changes the near-term arithmetic of a trillion dollars of standing bid.

SOURCE: Daily Chartbook DC Lite #660, Citadel Securities via Scott Rubner (Buyback Window Reopens, Over $1 Trillion Authorized)

Both Crowds Are Buying Again

Start with the crowd this letter has graded hardest. Back in #60 we scored institutions as the reluctant participant, fully aware of the melt-up case and unwilling to chase it. That reluctance is now measurably over: the monthly survey of institutional managers, nearly 300 of them running funds in excess of $3.5 trillion, printed its most bullish headline since January 2026. Professional money spent the spring being stopped into strength and has decided, at record highs, that the path of least resistance is participation. We note without cynicism that this is how every durable advance matures, and also how every top eventually forms; the reading is fuel first and warning second.

Retail never needed convincing. The Schwab trading activity index rose for a third straight month to its highest level since January 2022, with clients net buyers of equities and ETFs by two to one through July's chop, a ratio the survey's authors read as firm conviction among self-directed traders. Remember that this cohort bought the July lows, bought the tariff scare, and bought the AI credit wobble, and has been paid every time; the dip-buying reflex is now four wins deep this year. For the first time since early 2022 the fast crowd and the slow crowd are leaning the same way at the same time, and both crowds are buying again.

SOURCE: Daily Chartbook DC Lite #659, Charles Schwab (Trading Activity Index, Highest Since January 2022)

The cost of the convergence is concentration, and we keep it on the page deliberately. The positioning work embedded below puts mega-cap growth and technology exposure at the 95th percentile, firmly overweight, though still shy of the June and October extremes. The Techening flag from #64, implied tech allocations above 50%, has not resolved; it has compounded. Crowding is fuel on the way up and glass on the way down, and nothing about Wednesday changed the un-broadening alarm we wired in #64: if the melt-up proceeds by re-concentration rather than broadening, the equal-weight relative line trips first and the small-cap add is the designated first cut. On Wednesday the alarm read the opposite of tripped, with equal weight at a record. We watch it weekly all the same.

The Opposite of 1999, For Now

The clock debate that has run through this letter since #63 received its cleanest arbitration yet, from the analog work at Fidelity this letter has long respected. The 1999 comparison fails on the one dimension that matters most: by 1999 the earnings peak was already in place, and every point of upside that followed came from multiple expansion, speculation running unsupported by the underlying cash flows. Today the configuration is reversed, with earnings doing the driving and the multiple riding along. The chart's own parenthetical carries the discipline, for now at least, and we adopt it as our own. The clock from #63 said the hour was late; this chart says the engine is different, and the difference is the opposite of 1999, for now.

SOURCE: Daily Chartbook DC Lite #659, Fidelity via Jurrien Timmer (1999 Was All Valuation, Today Is the Opposite, For Now)

Wednesday supplied the live demonstration within hours. The AI infrastructure name that led the semiconductor rally did not rise on a narrative; it doubled revenue year over year and beat its operating margin guide by five points. The quarter behind us delivered the broadest revenue growth since 2021 across all eleven sectors, record margins in ten, and calendar-year estimates still being revised upward against their usual seasonal decay. The hyperscaler free cash flow trough that #64 documented now has its first confirming data points. A market can be expensive and earnings-driven at the same time; only one of those conditions has historically ended cycles on its own.

The runway argument arrived from an unexpected source: corporate card data. In July the top 1% of American businesses spent a median of $7,400 per employee on AI; the top decile spent $650; the median firm spent $11.95. Read that pyramid again, because it is the entire addressable-market case in three numbers: the most aggressive percentile is spending six hundred times the median, which means AI adoption at the typical American company has barely left zero. The counties hosting the buildout are already seeing the housing, employment, and property-value gains documented in the embed below, but the demand side of the trade, actual business adoption, remains at the base of the curve. Whatever one believes about the multiple, the adoption curve is barely started.

SOURCE: Daily Chartbook DC Lite #660, Ramp via Ara Kharazian (AI Spend Per Employee, the Top 1% vs the Median Firm)

What Is Still Armed

The print cleared September; it did not clear the board. Full-year hike odds still sit at 74%, the market is drifting toward December as the live meeting, and one more CPI report, the August print, lands before the September decision with three hawkish dissents already on record. This issue goes to press on the morning producer prices print, a number that reaches the Fed's preferred inflation gauge more directly than CPI does, so the first grade on this issue's optimism may arrive before some readers open it. We flag that timing rather than hide from it; the frame is built to be tested.

The inflation tail now runs through a chokepoint, not a model. Oil sits near $83 with the strait still closed, and one rates strategist put the problem plainly this week: "the Fed has no control over the Strait of Hormuz." Energy has been the swing variable in every hot-print scenario since June, the tariff fingerprint in core goods is no longer subtle, the AI buildout itself is emerging as an inflation impulse that the more careful commentary refuses to ignore, and the churn premium in wages that #65 filed, job-changers earning 7% raises against 4.4% for stayers, is still filed. None of these is a September problem anymore. All of them are December problems, which is exactly what the futures curve is saying.

Grade the hedge while we are here, because #65 promised it. Gold entered the print after a miners signal seen five times in seventy years and record central-bank accumulation, and we wrote that the bid was about the regime, not the number. The print came in cool and the regime thesis did not blink: the buyers accumulating at records were never trading Wednesday morning, and nothing in a 0.2% core print changes a currency system improvising yen defenses, a strategic reserve at 1983 levels, or a fiscal trajectory nobody defends. The hedge stays graded as a regime position, held alongside the equity book rather than against it, and the miners signal keeps its 2016 precedent on file until price says otherwise.

Two positioning reads complete the risk file. Speculators are now heavily long the US dollar, a crowding that has historically preceded dollar weakness more often than strength; a softer dollar would be fuel for the global breadth trade and the gold regime alike, but the unwind of a crowded position is rarely orderly, and the yen line at 160 remains on our board from #63. And the rate coil from #65 is only half unsprung: the September scare has passed, but rate volatility remains priced near lows, the private sector remains structurally short rates through its liabilities, and the long end has not withdrawn its bear-steepening argument that policy is not restrictive enough. A December repricing would find the same dry tinder September never lit. The 30-year and the 10y10y stay double-weighted on the dashboard.

Charts Worth Watching

Four threads from the week's feeds, each worth the click: the positioning meter behind section four, the dollar crowding flag, the data-center county evidence, and the manufacturing pulse that quietly ran to a four-year high this month.

The Verdict

The scorecard, graded against conditions written before the answer: the falsifier did not fire, and September is priced back to a hold. The floor was never tested. The coil sprang upward, through the valuation spring rather than the rate spring. The hedge traded on the regime rather than the print, as we said it would. The crowd got bigger on both the retail and institutional side, and the wall got a trillion dollars deeper the same morning. The small-cap add that led Monday's rehearsal lower led Wednesday's tape to a record high, the un-broadening alarm read untripped with equal weight at its own record, and the five-issue cyclical-defensive debt is closed, replaced with a breadth composite we can actually observe. Frameworks earn their keep on the days the answer arrives, and this one did.

The book changes nothing, which is again the decision. The full broadening book stays deployed with the small-cap add now validated at a record; the semis sleeve stays trimmed, because 95th percentile positioning is an argument against adding to the crowd, not a reason to fight it; and the rate-sensitive shorts stay on, formally repurposed from a September hedge to a December and oil-tail hedge, sized for insurance rather than conviction. No new risk gets added into this morning's PPI. The tripwires stand with one amendment: the #54 vol-regime read first, the equal-weight relative line as the un-broadening alarm, the 30-year and 10y10y through the issuance channel, the yen at 160, and the falsifier slot that July's CPI just vacated now belongs to the August print on September 10.

Cleared, not clear. The number gave the melt-up corridor its clean lane and the flow wall its reopened bid, and the same morning left oil lockstepped to a closed strait, positioning at the 95th percentile, a dollar crowd leaning one way, and a December meeting gathering everything September dropped. Both readings still live on this page, which is where this letter keeps them; #67 will grade this one the way this one graded #65, against conditions the reader can check while the answer is still unknown.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

Monday was useful precisely because it was bad in an instructive way. The day's losses ran straight down the fault-line map this letter has drawn since #62: rate-sensitive sectors led lower, semiconductors gave back 3% of last week's snapback, and small caps, our freshest addition, led the declines. Nothing broke; the index slipped a tenth of a percent. But the market showed us, one day early, exactly which trades a hot CPI liquidates first. Forewarned is forearmed, and the book goes into the print with its hedges described below and its exit sequence already written.

One More Day

Frame the stakes before the details. The bull market that began in October 2022 is up 116.9%, which sounds extended until it is ranked: fifth of the eight bull markets since 1966, behind runs that doubled and tripled before they died. The strategist who published the ranking attached the sentence that best captures the bull half of tomorrow's decision tree: investors who believe this market has run too far should look at what running too far can actually look like. A cool CPI does not merely avoid a hike; it reopens the melt-up corridor that the analog work says is the historical norm after a blowup clears positioning, with a flow wall behind it that we quantify below. This is what running too far can look like.

SOURCE: Daily Chartbook DC Lite #658, Yardeni Research (Eight Bull Markets Since 1966, This One Ranks Fifth)

The pricing math sharpens the picture. The futures curve carries roughly 27 basis points of hikes for the rest of the year, September sits at a coin flip that Monday nudged back above 50%, and the commentary into the print splits exactly along the line this letter has tracked since the Warsh minutes: one camp argues the jobs report bought the Fed patience and a cool CPI seals a hold, while the Macquarie desk offered the uncomfortable counter that even a low print might not be enough to negate the hawks, whose concern is four consecutive years of above-target inflation rather than any single month. A chair who has promised to be more data-dependent than his predecessor meets the single most-watched datum of his tenure tomorrow, with three governors already on record dissenting hawkish. The reaction function is as uncertain as the number.

The bear half of the tree is equally specific. A hot print does not merely price a September hike; it detonates in a market where, as the sharpest macro letter of the weekend put it, everyone is structurally short interest rates through mortgages, corporate debt, and leverage, while rate volatility is priced near historic lows. The asymmetry of tomorrow is not in the equity indexes; it is in the rates complex, where insurance is cheap, positioning is one-sided, and the long end has spent a month telling anyone listening that it thinks the Fed is not restrictive enough. The stakes are asymmetric on both sides of the same number, which is what makes it the print of the quarter.

The Floor

What limits the damage if tomorrow goes wrong is the same thing that has limited every drawdown since June: the earnings cycle refuses to cooperate with the bears. The newest exhibit is the calendar-year revision pattern. Earnings estimates for a given year almost always decline as the year approaches and decline further once it begins; it is one of the most reliable patterns in sell-side behavior. Right now the opposite is happening: calendar-year estimates are being revised up, on top of a quarter that delivered the broadest revenue growth since 2021 across all 11 sectors, record margins in 10 of them, and guidance momentum at all-time highs. An earnings cycle accelerating into a policy scare is the specific configuration in which CPI-driven selloffs have been bought all cycle, because lower prices against rising estimates simply reload the valuation argument. The revisions refuse to fall.

SOURCE: Daily Chartbook DC Lite #658, Truist via Josh Schafer (Calendar-Year EPS Revisions Rising Against the Usual Pattern)

The floor has a wage-side complication worth flagging before the print rather than after. The ADP wage data show pay growth for job-changers running at 7% against 4.4% for job-stayers, the widest churn premium in nearly a year. A reopening gap between switcher and stayer wages is historically an early symptom of labor-cost pressure returning, and it sits awkwardly beside the frozen-hiring narrative: companies are not adding heads, but they are paying up for the heads they poach. If tomorrow's CPI runs hot, expect the hawks to cite exactly this; if it runs cool, the gap becomes next quarter's problem instead. Either way it is the first genuine wage-inflation seed this letter has had to file since the spring, and we file it in the floor section deliberately, because margin cushions erode from the cost line first.

We are explicit about what the floor is not. It is not protection against the rates complex repricing, which flows through multiples regardless of earnings; it is not protection for the crowded cohort, which trades on liquidity first and fundamentals second in any shock; and it did not stop Monday's rate-sensitive selling. The floor is an argument about depth and duration, not direction: hot-CPI selloffs in accelerating earnings cycles have historically been violent, shallow, and short. Our book is positioned for that shape, not for immunity.

The Coil

The most compressed spring in the market right now is not in equities at all. The weekend's macro work laid it out in three linked claims: interest rate volatility is priced way too low; the entire private sector is structurally short rates through its liabilities; and the bond market's own behavior, bear steepening on the down moves, says the long end believes the Fed is not restrictive enough. Add the detail that the post-payrolls rally left bond futures sitting just under a defined resistance zone in a positive stock-bond correlation regime, and tomorrow's print lands on a rates market that is coiled, complacent, and one-sided all at once. If CPI runs hot, the shock transmits through rate volatility first, and cheap insurance everywhere reprices together. That is the mechanism by which a single inflation print becomes an every-asset event, and it is why we have kept the rate-sensitive shorts on through six issues of being early.

One structural detail makes the rate coil nastier than its 2023 ancestors: the stock-bond correlation is positive. This letter first flagged the death of the bond hedge in #58, when a 20% equity drawdown bought only a 3% bond rally, and the weekend's macro work confirms the regime holds: the post-payrolls move was a short-term top in bonds and equities together, and the resistance zone in bond futures is now the line that defines rate risk for both markets at once. In a positive-correlation regime, a hot CPI sells stocks and bonds simultaneously, the classic 60/40 portfolio has no working hedge, and the only assets that historically absorb the flow are cash, the dollar, and gold. Note which of those three is already being bid two sections below.

The opposite coil sits in the fundamentals of the crowded sector itself. Technology spent two months going sideways while its earnings estimates climbed relentlessly, and the arithmetic result is that tech is now cheap on a forward P/E basis relative to its own recent history, cheaper into the teeth of it. Two springs, wound in opposite directions, and one number releases one of them: a cool print springs the valuation coil upward through a re-crowding cohort with fresh estimate fuel; a hot print springs the rate coil through a structurally short system with no insurance. We do not pretend to know which; we position for the asymmetry and let the print choose.

SOURCE: The Weekly ChartStorm 9 August, @MacroCharts via Callum Thomas (Tech Forward P/E Cheapened Through the Range)

The Oldest Hedge Is Bidding

While the equity market debates tomorrow, one corner of the market has quietly stopped debating. The gold miners index just posted a weekly gain of a magnitude seen only five other times in more than 70 years; the most recent precedent, in 2016, was followed by an 85% gain in the index over the next six months. Gold itself printed its best week since January, breaking out against a dollar at two-month lows, on top of the record central-bank accumulation this letter documented in #63, 289 tonnes in a quarter, five times the prior pace. The oldest inflation hedge in the book is being bid by the most patient buyers in the system in the same week the equity market priced inflation risk at a coin flip. The oldest hedge is bidding.

SOURCE: Daily Chartbook DC Lite #658, Turning Point via Dean Christians (Gold BUGS Weekly Gain, Five Times in 70 Years)

The metal's divergence from its digital rival completes the picture. Bitcoin, the cycle's designated speculation thermometer, shows upside implied volatility at the lowest level in its recorded history, near 23%, with the call bid simply gone; the same week, hedge funds flipped net long Bitcoin futures for one of the few times on record, a positioning oddity we read as basis-trade unwind rather than conviction. Speculative energy has not left the system; it has rotated, out of the digital debasement trade and into the metallic one, at precisely the moment central banks are accumulating at records. When the fast money and the slowest money in the world converge on the same asset from opposite directions, the asset is usually telling you something about the denominator.

We resist the temptation to read gold as a verdict on tomorrow's number; gold's bid is about the regime, not the print. Central banks accumulating at records, a currency system improvising yen defenses in euros, a strategic petroleum reserve at 1983 levels, and a fiscal trajectory nobody defends: the debasement thesis does not need Wednesday to be right. But the timing matters for equity investors in one way: a market where gold, the miners, and the dollar are all moving like this is a market already paying up for the tail that the rates complex is pricing at lows. Somebody is wrong about the tail, and tomorrow adjudicates the first round.

The Winners of the Last Dip

The cohort with the most riding on the print is the one that just won. The Magnificent 7 complex and its equal-weight twin snapped back from the July lows fast enough that our published downtrend trigger never fired, another victory for dip-buyers in a cycle full of them, and the winnings were immediately redeployed: implied tech ETF allocations crossed 50% last week for the first time. Monday supplied the caution: semiconductors fell 3% on nothing more than a yield backup and an oil spike, because the most crowded cohort is also the most rate-sensitive cohort, long duration in its valuations and long liquidity in its ownership. The winners of the last dip.

SOURCE: The Weekly ChartStorm 9 August, MarketCharts.com via Callum Thomas (Mag-7 and Ex-Mag-7, the Snapback From the Lows)

Even the geopolitical tail carries a note of habituation worth recording: each Iran flare-up, as one portfolio manager observed Monday, has arrived with smaller market impact than the last, and the latest exchange, reparations demanded and counter-demanded, moved oil 6% but the index barely a tenth. Habituation is comfort until the flare-up that is not smaller, and the strait odds still sit under 10% for September.

The instruction for tomorrow follows from #64's alarm design. If the print is cool and the melt-up corridor opens, watch where the crowd's winnings go: into the equal-weight advance, which extends the broadening and validates the full book, or into a further narrowing of the same 50%-plus allocation, which starts the un-broadening film. If the print is hot, watch the reclaim behavior in the crowded names, the same reference-line discipline that called July correctly: dip-buyers defending their levels on a hike scare would be the strongest evidence yet that this market's underlying bid is structural rather than rented.

A Flow Wall Meets a Number

The last exhibit is the wall the number has to move. Cumulative 2026 equity ETF flows are blowing away every historical comparison, in the flow-tracker's words, and the corporate bid is back to roughly 85% eligibility after blackout, the two price-insensitive buyers from #64 now fully deployed. Even the speculative fringe is re-engaging in an unusual way: hedge funds just flipped net long Bitcoin futures for one of the few times on record, abandoning the basis trade that keeps them structurally short, while Bitcoin's own implied volatility sits at historic lows with the call bid simply absent. Positioning that was washed out in July is being rebuilt in August across every channel we track, and a flow wall meets a number.

SOURCE: Daily Chartbook DC Lite #655, Strategas via Todd Sohn (2026 Equity ETF Flows Blow Away Any Historical Comparison)

Two quieter flags from the week's reading belong in the record before the print, because both bear on how much wall is really there. The Financial Times documented a shift in Berkshire's equity buying alongside its cash levels, and the most patient capital allocator in the market building cash into record flows is a sentence that needs no elaboration. And the Wall Street Journal returned to private credit, the sector our #49 trigger has watched all cycle, with the sort of coverage that precedes repricing more often than it follows it. Neither is actionable tomorrow; both are reminders that the flow wall is made of participants, and the two most famously careful ones are leaning the other way.

The honest framing of tomorrow, then, is a collision test. A cool print and the wall becomes a runway: record flows, restored buybacks, cheap-on-growth fundamentals, and a re-opened melt-up corridor, with our whole deployed book participating. A hot print and the wall becomes the bull's only defense against the rate coil, and we learn whether record passive inflows and corporate bids can absorb a September hike repricing in a market with no volatility insurance. Either way the answer arrives at 8:30 tomorrow, and either way this letter's framework was built for it: the tripwires are numbered, the exit sequence is pre-written, and the floor, the coil, the hedge, the crowd, and the wall are all on this page in advance. One housekeeping item closes the file: the RSPD/RSPS window inherited from #54 expires this month without ever firing, through two stalls and a breakout, and we retire it formally with thanks; signals that do not fire are also information.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The eve-of-print summary is short by design. September hike odds re-crossed 50% on Monday's oil spike, making tomorrow's CPI the live test of the #63 falsifier list. The downside cushion is an earnings cycle revising calendar-year estimates up when they always fall; the upside corridor is the fifth-ranked bull of eight with a record flow wall behind it; the tail risk is a rates complex priced at volatility lows while the whole private sector sits structurally short rates and the long end argues the Fed is not restrictive enough. Gold and its miners are being bid like the regime question is already settled; the crowded tech cohort got cheaper on forward earnings even as its ownership crossed 50% of implied allocations; and Monday rehearsed, in miniature, exactly which trades a hot print sells first.

Positioning into the number is unchanged, which is itself the decision: the full broadening book stays deployed, the fresh small-cap add is the designated first cut if the alarm trips, the semis sleeve stays trimmed, the rate-sensitive shorts stay on as the direct CPI hedge, and no new risk gets added on the eve of a binary. The tripwires stand as written: the #54 vol-regime read first, the equal-weight relative line as the un-broadening alarm, the 30-year and 10y10y double-weighted through the issuance channel, the yen at 160, and, retired with this issue, the RSPD/RSPS window that never fired. The cyclical-versus-defensive spread remains overdue for a fresh reading, five issues running; it is first on the list for the verdict issue.

We hold both reads on the same page one more night. Tomorrow morning, one of them gets the tape, and #66 will grade the answer the way this letter graded Alphabet, the Fed, and the peaks report before it: against conditions written down while the answer was still unknown.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The update has two parts, and they are the two hands of the same clock. The hour hand, the one #63 set to late 1998, still points at an eventual reckoning: tech allocations just crossed 50%, the crowding is rebuilding, and the analog that put the fund blowup on the LTCM mark also schedules a top two analog-years out. But the minute hand, the one that pays this quarter, points straight up: the broadest revenue growth since 2021, forward multiples cheaper than three-quarters of recent breakouts, hyperscaler free cash flow at what the sell side now calls its trough, and a formerly all-in market that spent July getting washed out and is only now coming back. We promised in #63 that the small-cap add would come off pause on equal-weight confirmation. The confirmation printed. This issue executes it.

Confirmation

Start with the tape fact that settles the #62 and #63 argument sequence. The equal-weighted index has been steadily pushing higher with 200-day breadth trending up, and now the cap-weighted index, after two months of consolidation and one full semiconductor washout, is also out to new highs. Both averages agree for the first time since early June, and the disagreement between them was the entire architecture of the stall debate: when only the median stock rises, bears call it a last gasp; when only the giants rise, bears call it narrowing. When both rise together off a washed-out base, there is no bear reading of the pattern, only a valuation argument about how far it carries. Both averages agree at last.

SOURCE: The Weekly ChartStorm 9 August, MarketCharts.com via Callum Thomas (Equal-Weight and Cap-Weight, Both at New Highs)

What makes this breakout different from a mere bounce is what July's washout left underneath it. At the lows, positioning in the crowded index was the cleanest in years: Nasdaq futures long-short at a 17-year low, volatility ownership at the first percentile, retail net sellers for weeks, hedge funds finishing a record eight-week tech liquidation. Breakouts that start from crowded positioning fail because there is nobody left to convert; breakouts that start from washed-out positioning run because every new high forces underexposed money back in. This one starts from the second kind of base, with the buyback bid returning to nearly full strength in the same fortnight. The wall of worry was rebuilt in July, brick by brick, and the market is now climbing it on schedule.

The whiplash in expert opinion between July 26 and August 9 deserves its own paragraph, because it is a lesson this letter should keep. Two weeks ago the weekend review counted four simultaneous peaks and warned of a larger correction phase; this weekend it counts a global bull market and a path of least resistance pointing higher. Both readings were honest descriptions of the data in front of them, which is precisely the point: levels peaked, trends did not, and our own tripwire discipline, which required a broken trend rather than a stalled level before de-risking, is the only reason the book stayed long through the noise. The lesson is not that the reviewer was wrong; it is that in a dispersion regime, peak-counting fails and trend-following pays. We keep the tripwires and drop the peak-counting.

Execution follows the rules we published. The small-cap add, paused since #60 across four issues, goes on with the equal-weight confirmation in hand: the neglect thesis from #59 is intact, the fund-flow base never crowded, and the cohort now gets the broadening tailwind plus the buyback bid plus a Fed whose hike odds are fading. The semis sleeve stays trimmed; a two-day snapback does not un-crowd a fifth of the market.

The Global Tape Agrees

The strongest new fact of the weekend is that this is no longer an American story. Forty percent of the world's equity markets printed new 52-week highs this week and zero printed new lows, the widest global breadth reading of the cycle. Europe's STOXX 600 spent last week setting records of its own. And the relative-value context makes the global signal a US tailwind rather than a US indictment: American valuations have spent two months compressing against the rest of the world as prices stalled under rising earnings, so the US enters this global upswing cheaper relative to its peers than it has been in a year. Forty percent high, zero low.

SOURCE: The Weekly ChartStorm 9 August, Topdown Charts Professional via Callum Thomas (40% of Countries at New 52-Week Highs, Zero at New Lows)

The cross-asset tape supports the same reflation reading. Gold just posted its best week since January and is breaking out against a dollar sitting at two-month-low closes, with global money supply rising again after a long pause; the liquidity models that track it now argue gold is worth more than its recent range, and rising global liquidity is historically the tide that floats non-US equity breadth exactly as we are seeing it. Copper hovers near record territory. This is not the configuration of a world sliding into the hard landing the payroll number might suggest; it is the configuration of a world being re-liquefied while its largest economy runs a private-sector capex boom, and equities everywhere are pricing the second story.

Global breadth belongs in the same analytical bucket as domestic breadth, and the bucket keeps saying the same thing. Bull markets die narrow: the 2000 top and the 2007 top were both preceded by months of shrinking participation at home and divergence abroad. A tape where two in five world markets make annual highs simultaneously is the opposite configuration, and it pairs with the lending data, where the Fed's own loan-officer survey shows banks loosening standards for large and mid-sized firms. Credit that eases into an equity advance is fuel; credit that tightens into one is the warning we watch for. Nothing in this weekend's file tightens.

Revenue Does the Talking

The earnings season heading into its final innings settled on a statistic that outranks every sentiment reading in this letter: with 88% of the index reported, second-quarter revenue growth is on pace for its largest year-over-year increase since the fourth quarter of 2021, and all 11 sectors are growing their top lines. Earnings can be engineered, margins can be flattered, guidance can be managed; revenue is the hardest line on the statement to fake, and revenue is doing the talking. Stack the quarter's records in one place: the broadest revenue growth in nearly five years, margin expansion in 10 of 11 sectors, guidance momentum at all-time highs, and third-quarter estimates that rose during a month when they historically fall. Eleven sectors, one direction.

SOURCE: Daily Chartbook DC Lite #657, FactSet via John Butters (Q2 Revenue Growth, Largest Since Q4 2021, All 11 Sectors)

The labor paradox from #63 developed a second layer this weekend, and it belongs in the revenue section because the two are the same story. The hiring pop has, in the words of the economics desk that tracks it, largely evaporated: 20,000 jobs a month over three months, a negative July, deep revisions. Yet announced hiring plans just printed their strongest July since 2022 and their best year-to-date since 2023, while announced cuts sit at two-year lows. Companies are planning to hire and not yet doing it, firing almost no one, and meanwhile growing revenue at the fastest pace in five years. That is what an output boom without a labor boom looks like in its early innings, and it is why the margin records keep printing: the revenue line is compounding while the cost line waits. Whether that gap eventually closes with hiring, with wage pressure, or with the displacement scenario nobody has priced is the deepest question in this letter's file, and this quarter it is being answered, temporarily, in the shareholders' favor.

The valuation counterpart keeps the breakout honest. Forward multiples are on the high side of history, and nobody at this desk will pretend otherwise; but they are lower than they were at three-quarters of the breakouts of the past five years, because the sideways summer let earnings catch up to price. Breakouts from earnings-led resets have a materially better record than breakouts from multiple expansion, and this one belongs to the first family. The #62 statistic that started this thread, the cheapest price-to-growth in three decades, has now been joined by its price-action confirmation.

SOURCE: Daily Chartbook DC Lite #656, Bloomberg via Jonathan Levin (Forward P/E Lower Than in Three-Quarters of Recent Breakouts)

Cash Flow Finds Its Trough

The caboose-and-train question that organized #62 and #63 received its most specific answer yet, and it came with a date attached rather than a verdict. Goldman's estimate work now projects hyperscaler free cash flow to come back to life over the coming quarters, from a trough where only Microsoft currently generates positive free cash flow. That sentence contains both the bear case and the bull case in their purest forms. The bear case: three of the four largest companies on earth currently burn cash, something no index concentration this large has ever been built on. The bull case: the burn is the trough, the cloud revenue lines just accelerated through it, and if the estimates are right, the market spent July repricing a cash-flow valley that was already ending. Cash flow finds its trough.

SOURCE: Daily Chartbook DC Lite #656, Goldman Sachs via @mikezaccardi (Hyperscaler Free Cash Flow, Expected to Come Back to Life)

Our synthesis stays where the evidence puts it: this is now a timing question, not a viability question. The credit market that widened hyperscaler spreads in July has calmed with the macro; the equity market that sold Alphabet's capex guide bought the same complex back at better prices; and the revenue acceleration means the capex-to-revenue bridge from #62 is being crossed, not just promised. What would reopen the viability question is explicit: a second AI credit event, a cloud deceleration next quarter, or capex guides that rise faster than the revenue that justifies them. Until one of those prints, the caboose keeps its train.

The Techening

Now the seed inside the rally, because every issue of this letter carries the dissent forward. Investors are piling back into technology at a record pace, and implied allocations to tech ETFs have surpassed 50% for the first time. Retail's changed behavior is part of the story: households are still net sellers of single stocks but are committed buyers of tech through ETFs, a pattern shift that concentrates flows into the exact cohort that was crowded in July. The dip-buyers just won again in the Mag 7, and victories teach. Two issues ago the market punished crowding; this week it is rebuilding the same crowd at higher prices, and the crowd is back in tech.

SOURCE: The Weekly ChartStorm 9 August, Topdown Charts via Callum Thomas (Implied Tech ETF Allocations Surpass 50%)

The flow calendar amplifies whatever the crowd chooses. Buyback eligibility reaches roughly 85% of the index this week as the last blackout windows expire, restoring the corporate bid at full strength directly into the re-crowding retail bid. Two large price-insensitive buyers arriving in the same fortnight is how melt-ups mechanically start; it is also how blow-off tops mechanically start, and the difference between those two sentences is nothing but what the money buys. If the combined bid spreads across the equal-weight complex, the broadening gets its strongest month yet. If it funnels into the same 50%-and-rising tech allocation, the melt-up arrives pre-narrowed.

This is the mechanism by which the melt-up scenario carries its own ending, and it is worth stating as a process rather than a prediction. If the advance from here runs through re-concentration, tech allocations climbing from 50% toward the extremes while the equal-weight advance flattens, then the broadening that justified our whole book unwinds from the inside, and the 1999 phase of the analog begins in earnest: narrowing leadership, vertical prices, deteriorating internals under record indexes. None of that is today's tape; today's tape has both averages at highs together. But the watch instruction is now written: the equal-weight-versus-cap-weight relative line, which spent the summer as our confirmation signal, becomes from here our un-broadening alarm. Volatility agrees that vigilance is cheap: the VIX-level history shows current readings deliver decent median forward returns, but the third quarter's seasonal surprise window is open through October.

Both Hands of the Clock

The analog debate now has both hands visible, and the honest position holds them together. The hour hand is #63's: the blowup on the LTCM mark, the yen intervention rhyme, the crowding rebuild, all pointing at a reckoning that history schedules roughly two analog-years from the ChatGPT moment. The minute hand arrived this weekend from the same analog: overlay the indexes on the Netscape timeline and the S&P 500 and Nasdaq have, in the phrase of the chartist who posted it, a long way to go before they reach the analog's terminal altitude. Both statements come from the same chart. That is not a contradiction; it is the shape of every late-cycle bull market that ever paid its riders and then charged them for the privilege of overstaying. Both hands of the clock.

So that the alarm is actionable rather than rhetorical, we specify the response in advance. If the equal-weight relative line rolls over decisively while the cap-weighted index keeps rising, the sequence is: the fresh small-cap add gets cut first, the cyclical capex barbell gets trimmed second, and the tech core gets hedged rather than sold, because in the 1999 phase of the analog the crowded leaders are the last thing falling. That ordering is written now, calmly, at all-time highs, for the same reason the #61 tripwires were: the worst time to design an exit is while using it.

The secular framing that circulated this weekend rounds out the file: one respected shop counts this as roughly year six of a secular bull that typically runs about ten, and the strongest dissent in the stack argues the opposite corner, that detrended output and employment do not actually support the productivity-boom narrative underwriting everything above. We file both. The falsifiers from #63 stand unchanged: a September hike, a yen break at 160 with a disorderly long end, or a second AI credit event each kills the melt-up reading on contact, and this week's July CPI print is the first scheduled test of the fading-hikes assumption that fuels the whole structure.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The weekend confirmed the resolution and the book acts on it. Cap-weight joined equal-weight at new highs, 40% of world markets made annual highs with none at lows, revenue growth printed its broadest quarter since 2021, and hyperscaler cash flow found what the sell side calls its trough. The small-cap add is executed per the #63 condition; the broadening book is fully deployed for the first time since #59. The semis sleeve stays trimmed into the re-crowding, the software-and-memory core and capex barbell stay, and the rate-sensitive shorts stay on until the hike debate actually dies rather than fades.

The dissent carries forward in its new form: not a top call but a sequence warning. Tech allocations above 50% and climbing are the first frames of the un-broadening film, and our alarm is now the equal-weight relative line rather than any index level. The clock exhibit holds both truths, a long way to go and a scheduled ending, and the discipline is to collect the former without arguing with the latter. Housekeeping for #65: the RSPD/RSPS window from #54 expires around August 20 having never fired through two stalls and a breakout, and we expect to retire it formally; the cyclical-versus-defensive spread is now four issues without a fresh direct reading, which is a process failure this letter needs to fix, and we say so in print to make sure it happens.

We hold both reads on the same page one more time, and for once the page is calm about it: the minute hand pays, the hour hand waits, and the tripwires, as always, get the final word.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

But the window that settled our old argument opened three larger ones, and they are the reason this issue carries the strangest title we have used. In 11 days the market absorbed an AI fund blowup that lands precisely on the Long-Term Capital Management point of the Netscape-to-ChatGPT timeline, the first joint US-Japanese yen intervention since 1998, a 30-year yield at its highest since 2007, and a jobs report in which America lost jobs while the unemployment rate fell to a two-year low because a quarter million people stopped looking. Every analog our dissent file has carried since spring now points at the same year on the clock, and this issue is about what late 1998 does and does not license.

The Fourth Stall Resolved Upward

Grade the #62 report card first, because we promised specificity. We named three outcomes the bull case required: megacap gaps defended where Alphabet's was not, a Fed decision the two-year could live with, and breadth's 52-week-high expansion surviving the week. All three arrived. The earnings gaps held and were bought. The Fed held policy with three dissents, and after a hawkish press conference the two-year yield, our named tell, peaked on July 23 and has since fallen 22 basis points, with this morning's soft payrolls cutting priced hikes for the year from 45 basis points in late July to 27. And breadth did not merely survive: the daily chart letters spent the week announcing that the breadth tailwind is back, more stocks are moving into healthy uptrends, and the Dow sits at fresh all-time highs above a rising 50-day and rising 200-day average. Fresh all-time highs are bullish; it sounds obvious, and after six weeks of arguing about a stalled index, obvious is a luxury. The fourth stall resolved upward.

SOURCE: Daily Chartbook DC Lite #654, The Daily Number via Grant Hawkridge (Fresh All-Time Highs Are Bullish, Dow Above Rising Averages)

Honesty requires the same precision about what did not resolve. Semiconductors spent the window in a genuine washout: at the low, every single component of the SOX traded below its 50-day average, the index closed its worst month since June 2022, and tech companies beating estimates were still being punished on average. The split trade from #61 and #62 did not end; it stopped mattering to the index because everything else got strong enough to carry the weight. That is the handoff thesis of #58 completing in real time, and it is worth saying plainly: the equal-weight complex that peaked in late July did exactly what we said the prior two stalls did. It absorbed the leaders' de-rating without breaking, and then it made new highs.

The Reclaim Held

The earnings verdict deserves its own section because it answered the caboose-and-train question more decisively than we expected. The front of the train re-accelerated: all three reporting hyperscalers showed cloud revenue growth accelerating versus the prior quarter, Amazon joined Alphabet as an upward driver of index-level earnings growth, and analysts raised third-quarter estimates by 0.3% during July, a month in which they typically cut. The capex tolerance question from #62 was not answered with restraint, it was answered with revenue: the market decided it will fund the buildout as long as the cloud lines accelerate, and this quarter they did.

One mechanical note explains much of the week-to-week whiplash and is worth keeping on file: implied correlation almost always rises during earnings season, because single-name bets converge back into index bets as the reports land. July's dispersion extreme was partly a calendar artifact, and its partial reversal in August is not a regime signal either way. The durable signal is reaction quality, and there the split persisted: technology companies beating estimates were still punished on average even as the hyperscalers were rewarded, which says the market is still sorting the AI complex by balance sheet and revenue proof rather than buying it as a block. Our software-and-memory tilt is precisely a bet on that sorting continuing, and the window paid it again.

Margins carried the same message beyond tech: 10 of 11 sectors are expanding margins, which is the broadest margin advance of the cycle and the single best piece of evidence that the record index margin is a cycle phenomenon rather than a concentration artifact. Pair that with the #62 counter-file, record guidance momentum and the cheapest price-to-growth in three decades, and the fundamental case entered August stronger than it entered July, with prices only now catching up to it. The reclaim test that failed on Alphabet passed everywhere it was administered the following week, and the clouds accelerated on cue.

SOURCE: Daily Chartbook DC Lite #651, @fiscal_ai (All Three Hyperscalers Reported Cloud Revenue Accelerations)

The Largest Line Item in America

The macro data of the window described an economy that has quietly reorganized itself around one expenditure. US AI investment reached another record in the second quarter, exceeding a $450bn annual pace for the first time, and it is now the largest single category of American physical investment, larger than single-family housing, larger than factories, larger than power plants. GDP printed 1.5% against 2% expected on trade and inventory noise, but consumption and business investment both accelerated beneath the headline. The ISM manufacturing index reached its highest since May 2022 with production near five-year highs; services activity ran at its second-best since spring 2024. Productivity beat expectations while unit labor costs fell. The high-productivity regime the models flagged at 93% probability last issue keeps printing evidence, and the engine of it all is the largest line item in America.

SOURCE: Daily Chartbook DC Lite #651, @josephpolitano (US AI Investment Tops a $450bn Annual Pace, the Largest Category)

A caveat on the data itself belongs here, because the strangeness of the labor numbers is partly a measurement story. The JOLTS openings series that printed its best openings-to-unemployed ratio since January 2025 rests on a survey only a quarter of firms answer, and the private trackers that capture the rest have deteriorated all year; the official payroll count was just revised down by 103,000 for two months; and the household survey and establishment survey are now telling categorically different stories. In a regime shift, the surveys built for the old regime fail first. We weight the private data, the claims series, and the participation rate most heavily, and all three say the same thing: nobody is being fired, nobody is being hired, and a growing number have stopped asking.

Then there is the labor market, which is telling the opposite story at the same time. July payrolls fell 23,000, the first negative print since February, with 103,000 of downward revisions behind it and a three-month average of 20,000 that was 111,000 in the spring. Yet the unemployment rate fell to 4.1%, a two-year low, because the labor force shrank by 264,000 people on top of June's near-record exodus. Wages grew just 0.1% on the month. Claims sit near their lowest since 2022 and announced job cuts at two-year lows, so nobody is being fired; but hiring has stalled, participation is 61.4%, and the jobless rate keeps improving for the worst possible reason. This is the #59 participation thread at its logical extreme: a labor market that is neither growing nor breaking, attached to an economy investing half a trillion dollars a year in the technology whose stated purpose is to need fewer workers. The market read today's print as dovish relief. The honest reading is stranger, and nobody has priced it yet.

The Long End Has New Suppliers

The rates story of the window was not the Fed; it was the long bond. The 30-year yield touched its highest level since 2007 and the 10-year its highest in 18 months, into a hold decision, which tells you the long end has stopped trading the Fed and started trading supply. The cleanest explanation on offer names three drivers: inflation that will not finish the job, fiscal deficits that will not shrink, and, remarkably, hyperscaler bond issuance, the AI capex boom now competing with the Treasury for the world's duration budget. Last issue the credit market started voting on hyperscaler spreads; this issue it is moving the risk-free curve itself. When the biggest capital program in the country gets funded with corporate bonds, the whole curve feels it, and the long end has new suppliers.

SOURCE: Daily Chartbook DC Lite #652, Apollo via Torsten Slok (30-Year at 20-Year Highs on Inflation, Fiscal, Hyperscaler Issuance)

The policy layer compounded rather than calmed it. The updated Taylor rule, maintained by the chair's own Stanford mentor, prescribes rates well above current levels. Three governors dissented from the hold. September pricing sits at a coin flip even after the payroll miss, with three-quarters odds of a hike by year-end still on the board. And the week's genuine historic event came in the currency market: the United States and Japan executed their first joint yen intervention since 1998, rescuing the yen from a four-decade low near 164, and the Treasury did it by selling euros, blindsiding the ECB in what officials called an unprecedented breach of convention. Half the intervention's gains have already been surrendered; 160 is now the line the whole market watches. A world of record long yields, coin-flip hikes into negative payrolls, and improvised currency defense is not a stable equilibrium; it is a regime searching for its rules, with the this-morning consolation that the two-year, our named tell, has been falling for two weeks.

The Clock Says Late 1998

Now the exhibit that gives this issue its title. Mid-window, an AI-focused fund called Situational Awareness blew up, and the chartists overlaid the event on the analog our letter has tracked all year: the Netscape-moment-to-ChatGPT-moment timeline. The blowup lands almost exactly on the Long-Term Capital Management point of 1998. Add the rest of this window and the rhyme becomes uncomfortable to dismiss: 1998 delivered a leveraged fund collapse, a joint US-Japan yen intervention, and a Federal Reserve that responded by easing, and what followed was not the top. What followed was the melt-up, 15 months of the most violent upside of the cycle, and then the top. This week delivered a leveraged AI fund collapse, a joint US-Japan yen intervention, and a Fed whose hike odds are fading. On the clock, it is late 1998, and the clock says late 1998.

SOURCE: Daily Chartbook DC Lite #651, Bespoke via @bespokeinvest (Situational Awareness and LTCM on the Netscape-ChatGPT Clock)

We are precise about what this analog licenses, because analogs are how desks talk themselves into things. It does not license a price target, a date, or the assumption that this cycle must rhyme at all; the same week produced the sober observation that the most vicious rallies are bear market rallies, and the fundamental differences from 1999, real earnings, record margins, the cheapest price-to-growth in decades, cut in both directions: they argue this is healthier than 1999, and they argue the eventual reckoning is further away than the clock implies. What the analog does license is a scenario our book was not seriously carrying a month ago: that the resolution of the fourth stall is not a grind but a melt-up, retail re-engaging, the crowded fifth re-rating, the equal-weight advance going vertical. A desk positioned only for grind or top misses the middle path that history says is the common one after a blowup clears the overhang. We now carry all three scenarios, weighted, and the tripwires below decide between them.

And because a scenario without falsifiers is astrology, we write down what kills the clock. A September hike would do it: 1998's melt-up ran on easing, and a Fed hiking into negative payrolls is the opposite fuel. A break of the yen defense at 160 followed by disorderly long-end selling would do it, because 1998's interventions worked and this one is already half-unwound. And a second AI credit event that widens hyperscaler spreads rather than clearing positioning would mean the blowup was the start of something rather than the end of something, which is the one reading of the LTCM analog nobody should want to be right about. Absent those, the scenario stands.

The Corporate Bid Returns

The August setup gives the bull scenario its fuel schedule. Buyback eligibility, only about 45% of the index by weight during the earnings blackout, rises to 75% within a week and nearly 85% by mid-August, restoring the market's largest single bid precisely as positioning, in the estimation of the flow desk that tracks it best, has become materially cleaner. Retail is still a net seller of single stocks but no longer absent, dip-buying has reappeared in the AI complex, and our memory-and-software tilt kept its relative bid through the whole washout. Against that: August and September remain the weakest stretch of the calendar, this is a midterm year with the historical pattern of weakness into October before a strong reversal, and the seasonal warning we cited in #61 did, in fact, deliver a 10% peak-to-trough semis event on schedule. The corporate bid returns in force.

SOURCE: Daily Chartbook DC Lite #654, Citadel via Scott Rubner (Buyback Eligibility Rises From 45% to 85% by Mid-August)

The other flows worth naming are the ones outside equities, because they describe the regime rather than the week. Central banks bought a record 289 tonnes of gold for a second quarter, five times the prior quarter's pace, and gold is breaking out as the dollar prints two-month lows. The personal saving rate fell to 2.7%, a four-year low, which is a fully spent consumer. And Bitcoin, the cycle's speculation thermometer in #58, now shows implied volatility near record lows on multi-year-low volumes, the apathy signature of a finished unwind rather than a building mania. Speculative energy left the system through July's washouts. Where it re-enters, and whether the returning corporate bid escorts it back into the equal-weight advance or back into the crowded fifth, is the question that decides which 1999 we get.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The verdict window ruled for the cycle. All three of #62's bull requirements were delivered: the megacap gaps held with cloud growth accelerating through them, the Fed's hold came with language and a two-year path the market could live with, and breadth did not just survive but produced new all-time highs on the Dow, the NYSE Composite, and the index itself, with futures pointing at another record close as we publish. The semis washout ran its full course without infecting the broadening; the caboose de-rated and the train sped up. Our positioning, held through the stress test and trimmed rather than flipped, now re-engages: the small-cap add, paused for four issues, comes off pause on any equal-weight confirmation of these highs, the software-and-memory core stays, the capex barbell stays, and the trimmed semis sleeve stays trimmed until the SOX proves its washout is over rather than resting.

The dissent file did not shrink for being wrong about the week; it re-dated itself. The 30-year at 2007 levels with hyperscaler issuance named among the causes, a Taylor rule prescribing higher rates from the chair's own intellectual lineage, a September coin flip that a single hot inflation print could resolve hawkishly, a labor force shrinking by a quarter million people a month, an improvised currency regime defending 160 on the yen, and Saudi intelligence warning of coordinated attacks on Gulf energy infrastructure: any of these can end the party on its own schedule. And the clock exhibit cuts deepest precisely because it is bullish first: if this is late 1998, the melt-up comes before the reckoning, and the discipline required to ride the former without overstaying into the latter is the hardest trade in markets.

So the triggers, refreshed for the new regime. The #54 vol-regime read survived its maximal test unflipped and stays first. The cyclical-versus-defensive spread still lacks a fresh reading, two issues running, and getting one is now overdue; we say so again rather than pretend. The 30-year yield and the 10y10y forward are now wired directly into the AI capex thread through the issuance channel and carry double weight. The yen at 160 joins the list as the intervention tripwire. And the RSPD/RSPS window from #54 has roughly 13 days to expiry; it either fires or retires by the next issue. We hold both reads on the same page, as always; the difference this issue is that both reads now agree on the direction of the next move and disagree only about how it ends. That is what late 1998 felt like too.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

Because here is the strange part: the same week the tape rejected a beat, the fundamental file got better on almost every line. Profit guidance momentum printed a record. Index margins printed a record. The S&P 500's price against its own earnings growth reached its cheapest in more than 30 years. And a respected quant model now puts 93% odds that the US economy has entered a high-productivity regime. The market's most important argument has never been this sharp on both sides at once, and it resolves this week: 34% of the index's earnings weight reports inside five days, four Mag 7 names among them, wrapped around the most uncertain Fed decision in years. This is the setup issue for that week, with our tripwire report card on the table.

The Ruling on Alphabet

Take the ruling first, because it was clean. The report itself was good: a revenue beat from the world's most-watched advertising and cloud franchise. What the market punished was the spending. A capex guide of up to $205bn for 2026, layered on top of the roughly $725bn the four big hyperscalers had already signaled for this year, tipped the first of them into negative free cash flow, and the reaction told us the tolerance for spending without visible payback has run out. The distribution framework we adopted in #61 said to watch whether earnings gaps got reclaimed or rejected at the reference lines; Alphabet's gap was sold, the bounces since have been sold, and across the complex the pattern repeated. On the first and only megacap test of the window, the reclaim failed at the line.

SOURCE: The Weekly ChartStorm 26 July, MarketCharts.com via Callum Thomas (Lag-7 Lower High, the 493 Peak, MAGS $61 the Trigger)

The flow data behind the price action removes the ambiguity. Goldman's prime book showed hedge funds net selling US tech at a record pace for an eighth consecutive week, with the info tech book at roughly a tenth of sector gross market value, the deepest reading in the dataset. The market is also paying up for crash insurance in a specific way: large VIX call spreads in the 45-65 strikes for August and September, which is not a hedge against a dip, it is a hedge against an event, sized for the thin liquidity of late summer. Whatever the counter-file says, and it says plenty, the biggest and best-informed institutional books have spent two months reducing this theme, and last week they were joined by the reaction function of the market itself.

The Peaks Report

The weekend technical review delivered the sentence we have been bracing for since the handoff thesis began: the Mag 7, the other 493, the cap-weighted index and the equal-weighted index have now all peaked. The Lag-7 complex has put in a lower high, and the MAGS vehicle at $61 is the objective trigger: below it, lower high plus lower low, the textbook definition of a downtrend in the market's most important cohort. Breadth, the metric that carried our bull case through three issues, has peaked at the same level it peaked at the last two times, which gives the tape, in the reviewer's words, the look and feel of what might end up being a larger correction or consolidation phase. We promised we would report the tripwires honestly: the equal-weight uptrend that we called the bull case's spine has stopped rising. It is peaked, but not yet broken.

SOURCE: The Weekly ChartStorm 26 July, MarketCharts.com via Callum Thomas (Equal-Weight and Breadth Peak at Familiar Levels)

What does a triple-tested breadth ceiling actually tell us? Less than either camp claims. Ranges are ranges: breadth stalling at the same level three times defines the top of a corridor the market has respected all summer, and corridors resolve in both directions. The prior two stalls resolved with pullbacks that the equal-weight complex absorbed without breaking its uptrend, which is why the trend, not the ceiling, remains our line. If the coming week's prints push breadth through the ceiling on expanding 52-week highs, the consolidation reads as accumulation and the range breaks up. If breadth rolls from the ceiling while the highs count contracts for the first time since March, the corridor becomes a top. Neither has happened yet, and we decline to pre-trade a range that has not picked a side.

The honest report also includes the counter-reading, and it is not weak. Through the entire consolidation, the percentage of S&P 500 stocks making fresh 52-week highs has expanded, from 35% of the index at the July high to 46% now. A market where the count of individual winners grows while the averages stall is not the classic anatomy of a top; tops narrow, and this tape is still widening underneath. Both breadth claims are true on their own definitions, which is itself the diagnosis: the market of stocks is still healthy, the market of indexes has stalled, and the resolution depends entirely on whether the five reports and one Fed decision of the coming week break the range downward or resolve it up. We are not guessing in front of that; we are positioned for either and flat-footed for neither.

The Caboose and the Train

The best framework of the window came from the internet-era analog. The worrisome part of a boom cycle, the argument runs, is when the companies closest to final demand roll over even as the capital-spending beneficiaries of the cycle keep thriving; in 1999 the trouble started at the front of the train, not in the equipment vendors. The parallel today is uncomfortable and precise: soaring stock prices and free cash flow for semiconductor companies, the caboose, even as the hyperscalers at the front of the train see their stock prices stagnate and their free cash flow plummet. Alphabet's negative print made the metaphor literal. The train's engine is spending itself into cash burn to buy the caboose's product, and the market has begun to ask which end derails first.

This window's genuinely new information is that the question moved into the credit market. Spreads on hyperscaler bonds widened enough to reignite, in the sober language of a bank research desk, investor debate about the viability of their business models in light of their abundant capex. Equity investors argue about narratives; credit investors argue about cash. When both start asking the same question in the same month, the question is real, and the bond market started voting.

SOURCE: Daily Chartbook DC Lite #646, UniCredit via @wallstjesus (Hyperscaler Credit Spreads Widen on Capex Doubts)

For balance: the demand behind the buildout is not imaginary. Data centers are projected to grow from about 6% of US electricity consumption today to roughly 20% by 2035, household AI spending keeps compounding, and the split tape the stock pickers flagged on Friday rewarded the picks-and-shelter end of the theme lavishly: the largest data-center landlord printed its best earnings reaction ever, and the oil-services bellwether rallied 11% on a double beat. What was sold was not AI; it was crowding. Intel delivered its strongest revenue growth in 15 years, its seventh straight quarter above guidance, and fell 7.9%, while an under-owned data-center REIT with the same theme exposure rallied double digits on similar news. That is not a verdict on the technology; it is a verdict on positioning, and it is exactly the sorting mechanism our software-and-memory tilt was built for.

The Counter-File

Now the other side of the page, which grew faster than the bear side this week even as the tape fell. More S&P 500 companies are raising their outlooks than maintaining or cutting them combined, pushing profit guidance momentum to a record in data back to 2011. The index's blended net margin for the quarter stands at 15.7%, a record, and stripping out Alphabet still leaves the second-best margin ever. Seven of 11 sectors are expanding margins year over year. Set those earnings against prices that have gone sideways for two months and the arithmetic produces the single most surprising statistic of the window: the S&P 500's PEG ratio, price against expected earnings growth, is now the lowest in more than 30 years. Growth has outrun price so thoroughly that the most expensive-feeling market of our careers is, per unit of delivered growth, the cheapest since the early 1990s.

The labor inputs feed the same margin story from below. Posted wages are growing 2.4% year over year, below inflation and steady for four months, which is hard on households but is precisely the cost backdrop that lets record margins persist. New business applications with a high propensity to create jobs improved notably, the entrepreneurial churn that typically precedes employment broadening rather than breaking. A cycle where labor costs undershoot prices while formation accelerates is a cycle still funding the earnings line, whatever the tape thinks of it this month.

The macro layer of the file deepened too. A Markov-switching model, the standard tool for judging whether an economic series has genuinely changed state, now assigns 93% probability that the US has moved into a high-productivity period, which is the one macro condition that historically justifies sustained margin expansion and higher-for-longer multiples simultaneously. And the most careful of the cycle historians looking at the 1999 price analog concluded that while the price path tracks closely, fundamentally we are in a much better place than 26 years ago. The counter-file's summary sentence writes itself: the market being distributed at the top of the page is also the market printing record guidance, record margins, and the cheapest growth in three decades.

SOURCE: Daily Chartbook DC Lite #645, Bloomberg Intelligence via Daily Chartbook (Profit Guidance Momentum at a Record)

A Coin-Flip Fed and a Two-Year at the Highs

Wednesday's Fed decision is being described by people who measure such things as the most up-in-the-air meeting in years, and for the first time in this cycle the two live branches are a hike and a hold. The two-year yield made its highest level since February 2025 into the meeting. Real yields are rising, the curve is flattening sharply from the five-year sector out, and the desks make respectable cases on both sides: the doves argue the June CPI and PPI softness plus a cooling labor market argue for patience, the hawks point to supply-driven inflation the chair has said he will not look through. The market's own tell has been consistent for three weeks, and the two-year is the tell.

SOURCE: Daily Chartbook DC Lite #645, Koyfin via Daily Chartbook (Two-Year Yield, Highest Since February 2025)

The inflation inputs that arrived this window mostly armed the hawks. July's flash PMI put the composite at an eight-month high with input prices at a 14-month high. June import prices rose the most year over year since August 2022, with the AI trade itself now pushing up the ex-fuel index. The strategic petroleum reserve sits at its lowest since 1983, which caps the government's ability to blunt an oil shock just as the Iran situation, de-escalated over the weekend but unresolved, keeps a floor under crude; prediction markets still price only a 9% chance the strait normalizes by September. Layer a compressed slate of two-, five-, and seven-year auctions into the two days before the decision, plus GDP and PCE Thursday and the employment cost index Friday, and the rates complex has its heaviest week of the quarter.

The global backdrop reinforces rather than offsets the hawkish tilt. The Bank of England decides Thursday with a hold at 3.75% the consensus, the Bank of Japan on Friday, and Eurozone flash inflation prints are expected to show headline picking back up toward 3% with core drifting higher. The hawkish pivot our charts flagged in mid-July as the global direction of travel has not reversed anywhere: across the central banks the chart watchers monitor, hikes still dominate cuts. A Fed choosing between hold and hike inside a world already re-tightening is a very different animal from the lone-hawk scenario, and it is part of why the two-year refuses to rally on soft CPI prints.

One contrarian file belongs in this section because it is the mirror of everything above: the case that bonds, six years into a bear market with long-term Treasuries 50% off their peak, real returns negative over two decades, sentiment consensus-bearish, valuations cheap, and allocations at 25-year lows, now constitute one of the great contrarian setups of our time. We hold the #55 triggers under the hike-risk convention and have no position in that debate, but a desk that just watched positioning extremes resolve violently in equities should note when the same three conditions, unloved, undervalued, underallocated, appear in the asset everyone has left for dead.

Washout or Distribution

The positioning data of the window supports two incompatible stories, and the difference is the whole game. The long-short ratio on Nasdaq-100 futures has fallen to a 17-year low. Asset manager long positioning in VIX futures sits in the first percentile, meaning nobody owns volatility into the most binary week of the year. Retail's speculative flows have tapered sharply from last year's levels as losses in crypto, precious metals, and now semis sap enthusiasm and buying power. Read one way, this is a washout: the marginal seller has sold, positioning is stretched short in the crowded index, and any positive surprise this week meets a market with nobody left to sell and everybody underexposed. Read the other way, it is mid-distribution: the smart supply, insiders at $77.6bn for the half, hedge funds for eight record weeks, is exiting into a public that has stopped absorbing, and the futures shorts are informed rather than stretched.

Two details tilt our read toward caution without settling it. Margin debt rose last month even as stocks fell, buy-the-dip on steroids in the phrasing of the sentiment historian who flagged it, and that pattern has historically not worked out well; leverage that grows into weakness is conviction converting into fragility. And the dealer-gamma backdrop has flipped: the support that muted every dip this year is draining, with negative-gamma conditions now in place, meaning the structural dampener becomes an amplifier exactly as the calendar loads its heaviest week. A washed-out market with an amplifier attached can rally violently on good news; it can also gap through levels on bad. What it cannot do is stay quiet, and there is nobody left holding the hedge.

SOURCE: Daily Chartbook DC Lite #647, EPFR via Bloomberg (Nasdaq-100 Futures Long-Short Ratio, a 17-Year Low)

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The tripwire report card, as promised. The reclaim test failed: Alphabet's beat was sold, the gap stayed sold, and the Mag 7 surrendered $800bn in a day. The equal-weight spine has peaked at the same level as the last two peaks but has not broken, and MAGS $61 is the objective line between consolidation and downtrend. The cyclical-versus-defensive spread, our load-bearing counter-exhibit, has no fresh reading this window, and we flag that honestly rather than assume it; re-checking it is the first task of the coming week. The #54 vol-regime read remains unflipped, but with dealer gamma negative, volatility ownership at the first percentile, and the calendar maximal, the flip risk into Wednesday and Thursday is the highest it has been in the life of this letter.

One tripwire fired, one bent, one unknown, one loaded. That is not enough to flip a book built on an earnings cycle that is still accelerating, and it is far too much to ignore. So the book moves without flipping: we trim the semis-adjacent edge of the tech long further and raise the cash sleeve modestly, we hold the software-and-memory core and the cyclical capex barbell, the small-cap add stays paused a fourth consecutive issue, and the rate-sensitive shorts stay on under the hike-risk convention. The bull case now rests on three specific outcomes: Microsoft, Meta, Amazon, and Apple defending their gaps where Alphabet could not; the Fed choosing hold over hike with language the two-year can live with; and breadth's 52-week-high expansion surviving the week. Get all three and the fourth stall resolves upward with a record earnings quarter underneath it. Lose the reclaims again and the caboose-and-train analog stops being a metaphor and becomes the map.

We hold both reads on the same page, but for the first time since #58 the page has a deadline printed on it. By Friday we will know whether the market rejected Alphabet, or the cycle. Our money, positioned but trimmed, says the cycle survives the week; our tripwires, written down and numbered, say we will not argue with the tape if it disagrees.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

This issue covers the weekend reads of July 18 and 19 and sets the table for the fortnight that decides the argument, because the calendar has finally put a date on the referee. Amazon, Alphabet, Microsoft, Meta, and Apple all report within the next two weeks, into an index consolidation that is now two months old and unresolved. Every thesis this desk has carried since #58, the broadening, the handoff, the bubble file, the vol-regime watch, converges on those prints. We go into them with the book unchanged and the tripwires defined, and this note is about making those tripwires explicit.

A Tale of Two Indexes

The technical check is the cleanest statement of the divide. The cap-weighted S&P 500 is, in the reviewer's blunt phrasing, a mess: it peaked in early June and has been stuck below overhead resistance ever since, six weeks of failed attempts at the shelf. The equal-weighted index, meanwhile, has been largely onwards and upwards, and the share of constituents above their 200-day moving average keeps trending higher. One market is stalled at its highs; the other is quietly making them. They share the same 500 tickers, and right now they are two indexes, two different markets.

SOURCE: The Weekly ChartStorm 19 July, MarketCharts.com via Callum Thomas (A Tale of Two Indexes, Cap-Weight Stuck, Equal-Weight Rising)

The past week kept the pattern honest even through the Friday break. Ex-tech, it was a quieter week than the headlines suggested: small caps were the best performer of the group, down only about 0.6% in a week when the Nasdaq lost nearly 3%, and the violence stayed concentrated in single names rather than spreading to the index complex. That is the dispersion regime from #60 doing exactly what it has done for a month, repricing the market one stock at a time while the averages hold the range. The broadening thesis is not what broke last week; the leadership is.

It is worth restating why the equal-weight line carries so much analytical weight on this desk. Cap-weighted indexes are a bet on incumbency; equal-weighted indexes are a poll of the median business. When the poll makes new highs while the incumbents stall, history reads it as a leadership change inside a bull market far more often than a top, because tops are built on the opposite pattern: narrow indexes rising on a shrinking cohort while the median stock quietly breaks down. That was 2000's tell and 2007's tell. Today's configuration is its mirror image, which is precisely why the bears have had to argue from factor stress and valuation rather than from breadth, and why we keep granting breadth the deciding vote.

We flagged in the last issue that one more Friday-style day of everything-down would put the broadening on probation. The weekend charts say we are not there: breadth is still rising, the average stock is still trending, and the equal-weight line has not so much as bent. Probation deferred, verdict pending.

The Semis Problem

The other index has a specific problem, and it now has a size attached. Semiconductors rose to become close to a fifth of total US market capitalization, staged a rally of more than four times during the AI boom, and were, in plain terms, the main driver of the market for two years. That weight is what makes the current selling pressure an index-level event rather than a sector story: when a fifth of the market goes into a violent de-rating, the cap-weighted index cannot make a new high no matter what the other four fifths do. This is the arithmetic behind the stall at the shelf, and it is why the headline index will keep looking sick until the chips find a floor. The whole stall reduces to what happens when a fifth of the index wobbles.

SOURCE: The Weekly ChartStorm 19 July, MarketCharts.com via Callum Thomas (Semis Near a Fifth of Market Cap, Now Under Pressure)

For readers joining the story mid-episode, the sequence bears a one-paragraph recap. Ten days ago the sector's bellwether foundry reported a revenue blowout and raised its spending forecast, and the semiconductor index fell 4.3% on the day, the classic sell-the-news signature of a crowded trade. Then the Chinese lab Moonshot released a model strong enough to make open-source parity with the leading commercial systems a live question, which put a question mark over a hyperscaler capex path forecast at $740bn this year and nearly a trillion next. The demand side of AI keeps verifying, household AI spending is up roughly 25% year-to-date and adoption keeps widening, but the capex-to-revenue bridge is the thing being repriced, and semis sit exactly on that bridge.

The proximate catalysts have not improved. The Moonshot release still hangs over the capex story, the foundry bellwether's blowout quarter was sold within hours, and the pattern of the past two weeks, dips bought intraday but rallies sold at the reference line, is precisely the tape the distribution readers have been warning about. The desk view stays where #60 left it: this is a valuation event for the capex complex, not yet an earnings event, and semis remain the stretched end of a tech book we deliberately tilted toward software and memory. The #52 intra-tech framework has never earned its keep more visibly than in the past ten sessions.

Bubble Watch Goes Tactical

The bubble file has been a standing section of this letter since the spring, but it read as magnitude information, useful for the big picture, useless for timing. This weekend it turned tactical. The same reviewer who has patiently catalogued the warning signs now writes that things had become very much bubble-adjacent in semiconductors, that the recent price action makes the risk of deeper downside a real possibility, and, most tellingly, that the Korean equity market, where roughly half of total market cap is a couple of large chip stocks, is already starting to unwind. Korea is the offshore edge of the same trade, and it is moving first. Boom and bust 101, in the reviewer's words.

The falsification conditions matter as much as the flags, so we state them. If Korea stabilizes and the semiconductor index reclaims its own reference lines on the coming earnings gaps, the bubble-tactical read dies quickly and the whole episode files as the positioning reset the constructive strategists described. The BIS reminder from last week, that boom-bust cycles are a regular feature of transformative investment surges, is a statement about eventual outcomes, not about this quarter; even the loudest bubble calls of this cycle have been early by quarters rather than weeks. Our use of the bubble file is unchanged: it sizes the downside if the earnings referee rules against the bulls, and it does not, by itself, tell us the ruling.

The accompanying flags deserve their own line each. Margin debt growth is waving a risk flag. Seasonality, July through October, is waving another. Relative value indicators are, in the reviewer's phrase, presenting an omen. None of these individually is a signal this desk would trade, and we have said so for three issues running. But the density of simultaneous flags is itself information, and the honest reading of the weekend review is that a career-long chart watcher who has stayed constructive through the entire bull run just moved his risk language up a register. His caution cuts both ways, as he notes himself: bears get left behind when bubbles run, and bulls overstay when they turn, because bubbles punish both sides eventually.

SOURCE: The Weekly ChartStorm 19 July, Topdown Charts via Callum Thomas (Semiconductor Bubble Watch, Deeper Downside a Real Possibility)

Who Is Selling Into the Dip

The question nobody asks about a bought dip is who is doing the selling, and the weekend's sharpest read posed it directly. Every dip in semis and AI is getting bought, and a lot of the time the seller into that bid is institutional: distribution into strength, handing shares to traders who think they are getting a discount. The suggested tell is the volume-weighted average price line: a name that gaps down, tries to reclaim VWAP, and keeps rejecting from underneath is not being accumulated, it is being distributed. Lower highs into the line, sellers defending it, price unable to hold above. We would add that this is a testable claim, not a mood: the reclaim-or-reject behavior on the coming earnings gaps will be visible to anyone watching, and we will be watching.

The derivatives market has been telling the same story in its own language. Stock-level risk has spiked relative to index risk, meaning the market is paying up for single-name protection while index protection stays cheap, which is exactly the pricing you would expect if sophisticated players believed individual leaders would be repriced hard while the index was cushioned by rotation. It also rhymes with the generational-speculation thread this letter has carried since #58: record retail participation and record dip-buying were always going to meet institutional supply eventually, and the only question was at what level the handoff of shares would happen. The VWAP rejections suggest it is happening here.

The circumstantial evidence sides with the distribution read. Corporate insiders, the sellers with the best information in the market, sold $77.6bn of stock in the first half, up 20% from a year ago, a pace exceeded only by 2021. At the same time, the marginal retail bid is thinning: net single-stock buying has fallen to a post-Covid low even though retail has not turned seller. Institutions selling size into a bid that is getting thinner at the margin is what distribution looks like from the flow side, and it matches the price behavior at the reference lines. It is not proof, and the dip-buyers have been right all year. But the best-informed sellers keep selling.

SOURCE: Daily Chartbook DC Lite #642, Bloomberg via Michael Msika (Corporate Insiders Sold $77.6bn in H1, Most Since 2021)

Fresh Supply, Full Market

There is a slower-moving version of the same story, and it runs through issuance. The Nasdaq IPO Pulse, a leading indicator of listing activity, remains in an upturn and points to IPO activity staying in an upswing into late 2026. Pair that with the insider selling above and the picture is of equity supply returning to a market that is, by every positioning measure in the #60 file, already fully invested: allocations at record highs, cash and bond weights at the bottom of their ranges, defensives at a record-low share of market cap. Markets absorb new paper easily when there is idle cash waiting to be drawn in; they absorb it badly when everyone is already in, because the marginal buyer has to be created by lower prices. Supply is not a timing signal either, but it tilts the same scale as everything else in this issue, and it is fresh paper meets a full market.

SOURCE: Daily Chartbook DC Lite #642, Nasdaq via Phil Mackintosh (IPO Pulse Points to an Issuance Upswing into Late 2026)

The flow ledger is worth balancing carefully, because it is not one-sided. Long-only funds committed a record $180bn to equities in June, the retail dip bid, while thinning, has not reversed, and systematic strategies remain near full allocation. The bull case for absorbing the new supply is that the earnings season now starting delivers the strongest growth quarter since 2021 and converts nervous holders into confident ones, which is how heavy issuance gets digested in every durable bull market. The bear case is that June's record inflow was the crescendo, not the baseline. The distinction is unresolvable from flow data alone, which is why this issue keeps routing every question to the same referee.

The complacency reading in the currency complex belongs in this section too: the cost of hedging dollar swings has fallen to its lowest of the year, in a window in which the dollar broke a 52-week high, the Fed's futures curve prices hikes, and a shooting war sits on the world's most important oil chokepoint. Cheap insurance in a market this configured is usually mispriced insurance, whichever direction it eventually pays.

The Referee's Calendar

Now the fortnight ahead, which is what this weekend issue is actually for. The macro calendar is nearly empty: services PMI, new home sales, and a Fed that has gone into its blackout period, which removes both the data risk and the rhetoric risk that dominated the last window. Whatever moves this market in the next two weeks will be earnings, and the slate is maximal: all five of the megacaps that anchor the cap-weighted index report into the unresolved consolidation. The reports themselves matter less than the reaction function, and we have a precise framework for reading it, from the distribution logic above. Every earnings gap, up or down, is a fresh test of the reference line. Reclaims that hold mean the dip-buyers still control the tape and the two-month consolidation resolves upward, in which case the equal-weight index will have spent the whole stall quietly right. Rejections from underneath, repeated across several of the five names, mean the distribution read was correct, the fourth stall resolves like the first three, and the bubble file graduates from magnitude to timing.

It is equally worth writing down what a bullish resolution looks like, so we recognize it without rationalizing. Good, for this book, is not semis ripping back to highs; good is the megacap gaps holding their reclaims while the equal-weight index keeps grinding, software and memory holding their relative strength against stretched semis, and the cyclical-defensive spread widening back out. That outcome would complete the handoff this letter has tracked since #58: the index passes leadership to its median stock, the crowded fifth de-rates without infecting the rest, and the fourth stall resolves upward on broader shoulders. If instead the reclaims fail but everything else holds, we get more consolidation, which suits the broadening fine; only failed reclaims plus breaking breadth flips the book.

Our own tripwires into the fortnight, stated for the record. First, the cyclical-versus-defensive spread, still positive through the entire unwind: if it goes negative, the broadening thesis is wrong, not early, and the book de-risks. Second, the #54 vol-regime read, down-day variance against up-day variance: still unflipped, still the single most important line on our screens. Third, the equal-weight index itself: it has carried the bull case for six weeks, and a decisive break of its uptrend would mean the loud tape finally spoke. One of the strategists we track has already changed positions during the unwind and republished carry-trade risk models; we are not there, but we wrote the conditions down before the prints so that we cannot pretend later we never had them.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The weekend compressed the market into one sentence and a calendar. The equal-weighted index and rising breadth say the bull market is broadening exactly as the #58 positioning read predicted; the semiconductor complex, a fifth of total market cap after a fourfold run, is being distributed with Korea already unwinding at the fringe; insiders have sold more stock than in any first half since 2021 while the IPO window reopens into a fully invested market; and all five megacap anchors report within two weeks into a consolidation that has refused to resolve for two months. Both halves of the picture are true today. One will end up speaking loudest, and the earnings reaction function, reclaim or reject at the reference lines, is how we will know which.

The dissenting read keeps the first word it earned in #60: bubble language from patient observers, tactical distribution behavior, record insider supply, fresh issuance, complacent hedging costs, and hostile seasonality make a coherent top-building case, and if the megacap gaps get sold, every element of it confirms at once. The bull read still owns the book: breadth is rising, small caps outperformed through the worst tech week since spring, non-AI earnings are accelerating into the strongest reporting quarter since 2021, and a de-rating confined to the most crowded fifth of the market while the other four fifths trend higher is rotation completing, not distribution beginning.

Positioning is

unchanged and deliberately so: we hold the EPS-driven tech long with its software-and-memory tilt, hold the cyclical capex

barbell, keep the small-cap add on pause until the megacap prints clear, and stay short-to-flat the rate-sensitive long-duration cyclicals under the hike-risk convention. The five triggers stand as written in #60, with the #54 vol-regime read at the top of the list and the RSPD/RSPS window now roughly 32 days from expiry. No changes into the prints; the tripwires are on the page and the referee takes the field this week.

We hold both reads on the same page one more time, knowing the page is about to be graded. The market of stocks says stay; the market of chips says go; and by the next issue, one of them will have spoken loudest.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

This issue covers July 5 through July 17, and it is a stress test in the precise sense: every position and every trigger we carry got pressure-tested by new data. The earnings spine got stronger. The macro got more hawkish. The positioning got more crowded. The dissent bank got two of the loudest exhibits it has ever held. We take them in order, and we grade our own book honestly along the way, because one of our calls from #59 already needs a correction.

The Stall at the High Shelf

Start with the tape's most basic fact: the S&P 500 has now spent weeks sitting within 5% of its all-time high without printing a new one, and this is only the fourth time in the entire bull market that has happened. The prior episodes are not comforting company. Excluding 2024, every one of them resolved in a real pullback, and the bears have been circulating that table all week. The market has, quite literally, gone nowhere; the question the whole issue turns on is whether nowhere is a base or a top.

The base-rate studies that accompanied the stall lean cautious in a way they did not a month ago. The 200-day moving average has now risen for 300 consecutive days, a streak seen only 14 other times since 1975, and the forward 12-month return profile after such streaks is below average. The VIX printed a fresh 120-session low under 16 in mid-July, and the historical follow-through after those prints pairs an above-average hit rate with a below-average median return: markets tend to keep grinding, but the easy part is behind. None of these are sell signals. All of them say the distribution of outcomes has narrowed, and four stalls, three real pullbacks.

SOURCE: Daily Chartbook DC Lite #640, @bluekurtic (Fourth Stall Within 5% of the High, Prior Cases Saw Real Pullbacks)

That reading, stated plainly, is why we did not add to the broadening trade into this strength. The #59 plan was to add the small-cap leg on weakness, and weakness has now arrived; whether it is the buyable kind depends entirely on the sections that follow.

Quiet Index, Loud Tape

Under the flat index, the tape spent the window as loud as it has been in four decades. Stock-to-stock correlation inside the Russell 1000 touched a level seen only a handful of times in roughly 40 years. A quiet index with winners and losers this violent underneath is the signature of a market being repriced stock by stock rather than sold wholesale, and it is the direct continuation of the dispersion regime flagged in #59, taken to its statistical extreme. In a tape like this the index is the least informative price on the screen: alpha is loud, beta is quiet, and the winners and losers cancel out.

SOURCE: Daily Chartbook DC Lite #637, @renmacllc (Russell 1000 Stock-to-Stock Correlation at a 40-Year Low)

The rotation stayed violent all fortnight, and mostly in the direction our book wants. Last Thursday the S&P 500 fell half a percent while eight of 11 sectors closed higher and the equal-weighted index gained a full percent: a down day for the index that was an up day for the average stock. The cyclical-versus-defensive industry spread, the engine of the whole #58 to #59 thesis, remained positive through the entire window, a condition that has historically coincided with above-average forward returns. Financials kept working, the Dow outperformed all week, and a 114-year-old mega cap printed a record weekly close while the glamour names bled. As one strategist put it on Friday, the market has started to hope for some long-awaited broadening, and rotations tend to happen in quite a violent way; this is what violent looks like.

Friday is the honest caveat. On Friday the selling finally broadened, energy was the only sector that closed green, and the equal-weighted index fell 0.8%, barely better than cap-weighted. One day of everything-down does not overturn two weeks of rotation, but it is precisely the pattern we said we would respect: distribution is when the average stock stops absorbing what the leaders lose. One more week like Friday and the broadening thesis goes on probation.

The Hawk Meets a Cool CPI

The macro half of the window produced the strangest pairing of the year. On the data side, inflation broke lower: June CPI printed its biggest headline and core declines since April 2020 with broad-based softening, June PPI fell month-over-month by the joint-most since 2020, and the July Beige Book read materially friendlier than June, with growth and employment broadening and prices cooling. On the policy side, none of it mattered. The minutes from the new Fed chair's first meeting scored as the most hawkish since April 2022, the chair has said explicitly that he will lean harder on the data, and fed funds futures spent the window pricing one to two hikes over the next 12 months. The desk read on core PCE, tracking roughly 1.9% annualized for June, came with the sharpest one-line summary of the whole regime: welcome news, but hardly enough to derail the conversation around rate hikes. Cuts are not the debate anymore; the futures curve now says hikes.

SOURCE: Daily Chartbook DC Lite #637, Yardeni Research (Fed Funds Futures Now Price One to Two Hikes)

Why the disconnect between cool prints and hawkish pricing? Two words: Iran and oil. Since the strikes began, economists have been cutting growth and raising inflation in the same breath; the Bloomberg consensus for 2026 now sits at 2.1% GDP against 3.4% CPI, a mild but unmistakable stagflation drift. The blockade is biting: Iranian oil infrastructure is being hit, empty tankers are being disabled to deny floating storage, the rial is at record lows, and Tehran is now demanding ships seek permission to transit the strait. Oil firmed into the weekend on reports that escalation options are on the table within days. Import prices are already picking up the petroleum impulse. A Fed that just watched its best inflation prints in years arrive alongside a live energy shock is a Fed that stays hawkish, and the market has priced it.

The rest of the domestic data kept to the #59 script of a slow-motion economy. Retail sales for June came in as expected with the strength beneath the surface in autos and online, and the real-adjusted trend shows a resilient but price-sensitive consumer. Jobless claims stayed subdued in the middle of their multi-year range. Housing remained the weak spot: builder sentiment has now spent 27 straight months below 50 and sits at its joint-lowest since September, pending home sales snapped four months of gains with their largest drop since December, and June's headline beat in starts was entirely multifamily while single-family starts and permits went nowhere. Consumer sentiment improved to its best since February on softer gas prices, an improvement the new oil bid may take back. Nothing here breaks; nothing here accelerates; the earnings cycle keeps carrying the load.

For the rates complex this forces a correction to our own book. In #59 we argued a labor market cooling through participation would let yields drift lower without recession pricing. That was the wrong read on the wrong risk: the pressure on yields is now from the hawkish side, with the dollar coming off a 52-week high and hedging costs against dollar swings at their lowest of the year, which is its own complacency reading. The 30-year and the 10y10y forward stay on the trigger list, but the sign convention has flipped: we are watching for hike-risk steepening, not cut-drift rallying. When the data and our framing disagree, the framing loses.

The Earnings Spine and the Moonshot Shock

Now the bull half, which got stronger even as the tape got worse. Second-quarter earnings are tracking above 29% growth based on the average in-season improvement, which would be the best quarter since Q4 2021. Non-AI earnings growth is accelerating, which is the single most important sentence in the entire bull case because it funds the broadening with profits rather than multiple expansion. And the concentration math is turning over: the Magnificent 7 are expected to out-earn the other 493 again in Q2, but the growth gap narrows in Q3 and flips outright by Q4. The index's earnings engine is being handed from the few to the many on the sell-side's own numbers, and the other 493 take the baton.

SOURCE: Daily Chartbook DC Lite #642, FactSet via John Butters (Mag 7 vs the Other 493, the Earnings Gap Flips by Q4)

Against that spine came the window's genuine shock. A powerful new model from the Chinese AI lab Moonshot landed Friday and reframed the AI capex question in a way no US earnings miss has managed: if open-source Chinese models can rival the leading commercial offerings, the sustainability of a hyperscaler capex path forecast at $740bn this year and $996bn next becomes a fair question rather than a bear fantasy. The tape had already hinted at fatigue: the semiconductor index fell 4.3% the day a foundry bellwether reported a revenue blowout and raised its spending forecast, which is what sell-the-news looks like at a cycle's later innings, and end-user AI demand is visibly getting more price sensitive even as household AI spending runs up 25% year-to-date. The strategist consensus into the weekend framed it as the theme maturing rather than breaking, with the BIS reminding everyone that boom-bust cycles are a regular feature of transformative investment surges. Our synthesis: the Moonshot shock is a valuation event for the capex complex, not yet an earnings event, and the earnings season now doubles as the referee. As one European strategist put it, the key reassurance would probably come from the earnings season; we agree, and it starts now.

All-In and Uneasy

The positioning picture hardened into the shape late-cycle markets know well: investors saying cautious things while doing committed things. Portfolio allocations to equities are hovering around record highs while surveyed sentiment reads only mildly bullish, and we side with the actions over the words: investors are basically all-in. The supporting file is thick. The leveraged ETF long-versus-short ratio has surged into warning territory. The Investors Intelligence bull-bear ratio reached 3.12, approaching excessively bullish. The Nasdaq IPO Pulse points to an issuance upswing into late 2026. The average household now holds more of its wealth in equities than in real estate, which converts the next major drawdown into a real-economy event through wealth effects. And allocations to cash and bonds sit at the bottom of their historical range with defensive stocks at a record-low share of market cap: the typical portfolio has never been positioned with less ballast, which tells you nothing about timing but everything about the consequences of what investors do, not say.

SOURCE: The Weekly ChartStorm 12 July, Topdown Charts via Callum Thomas (Record Equity Allocations, Only Mildly Bullish Sentiment)

Two nuances complicate the froth reading, one on each side. On the cooler side, retail's net single-stock buying has fallen to a post-Covid low; the answer to whether retail has started selling is still no, but the marginal retail bid that defined the first half is fading, which removes a cushion the dip has enjoyed all year. On the hotter side, the sellers with the best information are active: corporate insiders sold $77.6bn of stock in the first half, up 20% year-over-year, with only 2021's spree larger. Record allocations, fading retail flow, and heavy insider supply is not a crash recipe, but it is exactly the configuration in which an external shock finds the market with no spare buyers at the margin.

The Unwind Gets a Number

The dissent section this issue is not about valuation tables; it is about the live factor event. Goldman's high-beta momentum basket is on track for its worst monthly performance in roughly 17 years. Read that sentence twice: worse than any month of 2022's rate shock, worse than the Covid crash months, a two-standard-deviation factor unwind happening inside a flat index. Stock-level risk has spiked relative to index risk across the derivatives complex, which is the market paying up for single-name insurance while index insurance stays cheap, and it is the structural echo of the correlation extreme in the second section. The momentum unwind that #59 called the first real test of the #54 vol-regime trigger did not pass and fade; it deepened for two straight weeks, and the unwind now has a number.

SOURCE: Daily Chartbook DC Lite #641, Bloomberg via Daily Chartbook (High Beta Momentum, Worst Month in 17 Years)

The bigger-picture exhibits stack alongside it. Global semiconductor stocks have risen 10 times in real terms, which history files under major-bubble behavior regardless of how good the underlying cash flows are. Seasonality now works against the tape, with July through October the most volatility-prone stretch of the year. And the longest-horizon chart of the window may matter most: global equities have now persistently outperformed US equities since the 2025 turning point, and the secular trend in favor of the US that emerged after the financial crisis looks broken. A US market priced for exceptionalism, positioned all-in, entering vol season with its momentum complex unwinding and a bubble-grade AI cohort being repriced by foreign competition: that is the bear case, and this window strengthened it more than any two-week stretch since we started numbering these issues.

For balance, the counter-exhibit is the one already noted: the cyclical-versus-defensive spread stayed positive through the whole episode. Factor unwinds inside positive cyclical leadership have historically resolved as rotations, not tops. That single spread is carrying a lot of weight right now, and we are watching it daily.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The two weeks delivered a stress test and the results are split. The index stalled within 5% of its high for only the fourth time this bull market while the historical base rates cooled. Dispersion hit a 40-year extreme as the rotation stayed violent but constructive until Friday's broad down day. The Fed's first Warsh minutes scored the most hawkish since April 2022 and futures moved to pricing one to two hikes even as CPI and PPI printed their softest readings since 2020, with an Iran-driven stagflation drift in the consensus. Q2 earnings are tracking their best quarter since 2021 with non-AI growth accelerating and the Mag 7 earnings gap set to flip by Q4, while a Chinese open-source model put the first credible question mark over the trillion-dollar capex path. Allocations sit at record highs against fading retail flow and the heaviest insider selling since 2021. And the momentum complex is finishing its worst month in 17 years inside a flat index.

The dissenting read, which after this window deserves the first word rather than the last: this is what a top being built looks like. The stall, the factor unwind, the insider supply, the all-in positioning with no defensive ballast, the external technology shock, and the seasonal turn all arriving together is how bull markets historically end, and if next week's earnings reassurance fails to arrive, the fourth stall resolves like the first three.

The bull read, which still owns our book: the correction is happening inside the market rather than to it. Earnings breadth is accelerating exactly as the broadening thesis requires, the cyclical-defensive spread never went negative, cool inflation gives the Fed room to disappoint hawkishly-priced futures rather than the reverse, and a positioning reset in the momentum complex creates the entry points we said we wanted. Positioning: we hold the strategic long in EPS-driven tech with the software-and-memory tilt, we hold the

cyclical capex barbell, we keep the small-cap add on pause until the equal-weight index proves Friday was the exception rather

than the new rule, and we stay short-to-flat the rate-sensitive long-duration cyclicals with the trigger convention flipped to hike-risk. The five triggers stand, with the #54 vol-regime read now the whole ballgame: down-day variance is the single line we watch next week. The #54 RSPD/RSPS window has roughly 33 days left to run.

We hold both reads on the same page: the earnings tape says the handoff is completing, and the factor tape says the exit is crowding. Until the cyclical-defensive spread breaks or the vol regime flips, we treat this stress test as a repricing inside a bull market, and we let earnings season grade the paper.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

This issue covers June 5 through July 4 and picks up the thread BSF #58 left on the table. We closed that issue with cyclical-minus-defensive positioning at the 17th percentile and argued the rotation had fuel. Four weeks later the fuel is visibly burning: the average stock is at new highs, value has printed an all-time high, small caps are stirring, and the old leadership is being sold at a pace with only one precedent in the modern dataset. The structural read, the tactical read, and the dissent all follow from that single handoff, so that is where we start.

The Broadening Is the Headline

Strip out the market-cap weighting and there is no stall to explain. The equal-weighted S&P 500 pushed to fresh all-time highs into early July while the cap-weighted index sat in consolidation near 7,483, and the share of S&P 500 constituents above their 200-day moving average climbed to roughly 67% and kept trending higher. This is the cleanest single signal of the month: the median stock is in an uptrend that the headline index is currently underselling.

The advance-decline data says the same thing more loudly. In the last week of June, with the index itself still sitting about 2% below its high, both the S&P 500 and NYSE cumulative advance-decline lines made new all-time highs. Compare that with the first quarter of last year, when breadth diverged negatively for months before the spring rollover: this time the internals are leading the index, not lagging it. Participation is the story, and the average stock has already voted.

SOURCE: The Weekly ChartStorm 5 July, MarketCharts.com via Callum Thomas (Equal-Weight S&P 500 at New Highs, Cap-Weight Stuck)

The style detail confirms it. The S&P 500 Value index quietly made a new all-time high in the final week of June while the Growth index peaked a month earlier, and the value-versus-growth relative performance line is attempting to turn from the bottom of a multi-decade channel. Financials are pressing higher on both sides of the Atlantic (the European financials ETF closed the window at record highs), aerospace and defense reclaimed its March highs, and micro caps have started to put in promising price action.

It also reframes the volatility of the past few weeks. The chop that unsettled the headlines was driven by sector rotation rather than broad-based selling: on the day the index printed its record participation readings it was still digesting a 2% pullback, and the dispersion between a rising Dow and a falling Nasdaq on the same session is what a leadership change looks like in real time, not what distribution looks like. Distribution empties the advance-decline line first; this tape keeps filling it.

Down the cap spectrum, the case is contrarian rather than confirmed. Small caps remain a neglected corner of the market: ETF market share and rolling 90-day fund flows are only now ticking up off record lows even as the Russell 2000 presses against its prior high shelf. Neglect plus cheap valuations plus improving technicals is the classic setup for the next leg of a broadening, and it is the piece of this theme we would add to on weakness rather than chase on strength.

The Handoff Underneath

The other side of the broadening is the exit. Last week's hedge fund net selling in US information technology was the largest in more than 10 years in both dollar and percentage terms, a four-standard-deviation event in the Goldman Sachs prime book. That is not trimming; that is a deliberate rotation out of the crowded leadership, and it landed in the same window in which the Magnificent 7 complex kept lagging the other 493 stocks. The prime brokerage data has given us the fastest tech exit in a decade.

SOURCE: Daily Chartbook DC Lite #629, Goldman Sachs via Robbie Stankard (Hedge Fund Tech Selling, Largest in Over 10 Years)

Price action followed the flows. The Philadelphia Semiconductor Index dropped more than 5% in a single session in early July after a multi-month bearish momentum divergence, the Nasdaq Composite closed below its 50-day moving average, and Goldman's high-beta momentum basket posted its worst two-day move since Covid. On July 2 the Dow rose about 1% while the Nasdaq 100 fell 1.7%: that is not a risk-off day, it is a hand-off day, and it is precisely the intra-market dispersion the BSF #52 rotation framework was built to catch.

Two details keep this from reading as a simple momentum crash. First, the seasonality work circulating this week shows July has been the weakest month for the momentum factor in each of the past five years even as it remains one of the most bullish months for the index, which historically resolves in favor of the broadening trade rather than against the market. Second, US equities have effectively stopped trading as one index; realized correlations have broken down and the dispersion backdrop into earnings is the richest in years. A market of individual stories is a stock picker's tape, not a topping tape, so long as the money leaving the leaders keeps landing somewhere else.

Our strategic long in EPS-driven tech is unchanged, but the internal mix matters more than ever: the #57 software surge and the memory-over-stretched-semis tilt did their job this month, and the semis air pocket is the strongest argument yet for keeping that tilt. The #55 software-energy rotation map also earned another confirmation, with energy leading the sector-level estimate revisions into the quarter end while stretched semis absorbed the factor unwind. One of the sharper summary lines of the month put it plainly: what is changing is not the fundamental story, it is the market structure, and a structure that rewards selection over concentration is the structure this desk has been positioned for since the #52 framework went on the page.

The Positioning Split

The flow data and the survey data spent the month walking in opposite directions, and the gap between them is the most interesting sentiment picture of the year. On the flow side, equity inflows from long-only funds hit an all-time high of roughly $180bn in June, and the institutional bid was matched by a retail bid with no modern comparison: retail investors bought nearly 3.5 times their average daily amount on S&P down days in the first half of 2026, the strongest buy-the-dip behavior in the Citadel dataset, and nine of the 10 largest retail trading days ever recorded on a major platform occurred within a single month of the window. Whatever this market lacks, it is not a record wall of committed money.

SOURCE: Daily Chartbook DC Lite #631, Barclays via Emmanuel Cau and Jamie Chisholm (Long-Only Equity Inflows, All-Time High Near $180bn in June)

On the survey side, the AAII bullish share just dropped by the most in nearly five years, pushing the bull-bear spread back into negative territory, and global gold ETFs gave up $4.7bn in a single week as the fear bid unwound. Positioning is committed while stated sentiment has washed out, which historically is fuel rather than froth: corrections start when everyone is both invested and euphoric, not when everyone is invested and nervous.

This is also the direct sequel to the #58 positioning read. Cyclical-minus-defensive positioning sat at the 17th percentile a month ago, the least crowded since 2023, and we argued the rotation had fuel precisely because nobody owned it. A record long-only inflow arriving in the same month the average stock breaks out is what that fuel looks like when it ignites: the crowding is building in the direction of the broadening, not against it, and by the positioning clock this rotation is early rather than late.

We hold the #58 caveat firmly in place, though. Record dip-buying and record retail volume are exactly the speculation flags the generational-speculation framing has been waving since the spring, and a flow regime this committed cuts both ways: it cushions every dip until the day the dip-buyers themselves need liquidity. The tactical read is that the split favors the bulls; the structural read is that it deepens the fragility file rather than emptying it.

Labor Cools at the Edges

June payrolls were the month's most important macro print and the easiest to misread in both directions. The economy added 57K jobs against a three-month average of 111K, less than expected but still probably above breakeven. The unemployment rate ticked down to 4.2%, which sounds reassuring until the composition is examined: the decline came from the labor force participation rate dropping to 61.5%, its lowest since March 2021. The jobless rate is falling for the wrong reason, an exit from the workforce itself.

SOURCE: Daily Chartbook DC Lite #632, Koyfin via Daily Chartbook (Unemployment 4.2%, Participation 61.5%, Lowest Since March 2021)

The softer undertone shows up in the survey data too. The Conference Board's labor differential (jobs plentiful minus jobs hard to get) made a new cycle low in June, which is the single best leading read on how workers themselves experience the market, and it has been deteriorating steadily while the headline prints stayed polite.

Set against that, the rest of the month's labor file refuses to crack. Challenger announced job cuts fell to 45.8K in June, down 53% on the month and the lowest total since December 2025. JOLTS openings surprised to the upside at 7.59 million. The Revelio alternative payroll series showed 258.8K jobs added in June. And the AI displacement story keeps failing to appear in the aggregate data: firms that adopt AI heavily grow headcount about 10% over the two years following adoption, while AI is simultaneously the most-cited reason for the layoffs that do occur (31% of June's announcements). Low churn, low hiring, low firing: a frozen labor market that is thawing at the edges rather than breaking.

The household backdrop rhymes with the frozen-but-fine read. The median US home-sale price hit a record $400,894 in the middle of the window, which is the housing market's version of the same story: constrained supply and locked-in owners keeping prices at highs while activity stays muted. A consumer who owns an appreciating house and holds a job he is afraid to leave is not a consumer who breaks the economy, but he is also not one who accelerates it, and that is broadly what the spending and confidence data delivered this month. Slow-motion macro is an underrated regime for equities: it keeps the earnings cycle intact while pinning the volatility that usually ends bull markets.

For the rates complex this lands dovish at the margin, and it arrives with the dollar at a fresh 52-week high, which tightens conditions on its own. The 30-year yield and the 10y10y forward stay on the trigger list from #55 exactly as written; a labor market cooling through participation rather than layoffs is the configuration that lets yields drift lower without recession pricing, and that would be jet fuel for the rate-sensitive half of the broadening.

The Earnings Spine and the Base Rates

The bull case in this tape still runs through earnings, and the earnings news this month was quietly excellent. During the second quarter, analysts raised S&P 500 EPS estimates by 3.4% over the course of the quarter, the most positive intra-quarter revision since Q2 2021, against a 20-year norm of cutting estimates by roughly 4% as the quarter progresses. Six sectors saw estimates rise, led by energy and tech. Forward margins were already at record highs in the #58 read; the revision behavior says the estimates under them are still too low.

This continues the earnings thread that has anchored the strategic long since the spring. The #58 read had forward 12-month profit margins at record highs and ADP on an 11-month winning streak; a quarter that opened with those margins and closed with analysts raising numbers into the print is a quarter in which the earnings cycle is broadening along with the tape. Revisions led by energy and tech are also exactly the sector pair the barbell wants to see.

Valuation has even rejoined the argument. For the first time this year, a bullish S&P 500 case can be built using nothing more heroic than consensus earnings and a historically average multiple, which pencils to roughly 8% upside. And within the most-feared corner of the market, the Mag 7 hyperscalers now trade at their lowest forward P/E since the launch of ChatGPT: the profit engines of the AI theme have de-rated even as the sell-side keeps finding consensus 2027 capex estimates too conservative.

The base rates lean the same way. The S&P 500 has never peaked for the year in June. More pointedly, the index made more than 20 new all-time highs in the first six months of the year, and in every prior such year it has never peaked before August, with the full year finishing positive 100% of the time. History is not a guarantee, but in a month when the bear case got louder, the historical analog stayed stubbornly constructive, and the calendar keeps siding with patience.

SOURCE: Daily Chartbook DC Lite #630, @bluekurtic (20-Plus First-Half Highs, the S&P Has Never Peaked Before August)

Bubble Watch

The dissent bank grew this month, and it deserves its section. Relative valuations for the TMT complex are approaching their peak dotcom levels, and defensives are already at theirs, which together say the market's pricing of the AI theme has reached the neighborhood of the only real precedent. Sentiment adds color here: the same week's charts flagged that the US, Korea, and China all share one bubbly thing in common, and it is their AI stocks. None of this is timing information. All of it is magnitude information, and the dotcom yardstick back in play.

SOURCE: Daily Chartbook DC Lite #631, Topdown Charts via Callum Thomas (TMT Relative Valuations Approaching Peak Dotcom Levels)

The sharpest analog of the month came from the 2000 semis playbook: in that cycle, price peaked in March 2000 while EPS kept rising into 2001, so anyone who waited for earnings to miss before selling flew into the side of the mountain. That is a direct challenge to the earnings spine in the prior section, and we take it seriously rather than rhetorically. The honest synthesis is that rising estimates tell you where the cycle is, not where the top is; they were rising at the 2000 peak too.

The technical warning is more concrete. The S&P 500 spent late June flirting with a rare diamond-top pattern whose measured move, if it resolves lower, projects toward the 7,000 area, conveniently the prior high shelf from the first quarter. And the structural fragility file gained a hard number: hedge funds' share of outstanding Treasuries has doubled since the beginning of 2022, financed through repo in basis trades and swap spreads. Pair that with the #58 finding that a roughly 20% equity drawdown now buys only a 3% bond rally and the extreme yields-stocks linkage flagged in mid-June, and the picture is unchanged but heavier: the hedges are thinner than the positioning assumes, and the unwind channel runs through the Treasury market.

None of the dissent book is firing as a timing signal today. The #53 composition signal, the #55 margin-debt read, and the #56 hawkish-minutes read all remain filed, and this month adds two exhibits rather than a trigger.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict. All macro.

SOURCE: Renaissance Macro Research

SOURCE: DataTrek

SOURCE: Kevin Gordon

SOURCE: Krishan Gopaul

The Verdict

The month handed the tape from the few to the many. The equal-weighted S&P 500 and both major advance-decline lines sit at all-time highs while the headline index rests, value has printed a record alongside a growth index that peaked a month ago, hedge funds just sold technology at the fastest pace in over a decade while long-only funds committed a record $180bn, June payrolls cooled to 57K with participation at a post-2021 low even as Challenger cuts hit a six-month low, second-quarter EPS estimates rose the most intra-quarter since 2021, and TMT relative valuations moved within sight of their dotcom peak. That is one coherent story told seven ways: the handoff is real, funded, and so far orderly.

The dissenting read is that every great distribution in history looked like a broadening first. Money leaving the leaders landed elsewhere in 2000 as well, right up until the equal-weight outperformance turned out to be everything falling less. If the momentum unwind that produced the worst high-beta two-day move since Covid is the start of the #54 vol-regime flip rather than a contained factor event, then the record dip-buying and the doubled hedge-fund Treasury book stop being cushions and start being accelerants, and the diamond-top target near 7,000 becomes the working map.

Positioning is unchanged in direction and adjusted in mix. We stay strategically long EPS-driven tech with the intra-tech tilt toward software and memory over stretched semis, we add to the cyclical capex barbell and the small-cap leg of the broadening on weakness rather than strength, and we stay short-to-flat the rate-sensitive long-duration cyclicals until the yield complex actually confirms. The five triggers are unchanged: the #54 vol-regime read just faced its first real test and remains the top of the list, the #55 software-energy rotation map, the 30-year yield, the 10y10y forward with the dollar now at a 52-week high beside it, and the #49 private-credit stress indicator. The #54 RSPD/RSPS window has roughly 46 days left to run.

We hold both reads on the same page: the average stock says the bull market just got healthier, and the valuation tape says it got closer to the only precedent that matters. Until the vol regime flips, the burden of proof stays with the bears, and the broadening stays the trade.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

What makes the current setup unusual is that two things are true at the same time. The fundamental case for owning equities got stronger over the last several sessions, with labor holding, margins pressing to new highs, and the rally itself flashing a constructive historical base rate. And yet the positioning that sits underneath that case is nowhere near stretched. We hold both reads on the same page, because the second one is what tells us the first one still has room. The risk, as always, lives somewhere else, and this week it moved from valuation to plumbing.

Positioning Has Not Caught Up to Price

Start with the single cleanest signal of the week. A measure of cyclical sector positioning minus defensive sector positioning now sits at the 17th percentile of its history, the least cyclically tilted the market has been since 2023. Read that against a tape making new highs and it is close to a contradiction. Price says the cyclical rotation is mature. Positioning says investors are still huddled in the defensive corners of the market, underweight exactly the sectors that have been leading. The least cyclically tilted positioning since 2023.

SOURCE: Daily Chartbook DC Lite #612, Deutsche Bank via @neilksethi (Cyclical Minus Defensive Positioning, 17th Percentile)

This is the missing piece from the BSF #50 to BSF #56 breadth story. We spent six issues documenting a divergence between a narrow cap-weighted advance and a broad set of laggards that refused to participate. When that divergence resolved in BSF #57, the bear interpretation was that the catch-up was the last gasp, the final group of buyers being pulled in at the top. The positioning data argues the opposite. If the average book were genuinely crowded into cyclicals, you would expect this measure to be sitting in the 80th or 90th percentile, not the 17th. The catch-up in breadth was real, but it was a catch-up in price, not in exposure. The money has not yet followed the tape.

The tactical implication threads directly into the strategic position we have carried since the spring. Our cyclical capex barbell, the industrials, copper miners, energy and clean energy basket, is not a crowded trade by this measure. Neither is the intra-tech rotation we mapped in BSF #52 and confirmed in BSF #55. A rotation that the rest of the market has not bought is a rotation with fuel left in the tank. We add nothing to the thesis here, we simply note that the contrarian condition we needed to keep holding it is intact.

Margins, Not Multiples

The second pillar of the week sits underneath the first. The forward twelve-month profit margin for the S&P 500 keeps spiking to new highs. This matters more than any single quarter of earnings beats, because it speaks to the durability of the structural-long thesis we have run since BSF #43. The bull case for US equities has never rested mainly on multiple expansion. It rests on the claim that this index earns more, per dollar of revenue, than any prior version of itself, and that the AI capex cycle is bending the margin structure higher rather than simply pulling demand forward. A forward margin at new highs is the data point that claim needs.

SOURCE: Daily Chartbook DC Lite #612, @kevrgordon (SPX Forward 12-Month Profit Margin at New Highs)

It is worth being precise about why margin matters more than the multiple here. A market that rallies on a rising multiple is borrowing return from the future, paying more today for the same stream of earnings. A market that rallies while margins make new highs is being paid by the earnings themselves. The first kind of rally is fragile to any change in rate or sentiment. The second kind has a floor under it, because the cash flow is real and growing. The desk has argued through the higher-for-longer regime that EPS-driven Tech could carry the index even with the rate complex offering no help, and the margin series is the receipt for that argument.

There is a rotation inside this margin story that the headline number hides, and it is the part we care about most. The intra-tech map we built in BSF #52 and confirmed in BSF #55 separated the parts of technology whose margins are still expanding from the parts whose margins are being competed away. Software, the sector the market left for dead a year ago on the assumption that AI would eat it, has instead turned its margin structure back up, which is why it surged the way it did into the close of BSF #57. Memory chips sit in the same camp, pricing power returning as the capex cycle tightens supply. The stretched, crowded end of semis is the part of the complex where the margin case is thinnest relative to the price, and our positioning continues to favor the expanding-margin names over the consensus winners. A record index margin built on a widening internal dispersion is exactly the kind of tape that rewards selection over beta.

The caution we owe the reader is that record forward margins are also, by construction, a high bar. Analyst margin assumptions embed a great deal of optimism about the AI earnings boom, and Callum Thomas was right to ask this week whether the long-term earnings growth estimate now closing in on record highs is a boom or a bubble. We will take the margin print as confirmation of the structural read, while keeping the BSF #53 four-times-in-history composition signal on the page as the standing reminder that record fundamentals and record concentration have historically arrived together.

The Labor Floor Holds

If margins are the structural story, labor is the cyclical floor, and the floor held this week. Private payrolls rose 122,000 in May, the largest of eleven straight monthly gains, with hiring broad-based across company sizes and small businesses contributing 67,000 of the total. Wage growth ticked down only marginally for job changers, to 6.5%, and held steady for job stayers at 4.4%. This is not the labor market of an economy rolling into recession. It is the labor market of an economy still adding workers at a measured, sustainable pace, with wage pressure cooling at the edges rather than collapsing. Eleven straight monthly gains in private payrolls.

SOURCE: Daily Chartbook DC Lite #612, Haver Analytics via Winnie Tapasanun (ADP Private Payrolls +122K, 11 Straight Gains)

The reason this matters for positioning is that the most common trigger for the rotation we are long to fail is a growth scare. A genuine deterioration in hiring would pull the cyclical barbell apart, send the defensive corners of the market from underweight to crowded, and validate the bears who have been waiting for the consumer to crack. May gave them none of that. The small-business contribution is the detail worth holding, because small firms are the most rate-sensitive employers in the economy, and their continued hiring through a higher-for-longer regime tells us the private-credit-stress lead indicator from BSF #49 has not yet started to bite the real economy.

We keep that BSF #49 indicator near the top of the tactical watchlist precisely because labor strength can mask financial-plumbing stress until it does not. A solid ADP print does not retire the risk, it simply tells us the transmission from tighter conditions to slower hiring has not arrived in the data yet. The desk reads the print as a green light for the cyclical position and an amber light, not red, on the credit watch.

When Bonds Stop Hedging

Here is where the week turned, and where the dissent earns its place on the page. For a generation, the reflex hedge against an equity drawdown was duration. You owned stocks, you owned bonds, and when stocks fell the bonds rallied and cushioned the blow. That reflex is breaking. Equity drawdowns are now being met by much shallower bond rallies than we were accustomed to. The roughly 20% drawdown in the S&P 500 after Liberation Day saw bonds gain barely 3% at the peak, a fraction of the offset that the same drawdown would have produced in the prior decade. Bonds gaining barely three percent at the peak.

SOURCE: Daily Chartbook DC Lite #612, TS Lombard via Daniel von Ahlen (Equity Drawdowns Now Meet Shallower Bond Rallies)

This is the most important chart of the week for anyone who runs a balanced book, and it is the cleanest expression of the risk we have been circling since the BSF #55 read on global yields rising in sync. When the stock-bond correlation flips positive, the diversification that investors assume they own quietly disappears. A portfolio that looks hedged on a spreadsheet is, in practice, long a single macro factor. That is fine while the tape climbs. It becomes the whole story the moment a real shock arrives, because the asset that was supposed to rally instead falls alongside equities, and the drawdown that the model said would be cushioned arrives at full force.

This pairs with the macro read coming out of the rates and currency desks, where the framing of the week was the global carry trade sitting on the edge of a cliff. The two ideas are the same idea seen from different windows. A world where duration no longer hedges equity risk is a world where leverage built on cheap funding has fewer escape routes when volatility returns. We do not think the cliff gets jumped this week. We do think the BSF #54 vol-regime trigger, the down-day variance flipping above up-day variance, is the single indicator most worth watching here, because it is the one that would fire first if the plumbing started to seize.

The Base Rate Still Points Up

Against that plumbing risk, the tape offers its own historical base rate, and it is constructive. The S&P 500 gained more than 16% over the two-month period ending May 26. Rallies of that magnitude and speed have, more often than not, preceded strong forward returns rather than marking exhaustion. This is the part of the analysis the permabears consistently get wrong. A sharp, broad, momentum-driven advance is not a contrarian sell signal by itself. History says it is closer to the opposite, that strength begets strength over the medium term, and that the moves which look unsustainable in the moment tend to resolve higher. More than sixteen percent over two months.

SOURCE: Daily Chartbook DC Lite #612, Daily Chartbook Studies (SPX Two-Month 16% Rally Forward-Return Study)

The base rate is not a promise, it is a probability, and we treat it as one. What it does is set the burden of proof. For the bullish path, the burden is simply that the historical tendency repeats and the economy avoids a shock. For the bearish path, the burden is heavier, because the bear has to argue that this time the momentum signal inverts, that the rally which historically led to gains instead leads to a top. That is not an impossible argument, the BSF #54 RSPD and RSPS three-month-forward signal is exactly such an argument, and its window still has roughly 77 days left to run as of this issue. But a momentum base rate this constructive means the bear is fighting both the tape and the history of tapes like it.

There is a second confirmation hiding in the composition of the move. The software and energy rotation map from BSF #55, the one we marked as confirmed in BSF #57, is precisely the kind of leadership that turns a momentum base rate from a statistic into a thesis. Rallies that lead to strong forward returns tend to be led by cyclicals and by the under-owned corners of the market rather than by a last, narrow push in the names everyone already holds. This one is being led by software off a washed-out base and by the cyclical capex complex, which is the leadership profile the base rate likes. The advances that fail are usually the ones carried by a shrinking handful of mega-caps while everything else rolls over, and the breadth resolution we documented in BSF #57 is the evidence that this is not that.

We hold the base rate as a tailwind, not a thesis. It tells us which way to lean while the harder signals, the vol regime and the credit watch, stay quiet. It does not tell us to stop watching them.

The Tape and the Speculation Flag

Step back to the monthly chart and the picture composes. The S&P 500 closed May up 5.15%, up 10.73% on the year, comfortably above its ten-month moving average, confirming the first quarter wobble as the third major bull-market correction since the 2022 low. On the trend-following read that has guided the strategic position all year, the message is unchanged. The trend is still your friend. Confirming the third bull-market correction since the 2022 low.

SOURCE: The Weekly ChartStorm 31 May, Topdown Charts via Callum Thomas (SPX Monthly, Above the 10-Month Average)

But the same source that drew that monthly chart spent the rest of the week cataloguing the other side, and we would be doing the reader a disservice to leave it out. The share of investors expecting higher prices a year from now reached its second-highest reading ever in May. The average investor is holding a record allocation to equities. A composite of stretched market indicators sits at levels that do not signal an imminent crash but do mark these as unusual times. Semiconductors are as overbought as any sector has been in years. Median stock short interest has surged to decade-plus highs. And the buyback bid that powered so much of the post-2022 advance has faded from a major tailwind to nearly irrelevant just as a wave of new issuance approaches.

This is the speculation generation we first flagged through Callum Thomas in BSF #57, and it has not cooled. The honest way to hold it is to separate the two clocks. Sentiment and allocation are stretched, which is a condition, not a catalyst. Positioning in the cyclical rotation is not stretched, which is why the rotation can keep working even as the broad sentiment runs hot. The market can be expensive and under-owned in the same breath, expensive at the index level on sentiment and valuation, under-owned at the sector level where the leadership actually sits. That tension is the whole story of BSF #58, and it is why we are neither chasing nor fading here.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

The picture is more coherent than any single chart. Labor is still adding jobs, with private payrolls up 122,000 in May and small businesses still hiring. Margins are pressing to new highs, keeping the EPS-driven structural-long thesis intact. The two-month rally carries a constructive historical base rate. And the positioning that sits under all of it is the least cyclically tilted since 2023, which means the rotation we are long is under-owned rather than crowded. On the weight of the evidence, the tape has earned the benefit of the doubt, and we lean with it.

The single dissent we will not wave away is the one coming from the bond market rather than the stock market. When equity drawdowns are met by 3% bond rallies instead of the double-digit cushions of the past, the diversification most balanced portfolios assume they own has quietly gone missing, and the global carry trade built on cheap funding has fewer exits than it thinks. This is a structural vulnerability, not a timing signal. It sits alongside the standing bearish book: the BSF #53 four-times-in-history composition signal, the BSF #55 margin-debt read, the BSF #56 FOMC minutes that were the most hawkish since July 2023, and the BSF #54 RSPD and RSPS window still live for another 77 days. None of these is firing today. All of them would matter more the day the vol regime turns.

So the strategic position is unchanged and the conviction is a notch higher. We stay long EPS-driven Tech and the intra-tech rotation, long the cyclical capex barbell across industrials, copper, energy and clean energy, and short-to-flat the rate-sensitive long-duration cyclicals that the bond-hedge breakdown makes most fragile. The five tactical triggers hold their order, with the BSF #54 vol-regime read at the top of the list, then the software and energy rotation map, the 30-year yield, the 10y10y forward, and the BSF #49 private-credit-stress indicator. The cleanest single signal of the week was the 17th-percentile positioning print, and it says the same thing the tape says, only louder. The rotation is not crowded yet.

The closing punch is the one the data keeps making for us. A market top usually arrives when everyone is finally in. This is not that. Not crowded yet.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.

The BSF #56 tactical caution was correctly framed; the resolution direction it explicitly named as the bull-case path was the one that landed. The breadth catching up is exactly the scenario the verdict from May 21 said would keep the index grinding higher and the cap-equal gap closing. Below the fold: the breadth resolution that just changed the entire tactical picture, the software seven-week stretch that anchors the rotation, the macroeconomic reads that explain why the structural rally has held (Chicago PMI four-year high, only one of three cyclical GDP components contracting, ten yield-curve pivots since 2023), the Q2 EPS revision read that is the largest since Q3 2021, and what to actually do with a tape that just produced the longest weekly winning streak since 1985.

The Breadth Resolved Up

The single most important observation this week is structural. The read that crystallized the resolution is that the median stock in the S&P 500 has broken out. The equal-weighted S&P 500 has now broken out alongside the cap-weighted index. The Software ETF (IGV) has posted its second-strongest seven-week advance on record. Small caps (IWM) are roughly 70% off the 2025 lows. Each of those data points, in isolation, would be a constructive observation. Read together, the picture is that the breadth divergence the desk has been warning about since BSF #50 has resolved through the laggards catching up — not through the leaders correcting.

This is the positive-skew resolution that the BSF #56 verdict explicitly identified as the path of constructive interpretation. The framework from May 21: "record breadth divergences resolve in two ways. Either the laggards catch up (positive-skew resolution, late-cycle rotation) or the leaders correct (negative-skew resolution, the index re-rates to the equal-weight median)." Eleven days later, the resolution direction has been determined. The laggards caught up. The cap-equal gap that hit a record reading in BSF #55 is now closing through the equal-weight, not through the cap-weight.

The implication for positioning is that the strategic bull case has held, and the tactical caution from BSF #56 has been resolved without the BSF #54 RSPD/RSPS 3-month-forward signal cashing. The historical record is not deterministic, and 11 trading days is not a 90-day window — we should be honest about that. But the structural read is clear: when the breadth divergence resolves positively, the historical playbook says the rally extends because the index re-rates to a broader earnings base. The BSF #52 → BSF #56 intra-tech rotation thesis (memory, software, energy, copper, ag) is now part of a broader rotation pattern.

SOURCE: Topdown Charts Professional via Callum Thomas Weekly ChartStorm (Happy New Month — S&P 500 May close)

Software's Historic Seven-Week Stretch

The Software ETF (IGV) just posted its best single-day gain in over a year and closed higher for a seventh consecutive week. IGV has rallied more than 36% over that span. The dataset milestone: this is the second-strongest seven-week advance on record for IGV, trailing only the rebound off the post-dot-com lows in 2002. Software has also reclaimed its long-term 200-day moving average, and the relative breakout versus the S&P 500 confirms a regime change in the intra-tech leadership cycle.

This is the cleanest possible confirmation of the BSF #52 / BSF #55 / BSF #56 intra-tech rotation thesis. The framework from BSF #52: NVDA at 2-year relative lows, memory chips heating up, software ETF starting to base, the leadership is rotating inside the Tech complex. The framework from BSF #55: TrendSpider's IGV breakout was the leading indicator. The framework from BSF #56: software picks up where semis stalled. Eleven trading days later, the leadership cycle has produced an outcome the historical analog set says only happens at major regime shifts — the second-strongest seven-week run since the post-dot-com bottom.

The implication for positioning is that the IGV trade is now structural rather than tactical. We have held the software bias since BSF #55. The May 21-29 stretch is the confirmation that the bias was correctly sized. We do not chase the +36% move; we maintain the position and extend the framework to the adjacent sub-sectors (cybersecurity, vertical SaaS, data infrastructure). The +70% IWM move off the 2025 lows is the structural confirmation that the broader bull breadth is now extending into the small-cap complex.

SOURCE: Daily Chartbook DC Lite #609 — Daily Chartbook Studies (NDX rolling YoY > +1 std signal)

Chicago PMI Hit a Four-Year High

The macro read this week came in significantly more constructive than the BSF #56 framing assumed. The Chicago Business Barometer jumped to a four-year high in May, driven by increases in new orders (highest since January 2022), order backlogs, and supplier deliveries. The single most important data point: prices paid rose to the highest reading since May 2022. The bullish read on this print: the manufacturing complex is accelerating, which supports the BSF #56 US EPS revision differential. The bearish read: the prices-paid component at three-year highs is the inflationary input that reinforces the FOMC hawkish drift from BSF #56.

We do not reverse the BSF #56 framing on the back of one PMI print. We do update the tactical bias. The growth read is meaningfully more constructive than the consensus assumed, which supports continued strength in the cyclical complex (the BSF #52 → BSF #56 industrials, copper, ag, energy positioning). The inflation read is meaningfully more concerning than the consensus assumed, which supports continued weakness in the rate-sensitive long-duration complex (homebuilders, REITs, unprofitable growth). The net read: the Chicago PMI print sharpens the bull case for the EPS-driven Tech and cyclical-capex sectors, and sharpens the bear case for the rate-sensitive complex. The barbell positioning the desk has held since BSF #50 has now received fresh fundamental confirmation.

SOURCE: Daily Chartbook DC Lite #609 — Koyfin (Chicago PMI Four-Year High, Prices Paid May 2022 Highest)

One of Three Cyclical Components Is Contracting

EPB Research surfaced the macroeconomic framework that explains why the rally has held through every macro pivot the desk has tracked. The framework: less than 20% of GDP drives 100% of recessions, sitting in three components — residential investment, durable goods consumption, business equipment investment. Today, only residential is contracting. One of three. Not enough to break the cycle. This is the cleanest single-sentence formulation of the structural anti-recession read that has been the floor under the bull since BSF #50.

The structural framework is right. Recessions require a coordinated contraction across the cyclical components of GDP — historically, when only one of the three is contracting, the cycle holds. The residential weakness has been visible since BSF #50 (homebuilders at relative lows, rate-sensitive credit underperforming). The durable goods consumption read remains positive (consumer durables holding up). The business equipment investment read is the AI capex backbone (BSF #45 → BSF #56), which has been structurally accelerating rather than contracting. Two of the three legs are firm; the third (residential) is contracting in isolation. This is the structural reason every BSF in the run has held the strategic long bias even as the tactical signals have warned of dispersion.

SOURCE: Daily Chartbook DC Lite #609 — @epbresearch (Cyclical GDP Components — Only Residential Contracting)

Ten Pivots in Rates Since 2023

Treasury yields have now produced their tenth pivot since 2023. The reading is structural: the rates complex has been over-trading the Fed-pivot narrative for nearly three years, with each pivot reversing into either a higher-for-longer or a deeper-cut-than-priced regime within weeks of forming. The cumulative implication: the consensus has consistently been wrong on the direction of yields, and the most recent FOMC minutes (BSF #56) have just established the highest-frequency hawkish-drift signal since the cycle started.

We hold the BSF #56 read on yields: the long-duration trade has been structurally repriced through the back of the curve, and the front of the curve has now started to absorb the same read after the FOMC minutes printed as the most hawkish since July 2023. The 10th-pivot framework adds a useful asymmetry observation: when the curve produces this many pivots in a 36-month window, the next directional move historically takes the path of least resistance rather than the consensus path. That path of least resistance is currently higher-for-longer, which is what BSF #54 → BSF #56 has been mapping. The Chicago PMI prices-paid read confirms it.

SOURCE: Daily Chartbook DC Lite #609 — @michaelkantro (Ten Treasury Yield Pivots Since 2023)

The Largest Bottom-Up Revision Since 2021

The bottom-up Q2 EPS estimate for the S&P 500 has been revised up by 3.8% during the first two months of the quarter. This is the largest such increase since Q3 2021. The structural significance: in a typical quarter, analysts reduce earnings estimates during the first two months. The Q2 2026 print has produced the cleanest single read on the strength of the underlying earnings backbone that the desk has tracked this cycle.

This is the structural confirmation of the BSF #56 US-EPS-revision-differential thesis. The framework from BSF #56: Nasdaq EPS revisions 2.5, SPX 1.8, S&P 400 1.5, S&P 600 1.3 — all above global peers. The framework from BSF #55: tech P/E getting cheaper (forward 23.6, down from 30 last fall). The framework from BSF #45: AI capex cycle generating earnings at structurally faster rate than the macro-rate repricing is compressing the multiple. Eleven trading days later, the Q2 EPS revision print has confirmed all of the above. The cleanest historical analog (Q3 2021) was followed by a final upside leg of the post-COVID equity rally before the 2022 bear set in. We do not claim the 2026 setup is identical to 2021; we do note that the historical pattern says strong bottom-up Q2 revisions extend the rally through the back half of the year before the cycle reaches its terminal phase.

SOURCE: Daily Chartbook DC Lite #609 — John Butters, FactSet (Q2 Bottom-Up EPS Revision Largest Since Q3 2021)

The companion bullish base-rate read came out of the NDX rolling YoY framework. The Nasdaq-100 rolling YoY metric crossed above the +1 standard deviation threshold in April 2026 for the first time in over a year. This signal has only flashed seven prior times since 1985 (small sample, acknowledged). The NDX was higher 12 months later in six of those seven cases — an 86% hit rate. That is the cleanest single bullish base-rate signal on the desk.

Charts Worth Watching

A handful of tweet narratives worth their own callout before The Verdict.

The Verdict

Let us call the picture clearly. The breadth divergence the desk has been mapping since BSF #50 has resolved positively — the median stock broke out, the equal-weighted SPX broke out, software has posted its second-strongest seven-week advance on record (trailing only the 2002 post-dot-com bottom), small caps are +70% off the 2025 lows. The S&P 500 has notched nine straight weekly gains — matching the longest weekly winning streak since 1985. Software has rallied +36% in seven weeks. The Chicago PMI has hit a four-year high, with prices paid at the highest level since May 2022 — bullish for the cyclical complex, inflationary for the FOMC hawkish drift. Only one of the three cyclical GDP components is contracting; the cycle is not breaking. The bottom-up Q2 EPS estimate has been revised up by 3.8% — the largest first-two-months revision since Q3 2021. The NDX rolling YoY +1-std signal has flashed for the eighth time since 1985 with an 86% historical 12-month positive hit rate.

The single dissenting bearish read is the structural-speculation framing from the weekend ChartStorm. Survey expectations of higher stock prices a year from now reached their second-highest reading ever in May. Investor allocations to equities are at record highs. The long-term average earnings growth estimate is closing in on record highs, supported by the AI earnings boom — the framing used is "generational run in stockmarket speculation." Median stock short interest at decade-plus highs (which is dry tinder for further short-squeeze fuel if the rally extends, but also a sentiment-extreme read). The BSF #55 "Margin Debt Says 2021" read remains structurally intact. The BSF #56 FOMC hawkish-drift read remains intact. The BSF #54 RSPD/RSPS divergence is still present — the 3-month forward window from May 18 still has roughly 80 days to run.

Our stance: we maintain the strategic long position, we extend the BSF #55 / BSF #56 software bias into the broader tech-rotation framework (cybersecurity, vertical SaaS, data infrastructure), we lean into the BSF #50 → BSF #56 cyclical-capex barbell (industrials, copper miners, ag, energy) which the Chicago PMI four-year high directly supports, and we avoid rate-sensitive long-duration cyclicals (homebuilders, REITs, unprofitable growth) which the prices-paid component and the FOMC hawkish drift both pressure. The BSF #56 tactical tightening from May 21 has been resolved without the RSPD/RSPS signal cashing; we relax the tactical caution but keep the structural watchlist intact. The five cleanest tactical-trigger leading indicators stay the same: the BSF #54 vol-regime read (down-day variance flipping above up-day variance), the BSF #55 Software/Energy rotation map (now confirmed), the 30-year yield, the 10y10y forward, and the BSF #49 private-credit-stress lead indicator.

The breadth resolved up. Nine straight weekly gains, the longest since 1985. Software has posted its second-strongest seven-week advance ever. The cycle is not breaking — only one of three cyclical components is contracting. The Q2 EPS revision is the largest since 2021. The historical NDX rolling YoY signal has an 86% 12-month hit rate. We stay positioned strategically long, with the BSF #55 software bias extending and the BSF #50 → BSF #56 cyclical-capex barbell intact. The bear-case dissent from the structural-speculation framing remains on the watchlist. The trend is still your friend for now.

Disclaimer

This newsletter is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The views expressed are those of the author and may change without notice. All investments carry risk, including the risk of loss. Past performance does not guarantee future results. Always conduct your own research and consult a licensed financial professional before making investment decisions.