That is the whole business. No platform, no ecosystem, no AI angle. Just a small team of people who are good at buying real estate cheaply and patient enough to sell it dearly.

Here is what makes it interesting. The stock trades at roughly 0.66 times its reported book value. That alone is not remarkable; plenty of Japanese small caps trade below book for good reasons. What is remarkable is what sits behind that book value.

The number nobody puts on the balance sheet

Japanese accounting requires companies to carry rental real estate at depreciated historical cost. If you bought a building in 2013 and it has doubled in value, your balance sheet does not know that. The gain shows up in a footnote, in the notes to the financial statements, under a line item most investors never open.

For this company, that footnote currently reads 16.67 billion yen of unrealized gains on rental property. Book equity is roughly 26.0 billion yen. Management themselves describe the hidden gain as equal to just over sixty percent of shareholders’ equity.

So the arithmetic goes like this. You pay about 17.8 billion yen of market capitalization. You get 26.0 billion yen of book equity. Behind that equity sits another 16.67 billion yen of value that the accounting refuses to acknowledge. Tax the hidden gain at thirty percent and you land somewhere around 37.7 to 38.4 billion yen of adjusted net asset value, or roughly 2,740 yen per share against a share price near 1,270 yen.

That is forty six cents on the dollar. And it assigns exactly zero value to the fee earning asset management arm, which runs 14.56 billion yen of client capital and carries no book value whatsoever.

Is the footnote real?

This is the right question to ask, and I want to answer it with evidence rather than assertion, because appraisal values are opinions and opinions can be flattering.

In March of this year, the company sold a lodging facility in Kamakura. The property sat on the books at 1.71 billion yen. It sold for 2.68 billion yen, a fifty seven percent premium to carrying value, producing a 970 million yen extraordinary gain. Separately, it exited an equity stake in a drone technology company for 590 million yen, booking a 280 million yen gain on a position most shareholders did not know existed.

When a company sells assets meaningfully above the carrying value implied by its own disclosures, the disclosures are conservative rather than promotional. That is the single most important fact in this write up.

What the cash actually looks like

Reported operating cash flow for this business is close to useless, and I want to explain why, because it is the reason screens miss the name entirely.

Japanese property investors book purchases of inventory real estate through operating cash flow. In a year when the company is buying, operating cash flow looks catastrophic. In a year when it is selling, it looks heroic. Neither tells you anything about the underlying economics.

Two things do. First, the rental book: 64.8 billion yen of property at acquisition cost across ninety two assets, generating a 7.1 percent net operating income yield against a weighted average borrowing cost of 1.80 percent. That spread of roughly 5.3 points is the durable engine, and it runs whether or not anything is sold in a given quarter.

Second, and more tellingly, book equity has gone from 7.8 billion yen at the time of listing to 26.0 billion yen today, a compound rate of about 12.8 percent, with no loss year in a decade. That is the honest owner earnings signal for a business like this. The company grew that equity without issuing stock and diluting anyone, which I regard as the cleanest tell of management quality available in this market.

Last fiscal year ordinary profit came in at 4,433 million yen, roughly double the prior year. Through the first half of the current year, net income had already reached ninety three percent of the full year plan while management left guidance untouched, which tells you they would rather beat quietly than promise loudly.

The company

Read more

The rock this time was Japan. At the bottom of the list, below several hundred names I had never heard of, sat a company with a market capitalization of about 165 million yen. Roughly 1.1 million dollars. Not 1.1 billion. One point one million, which is what a modest apartment costs in the city where the company is headquartered.

The name was 軽自動車館, Keijidoushakan, ticker 7680. It translates to something like “Kei Car House.” It sells used kei cars, the yellow-plated 660cc microcars that make up roughly a third of the Japanese vehicle fleet, out of eleven stores in Hokkaido and one corner of Tohoku. Headquarters is on the eleventh floor of an office building in central Sapporo.

I decided to underwrite it properly. Not skim it. Actually work through it the way you would if you were buying the whole thing. What follows is what I found, including the part where I discovered I could not buy a single share.

The price has not moved since 2019

The first thing that looked wrong was the chart, in the sense that there was no chart. The last trade was 3,190 yen. So was the fifty-two week high. So was the fifty-two week low. So was the all-time high and the all-time low. Volume: zero.

That price is from September 2019, the day the company listed. In the nearly seven years since, the stock has essentially not traded.

There are 51,600 shares outstanding. Not 51.6 million. Fifty-one thousand six hundred, which is fewer shares than a mid-size company has employees. The founder controls the overwhelming majority of them. Whatever float exists is a rounding error.

The explanation turned out to be the market segment. This is not listed on the Prime, Standard, or Growth boards of the Tokyo Stock Exchange. It is listed on TOKYO PRO Market, a professional-investors-only venue that the TSE runs alongside its main boards, originally set up in 2009 as a joint venture with the London Stock Exchange and modeled on London’s AIM.

TOKYO PRO Market has no minimum shareholder count, no minimum float, no minimum market capitalization, and no minimum profit. That is the entire point of it. A founder can list the company, get the credibility and the recruiting benefit and the bank relationships that come with being a listed company in Japan, and keep essentially one hundred percent of the equity. There is no forced distribution of shares, so there is no float, so there is no trading.

An unexpected loophole, and why it does not help

Buying on TOKYO PRO Market is restricted to two categories of investor. The first is specified investors, the Japanese regulatory category for professionals. Institutions qualify automatically. An individual has to apply to a Japanese broker and clear one of several tests, roughly 300 million yen of investable assets plus regular trading activity, or 100 million yen combined with a relevant professional qualification.

The second category is non-residents of Japan.

Read that again, because it produces a strange result. Non-resident is a residency test under the Foreign Exchange Act, not a nationality test. An American living in Ohio is a non-resident of Japan and is therefore categorically eligible to buy TOKYO PRO Market shares, with no wealth test at all. An American living in Tokyo is a resident, does not qualify, and would have to clear the 300 million yen hurdle. A Japanese citizen living in London is eligible.

The logic is not as strange as it looks. Japan’s investor protection regime is territorial. The FSA is not trying to protect retail investors in Ohio, on the reasonable theory that this is the SEC’s job.

I spent an hour being pleased with myself about this before running into the wall. Eligibility is permission to be on the other side of a trade. It is not a route to the order book. You still need an account at a Japanese securities firm that routes orders to TOKYO PRO Market, and no foreign broker does. Interactive Brokers, Schwab, Fidelity, Saxo: all of them give you Prime, Standard, and Growth and stop there. The firms with PRO Market capability are the domestic underwriters and J-Advisers, and opening a retail account at a Japanese broker as a non-resident with no Japanese address and no My Number is close to impossible. Most will not take you, and many will convert an existing account to sell-only if a client moves abroad.

The non-resident channel exists in practice for institutions. Foreign funds trading Japan through a global custodian or a prime broker can access it. That is who the rule was written for. A person with a brokerage account cannot.

And even if you cleared all of that, there is no resting offer in 7680. You would be negotiating a block directly with the founder, which is a relationship problem rather than a brokerage problem.

It is not the Expert Market

The obvious comparison for an American is the OTC Expert Market, where securities go when the issuer stops publishing current information and broker-dealers can no longer quote them. Retail can sell but not buy. The practical effect is similar. The cause is close to opposite.

Expert Market names are dark because the issuer went dark. TOKYO PRO Market names are dark because the buyer pool is walled off while the issuer keeps filing. Keijidoushakan publishes audited semiannual earnings releases with a full balance sheet and cash flow statement, plus an annual issuer information document lodged with the exchange. It has an auditor. It files controlling-shareholder disclosures every year. The disclosure regime is lighter than the main board, with no quarterly reporting and no internal control report requirement, but it is a real regime and the company complies with it.

The Expert Market is a penalty box. TOKYO PRO Market is a waiting room. Companies do graduate from it to the general market, and the exchange promotes it as an entry venue for exactly that reason.

Which is why I kept going. If a company files audited financials, you can underwrite it, whether or not you can buy the stock.

Reading the filings

One practical note for anyone who tries this. Keijidoushakan does not file to EDINET, Japan’s equivalent of EDGAR. TOKYO PRO Market issuers file to the exchange instead. That means there is no XBRL, which means no data vendor has anything. I checked four of them and every field came back empty. Everything below was hand-keyed out of PDFs.

The recent history goes like this.

For the twelve months ended April 2025, the company did 4,181 million yen of revenue, about 28 million dollars, and generated 9 million yen of operating profit. Nine. On four billion yen of sales. It lost money at the net line. Used vehicle acquisition costs were rising and the company could not pass them through fast enough. Unit sales fell almost twelve percent in the first half of that year as it chose price over volume.

Then it worked. The company changed its fiscal year end from April to October, producing a six-month stub period, and closed a loss-making store in Sapporo at the end of May 2025. In the six months to October 2025 it did 2,366 million yen of revenue and 74 million yen of operating profit. In the following six months, to April 2026, it did 2,373 million and 46 million.

Stitch those together and the trailing twelve months to April 2026 shows 4,739 million yen of revenue and 120 million yen of operating profit. Operating profit went from 9 million to 120 million in a year, on thirteen percent revenue growth.

The mechanism is visible in the cost structure. Selling and administrative expenses were 812 million yen in the bad year, against 821 million of gross profit. The entire gross margin was consumed by overhead. In the trailing year, gross profit grew by something like 125 million yen while overhead stayed roughly flat. Every incremental yen dropped through. That is operating leverage on a fixed cost base, and it works in both directions, which is the whole problem.

Here is the part that gave me pause. Management is guiding the current year to 80 million yen of operating profit, not 120 million. They reaffirmed that guidance in June 2026, after a strong first half, implying a second half of about 34 million against 74 million in the year-ago period. Either this is the sandbagging that Japanese small caps are known for, where guidance functions as a commitment rather than a forecast, or management sees the margin recapture reversing. It is a fifty percent swing on the earnings you would be buying and I could not resolve it from public documents.

What the business actually is

Eleven stores. About a thousand cars in stock. The company buys kei cars at auction in Honshu, specifically because Honshu cars have less rust than cars that have spent their lives on salted Hokkaido roads, ships them north, and sells them with free inter-store transport, no accident history, and a warranty. It also buys trade-ins, does maintenance and inspection work, sells insurance as an agent, and offers residual-value credit on new cars.

The arithmetic per car is roughly this. Buy for about 900,000 yen. Hold it for ninety-nine days. Sell for something like 1.15 million. Keep about 235,000 yen of gross profit. Do that between 3,500 and 3,900 times a year. Gross margin runs a little under twenty percent.

Two of those numbers matter more than the rest.

Ninety-nine days of inventory is slow. Well-run Japanese used car retail turns stock in forty-five to sixty days. Every extra day is depreciation, financing cost, and an eventual markdown. Getting to sixty days would release something like 360 million yen of working capital, which is more than the entire equity of the company.

And a business with a twenty percent gross margin and nineteen percent operating costs has no margin of safety at all. It works at twenty and it dies at eighteen. The year ended April 2025 was not a hypothetical stress case. It was two years ago, and it produced nine million yen of operating profit on four billion of revenue.

Where the money is, and where it is not

The balance sheet at October 2025 shows total assets of 1,526 million yen, of which 907 million is used cars. Cash is 133 million. Against that sits 889 million yen of borrowings and bonds, leaving 259 million of equity and an equity ratio of seventeen percent. The quick ratio is 0.21, meaning there is almost nothing liquid behind the current liabilities except the cars themselves.

I went looking for hidden assets, because that is usually the reason to be interested in a Japanese microcap. Land carried at 1970s cost, cross-shareholdings, a securities portfolio, rental real estate at a fraction of market value. There are none. Land on the balance sheet is 121 million yen for a company operating eleven sites, which tells you that it owns almost none of its real estate. There are no cross-shareholdings and no investment securities. The largest asset is a depreciating pile of used cars.

That has a consequence worth stating plainly. Run a liquidation and the recovery lands at or below zero for equity. Haircut the inventory to wholesale, write off the deferred tax asset, recover most of the lease deposits, subtract 1,266 million of liabilities and the cost of winding down eleven stores, and there is nothing left. Every yen of value in this company is going concern value. There is no asset floor.

The multiples, and why they lie

At the last trade of 3,190 yen, the trailing price to earnings ratio is 2.1 times. Price to book is 0.63. The earnings yield is forty-seven percent. On the screen, this is the cheapest thing in the country.

It is also almost entirely an artifact of leverage. Market capitalization is 165 million yen sitting on top of 756 million yen of net debt. The equity is fifteen percent of the enterprise. Value the business rather than the stub and enterprise value to EBITDA is about 5.9 times on trailing numbers, which is a perfectly ordinary price for a Japanese small cap auto retailer. On management’s guidance it is eight times, which is not cheap at all.

That gap between a two times earnings multiple and a six times enterprise multiple is the single most useful thing this exercise produced. It is the same gap that shows up on every over-levered screen hit, in every country, and the screen will never show it to you.

Working through it properly, treating the short-term borrowings as vehicle floorplan the way auto retail convention requires, and using a Japanese small-company M&A multiple of three and a half to five times EBITDA rather than a listed comparable, I get an equity value somewhere between 200 and 460 million yen, with a base case around 330 million, or 6,400 yen a share. That is about 2.2 million dollars. It is roughly 1.3 times book and roughly double the stale listing print.

I would not put much weight on that gap to the last trade. A 2019 price that has never been tested is not evidence of mispricing. It is evidence of no pricing.

What I would do if I owned it

Since the underwriting exercise was hypothetical anyway, I took it to the end.

The interesting lever is not cost cutting and it is not new stores. It is back-end gross profit. The company already holds an insurance agency licence and already offers a residual-value credit product, so the infrastructure exists. On 3,500 units a year, every 20,000 yen of incremental attach from extended warranty, paint protection, maintenance packages, and finance commission is 70 million yen of gross profit, which is comparable to the entire current operating profit of the business. The large listed used car chains in Japan run this as a core discipline. Whether this company already does is the question that decides whether the whole thing is interesting, and it is not disclosed.

Second is the inventory turn, which needs pricing discipline and aging triggers rather than capital.

But the first question I would ask is not operational at all. The company owns 121 million yen of land while operating eleven stores. So who owns the other sites? In a Japanese owner-managed business the answer is very often the founder personally, or a family asset-holding company, leasing the property back to the operating business. If that is the case here, the rent is negotiable, the leases are a change of control risk, and the seller has a second asset to sell you. That single fact could move value more than any of the operating levers, and it is invisible from the public filings.

What the exercise was worth

I cannot buy this stock. Neither can you, in all likelihood. It will sit at the bottom of every Japan screen either of us runs, at 3,190 yen, forever, or until the founder decides to step up to the main board or sell the company.

The screen filter that removes it is a minimum twenty-day average traded value. Five million yen a day clears out every PRO Market listing at once. Fukuoka launched its own professional market at the end of 2024, so there will be more of these artifacts, not fewer.

But I would run the exercise again. Two things came out of it that I would not have gotten from a name I could actually trade.

The first is the leverage lesson, which is old but which lands harder when you have watched it happen inside a specific set of numbers. A two times earnings multiple on a company where the equity is fifteen percent of the enterprise is not a cheap company. It is a cheap option on a company, and the option can go to zero in a year that looks a lot like the year ended April 2025.

The second is that the thing which makes this untradeable as a security makes it clean as an acquisition. One controlling shareholder, no minority squeeze-out mechanics, no tender offer rules to navigate, and a voluntary delisting process. The illiquidity that makes it impossible to buy 100 shares makes it simple to buy 51,600.

That is not a useful conclusion for anyone with a brokerage account. It might be a useful one for somebody in Sapporo with a strategic interest in eleven kei car lots and a founder who has been running them since 1997.

In May this company rang the closing bell at the Nasdaq MarketSite in Times Square. The founder gave a speech about a new chapter. The stock had priced at eight dollars. On its fifth day of trading it touched thirteen dollars and ninety five cents.

As I write this it trades at about a dollar twenty one.

Read that again slowly. Down eighty five percent from the offer price in ten weeks. Down ninety one percent from the high. Not on a fraud report, there is none. Not on a lost contract, none has been announced. Not on an earnings miss, the last report showed operating profit up nine percent, net income up twenty percent, and a record gross margin. Not on short sellers, because I checked the exchange data and the most recent published count of total short interest was five hundred and thirty six shares. Not five hundred thousand. Five hundred and thirty six, a position a dentist could cover with a market order.

The stock has simply fallen, day after day, ten percent at a time, on volume of a couple hundred thousand shares, the way a stone falls down a well. On July 8 it was three nineteen. On July 15 it was two forty six. On July 17 it closed at one seventy seven, down another ten percent, at a fresh all-time low, and it has kept going since.

So what does a dollar twenty one buy you?

It buys you a company with roughly seventy two million dollars of market capitalization. Against that, the last audited balance sheet showed about fifty three million dollars of cash, before the twenty two million or so of net proceeds from the IPO landed in the till in May. Call it seventy five million of cash, against about thirty five million of borrowings. The enterprise value, the price of the actual operating business after you net out the money, is somewhere around thirty million dollars.

For that thirty million you receive a business that produced fifteen million dollars of operating profit last year and thirteen million dollars of operating cash flow. You are paying a bit over two years of cash flow for the whole enterprise. Under two times EBITDA. Around seven times earnings. A quarter of one year’s revenue. For a profitable, growing, net-cash, twenty-year-old supplier whose two largest customers are Toyota and Honda and whose auditors found the cash exactly where the company said it was.

And the insiders. The founder owns ten and a half percent. His CTO owns nearly eight. His CFO owns six. The COO owns another six. Add the two automakers and a third strategic holder and roughly two thirds of this company is in the hands of people who have watched, together, on paper, roughly a quarter of a billion dollars evaporate since May, and not one of them has filed to sell a single share.

I have spent the last stretch underwriting this company the long way. The prospectus, the annual report, the earnings call, the Japanese-language disclosures and press releases the American market has never read, the ownership register, the lock-up mechanics, the short data, the product releases from the Kobe headquarters, and the technology the company is betting its future on, which turns out to involve dashboard cameras, government 3D city maps, and a plan either five years ahead of its time or five years too late.

Below the paywall I will give you the name, the ticker, the full valuation workup, the full ledger, the reason I believe the stock is falling when no one else has published one, the identity of the single customer that decides everything, the honest and uncomfortable findings my digging produced, including two discoveries that cut directly against the easy bull case, the exact date in November when this situation either resolves or ruptures, and the three roads forward with numbers on each. Because here is the thing about a stock at two times cash flow: the valuation is never the question. The question is always whether the cash flow, and the cash, will ever belong to you.

This one is not like the paper company. This one is small, foreign, illiquid, and structurally orphaned, and it could genuinely go either way. That is not a disclaimer. That is the analysis.

Read more

The whole company, all of it, the trillion yen of revenue and the forests and the mills and the cash and the lawsuit, trades for about one hundred fifty seven billion yen.

Read that again slowly. The stock market will sell you this enterprise for 0.31 times its book value. The company itself discloses, with a straight face, in its own filings, that its market capitalization equals 8.4 percent of its total assets. Its price to book ratio has spent the last decade wandering between 0.2 and 0.7 times, and it sits today closer to the bottom of that range than the top, while three separate things have changed in the last eighteen months that have never all been true at once in this company’s modern history.

First, a new chief executive took over in June 2025, the first leadership change in six years, and he is not a finance man or a sales man but a forester, a man who joined in 1988 out of a forestry department and spent his career buying wood, and he was handed the job specifically to fix the two things that have bled this company for a decade.

Second, in May 2026 that new chief executive published the first five year plan in the company’s history that reads like it was written for shareholders instead of at them. Return on equity of eight percent or better. Return on invested capital of four percent or better. Operating profit more than doubling. Net debt to equity cut from 1.7 times to below 1.0. And, buried in the plan like a confession, a policy of abolishing cross shareholdings entirely, in principle, all of them, starting with twenty five billion yen of sales over five years.

Third, in April 2026 the company signed a memorandum of understanding with two other listed players to explore combining forces in containerboard, the cardboard that every Amazon box in Japan is made from, in an industry that everyone has known for twenty years must consolidate and never has. The last time Japanese paper consolidated, fortunes were made by the people holding the cheap paper.

The dividend was just raised fifty percent. The payout ratio is still under fifteen percent, which tells you how much room is left. The market has noticed none of it.

I spent the last two weeks underwriting this company the long way: ten years of results, the segment economics line by line, the register, the credit file, the officer filings, the franchise-style dependency at the heart of its worst division, the litigation docket in Australia. Below the line I will give you the name, the ticker, the ten year ledger, the segment by segment unit economics including the one division earning a thirteen percent margin that nobody talks about, the honest bear case, and the three roads with numbers on each. This is the longest piece I have written this year and I think it is the most asymmetric large cap situation on the Tokyo exchange.

Read more

Credit where it’s due before we roll: I first came across this name over at Tokyo Deep Value, which is exactly the kind of dusty roadside diner of an idea source you want on this beat — the work below is my own, but the tip of the hat goes there.

I want to tell you about this company the way it deserves to be told — slowly, completely, the way you’d describe a town you drove through so many times you finally stopped and stayed. Because Buffalo is not a story stock. There is no AI in it, no dream, no narrative arc that ends in a keynote presentation. There is a parking lot in Saitama, a pit bay, a franchise agreement signed in 1983, and a pile of money the market has decided to ignore. Let’s go all the way through it.

Read more

Today’s company is built on the opposite premise, and I want that contradiction on the table from the first paragraph, because the entire letter turns on it. This is not a wager on entropy. It is a wager against a family that has spent a hundred and two years refusing to let anything fall apart at all — that has, in fact, organized the whole enterprise around the explicit, stated, slightly vertiginous ambition of surviving for a thousand years. You do not, here, buy the cash for less than the cash; this is no net-net, and if you came for a Graham corpse you will leave disappointed. You trade above book. You receive, in cash, only about a third of what you pay, not more than all of it. What you get instead is the rarer and sturdier specimen — a genuinely excellent business, throwing off real money, sitting on a fortress balance sheet, run by people who think the word “shareholder” describes a tourist who wandered into a cathedral they have been building for five generations and intend to keep building for thirty more.

The company is Bourbon Corporation (TSE: 2208). It makes biscuits — the good ones, the ones in every convenience store from Wakkanai to Naha: Alfort, the little chocolate sailing-ship; Lumonde, the crumbling cocoa-cream roll; the Petit series; Fettuccine Gummy. It operates out of Kashiwazaki, a small city on the Niigata coast in the snow country, and the market values the whole 102-year-old enterprise at roughly ¥78 billion — about $520 million, or rather less than a single mid-rise Tokyo office tower, for a company that sold ¥120 billion of snacks last year and has never in its listed life come close to losing money. Against that ¥78 billion it holds something on the order of ¥21 billion of cash and securities against almost no debt worth the name — a net cash position near ¥18 billion, enough to buy back nearly a quarter of its own stock and not feel it. The cash is a third of the price. The operating business — the number-one biscuit franchise in the country — is the other two-thirds, and as we will see, even that is being handed to you at a price that does not make a great deal of sense.

So why does it trade here? The same reason they all do: something is wrong, or looks wrong, or is structurally arranged to stay wrong for longer than most investors can stand. With Bourbon the business is not the problem. The problem is the family, the thousand years, and the wall they have built around the money. But we will get there.

Read more

Today’s company is a different specimen, and I want to be honest about that from the first paragraph, because the whole letter depends on it. This one is not dormant. It is not melting. It does not trade below the value of its own bank account, and if you came here for a Graham net-net you should know now that you will leave slightly disappointed and, I hope, slightly more interested. What we have instead is the rarer animal — the cash box that grows, funded by the strangest and most attractive liability on any balance sheet I follow: its own customers, paying in advance, for a service they have not yet received.

The company is called KIYO Learning (TSE: 7353), it sits in an office near Nagatacho, in the political quarter of Tokyo, and the market currently values the whole thing at about ¥3.9 billion — call it $26 million, roughly what a mid-tier Tokyo convenience-store chain spends on rice balls in a long weekend. Against that ¥3.9 billion the company holds about ¥4.09 billion in cash and almost no debt — ¥450 million of borrowings, the kind of number a company this size could repay by accident. Net of everything it owes the banks, the cash alone is worth something on the order of ¥3.6 billion. Which is to say: the market is paying around ¥3.9 billion and receiving, before it values a single course, employee, or line of code, roughly ¥3.6 billion back in a bank account.

You are not, quite, being paid to take the business. ItoKuro and its cousins will do that for you. Here you are paying perhaps ¥300 million of true enterprise value — call it a rounding error — for a company that earned ¥304 million of operating profit last year, generated ¥726 million of operating cash, and grew its revenue 13% while doing it. The cash covers ninety-odd percent of the price; the operating business, one of the better small software-and-education franchises in the country, is being thrown in for the spare change in the market’s coat pocket. Graham did not have a tidy word for this one. The screener calls it an enterprise value near zero attached to a growing business, which is rarer than the negative-EV corpse, and considerably harder to dismiss.

So why does it trade here? The same reason all of these trade where they do. Something is wrong, or looks wrong, or used to be wrong. With KIYO the list is real, and the most damning item on it is not the business at all. It is the man who owns it. But we will get there.

How the machine got built

Read more

Today’s company is one of those. It is called ItoKuro (TSE: 6049), it occupies the fourth floor of an office building in Kami-Osaki, Shinagawa, and as of this writing the market values it at roughly ¥5.1–5.6 billion — call it $35 million, or approximately the price of a nice apartment building within walking distance of its own headquarters.

Here is the punchline, delivered early because it deserves to be: ItoKuro finished its October 2025 fiscal year with about ¥7.3 billion in cash and short-term investments. Its total liabilities — all of them, every account payable, every accrued bonus, every yen the company owes to anyone on earth — came to about ¥0.7 billion. Its borrowings are, to the nearest rounding error, zero: ¥9 million of debt against ¥9.9 billion of total assets, an equity ratio above 90% that has not dipped meaningfully in years. Subtract everything the company owes from the cash it holds and you get roughly ¥6.6 billion.

The market capitalization, again, is about ¥5.5 billion.

You are being paid, in the market’s own arithmetic, roughly a billion yen to take the business. The websites, the brand, nearly two decades of accumulated parent reviews, the employees, the office furniture — the market has appraised all of it at less than zero. Graham had a word for this; he called it a bargain. The modern screener calls it negative enterprise value, which is less poetic but means the same thing: the operating company is, officially, worth less than an empty box.

I should note that I run a screener covering nearly four thousand Japanese equities, and even in this market — where more than 1,200 companies trade below tangible book — a negative enterprise value attached to a profitable business is rare. Plenty of companies trade below their cash because they are incinerating it. ItoKuro generated about ¥800 million of free cash flow last fiscal year. The drawer refills itself.

Read more

In 1909, a man set up a workshop in Nakanomachi village on the Hamana plain near the Tenryu river mouth. His business was repairing circular saws. The sawmills along the river used heavy steel circular saws to break down the logs into lumber, and the saws got dull and broke teeth and warped under the strain of cutting all day. The shop fixed them. He called it Tenryu Tekko Goshi Gaisha — Tenryu Ironworks.

In 1910 he moved it closer to the river itself, to a village called Wada in Hamana District. In October 1913 the partnership was reorganized as a joint stock company under the name it still uses today.

Every circular saw used in Japan in 1913 was imported. The blades came from Britain and from a handful of other Western manufacturers. Japan had no domestic capacity to make a precision steel cutting tool of that size. Then the First World War broke out. Imports from Europe disrupted, then stopped. The Japanese sawmills could not get parts.

In April 1920, this company did the only thing it could think to do. It sent two of its engineers to England, to study the British saw-making process. They lived there for some time, learned what they could, and came back. Two years later, in 1922, they produced the first fully domestic Japanese circular saw, the Star Mark. In 1930 the Ministry of Commerce and Industry designated their woodworking saw an Excellent Domestic Product. The company has been making saws in central Japan ever since.

Today the head office is no longer next to the Tenryu River. In 1984 the company moved fourteen kilometers east, to a flat industrial site in the small city of Fukuroi, in Shizuoka prefecture. The address is 3711 Asaba. There is a manufacturing complex, an R&D building, a head office, and a parking lot. From there the company runs a global business with 970 employees, factories in Japan and China, sales offices in Thailand and the United States and Germany, and a deep product catalog of saw blades and cutting tools for wood, metal, stone, ceramic, cement, paper, food, and pretty much every industrial material humans cut. The current brand is called Pass Mark.

The company is one hundred and seventeen years old. The current CEO, Takaaki Ohishi, is its president. The stock trades on the TSE Standard market at ¥2,510 a share, giving the entire enterprise a market value of about ¥22.4 billion.

This is less than the financial assets sitting on its balance sheet.

THE NUMBERS

• Price: ¥2,510

• Market cap: ¥22.4 billion

• Cash and deposits: ¥9.1 billion

• Short-term securities: ¥3.1 billion

• Long-term investment securities: ¥12.4 billion

• Interest-bearing debt: zero

• Total equity: ¥39.1 billion

• Book value per share: ¥4,377

• Price to book: 0.57x

• Revenue (FY3/26): ¥13.5 billion (+2.6%)

• Operating profit: ¥1.74 billion (-5.0%)

• Ordinary profit: ¥2.19 billion (+4.3%)

• Net income: ¥1.52 billion (+0.7%)

• EPS: ¥168.94

• P/E: 14.9x

• ROE: 4.0%

• Dividend (FY3/26): ¥85 (50.3% payout, 2.0% DOE)

• Dividend yield at ¥2,510: 3.4%

• FY3/27 dividend guidance: ¥83 (3.3%)

Read more

It is a castle town on the Sea of Japan coast, the old capital of the Maeda clan, three hundred years of accumulated wealth poured into gardens and lacquerware and gold leaf and tea ceremony. Walk through Kenrokuen at dawn in November and the maple leaves look almost designed. The geisha quarters at Higashi Chaya and Nishi Chaya are still working districts where the lanterns come on at dusk. The Ohi family has been making the same style of pottery in the same workshop since 1666.

The B-29s never came. Kanazawa was on the American firebombing list in 1945 but the war ended before its turn. Most of the wooden city is still standing. There are streets in central Kanazawa where the buildings look essentially as they did when Imamura Naoji opened his trading shop in the spring of 1921.

For a long time it was a quiet provincial city, prosperous in the way old castle towns can be, off the main drag of postwar Japan. Then in March 2015 the Hokuriku Shinkansen extension arrived. Tokyo is now two and a half hours away. The tourists came. The hotels filled. The kaiseki restaurants started taking reservations a month out. Money has been flowing into Kanazawa in a way it had not for decades.

And in the middle of all of this, on a narrow street called Jikkenmachi in the old commercial heart of the city, sits a low office building behind a stone facade. Building number 25.

The company started life on March 19, 1921, as Imamura Naoji Shoten, named after its founder. It was a small trading shop, one of dozens in Kanazawa back then. What separated it from the others is that it survived. Most did not.

In July 1944, with Japan deep into the war and the economy being reorganized for total mobilization, the government issued the Industrial Consolidation Order. Naoji Shoten was forced to merge with three other small trading houses, Shitaka Kichizo Shoten, Fujii Sotoji Shoten, and Kojima Kishiroh Shoten, into a single new entity. The wartime authorities did this all over Japan. Most of the resulting companies did not outlast the postwar collapse. This one did. In October 1948, with the new Securities and Exchange Law in place, Imamura registered as a securities broker and started the business it still runs today.

The company has been in continuous operation in Kanazawa for 105 years.

THE NUMBERS

• Price: ¥1,330

• Market cap: ¥6.8 billion

• Cash and deposits: ¥8.9 billion

• Corporate borrowings: zero at year-end

• Total equity: ¥13.2 billion

• Book value per share: ¥2,576.79

• Price to book: 0.52x

• Operating revenue (FY3/26): ¥4.91 billion (+17.4%)

• Operating profit: ¥1.41 billion (+42.7%)

• Operating margin: 28.6%

• Net income: ¥1.06 billion (+38.8%)

• EPS: ¥206.29

• P/E: 6.4x

• ROE: 8.4%

• Dividend: ¥73 (5.49% yield, 35.4% payout, 3.0% DOE)

• Assets under custody: ¥4.329 trillion

• Capital adequacy ratio: 675% (regulatory minimum 120%)

• Employees: 191

• Branches: 11 across Ishikawa, Fukui, and Toyama

Read more

This is the least glamorous business imaginable. There is no technology story, no brand, no pricing power to speak of. It is a distribution business, which means it lives on a thin slice of gross margin and survives on volume, logistics, and relationships built over decades. The core operating company, Japan Kenzai, has been doing this since 1937. JK Holdings ranks at or near the top of the Japanese building-materials distribution industry by sales every year, trading the top spot with names like Watanabe Pipe and Nice.

For the year that just ended in March 2026, JK did ¥398.8 billion in revenue. The whole company is worth about ¥35 billion. You read that correctly. The market values this business at less than a tenth of its annual sales, and as we’ll get to, at less than two times the cash earnings of the operating business once you back out the cash on the balance sheet.

Let’s walk through why.

Read more

Tohoku Electric Power serves 7.6 million individual and corporate customers across six prefectures in the Tohoku region plus Niigata Prefecture.

It’s the fourth-largest electric utility in Japan by revenue, behind TEPCO, KEPCO, and Chubu Electric Power.

The Tohoku region is beautiful and it’s emptying out.

Population decline in the Tohoku region is among the most severe in Japan, with Akita losing 1.52% per year, Aomori 1.35%, and Yamagata 1.23%.

In 2021, the region had a population of 8.52 million, accounting for just 7.2% of the national total.

The productive-age population is projected to shrink by 31% by 2040.

That’s the kind of demographic headwind that keeps most investors away. And it’s why you can buy over ¥1 trillion of equity for about ¥506 billion.

THE NUMBERS

Read more

I do this because somewhere in the catacombs of the Tokyo Stock Exchange there are around three thousand companies that haven’t been thought about by anyone outside of their hometown in roughly forty years. Some of them are dead in the water. Some of them are quietly making money. Some of them are sitting on more cash than their entire stock-market value, while still going about their daily business of stamping out floor mats or grinding lenses or canning pickled plums or whatever it is they do. They are the corporate equivalent of those old peeling-paint houses out on the back roads where you wonder whether anyone lives in there at all, and then one day you see the porch light on and an old lady in a housecoat shuffling out for the newspaper, and you realize: oh. She’s been here the whole time. She just doesn’t make a lot of noise.

Today I want to tell you about one of these companies. Its name is Eidai Kako. It trades under the number 7877 on what the Japanese now call the “Standard” market — basically the second tier, where the small fry live. It makes plastic floor mats for Hondas and Suzukis, and a kitchen-sink assortment of other extruded plastic odds and ends. It is seventy-six years old. Average daily trading volume is about a thousand shares. I would estimate the chances that you have ever heard of it at approximately zero.

Bear with me. There’s a person in this story.

I.

When you pull up Eidai Kako on a stock screener — and I won’t pretend the kind of person who pulls things up on stock screeners isn’t its own breed of strange — what you see is the kind of profile that makes value investors lick their chops. The company is worth about three billion yen on the market, which is roughly nineteen million American dollars. It has about two and a half billion yen in cash sitting in the bank, and around one billion in debt. So almost half of what you’re paying for the company, if you buy a share, is just the cash itself. The rest of it is the actual business — the factories, the machines, the customer relationships, the seventy-six years of know-how about how to push molten plastic through a die without it coming out lumpy.

And what is the actual business doing? It’s making money. Not a lot — about a hundred and eighty-five million yen of net income last year, which is around a million two hundred thousand U.S. dollars. But it’s making it every year, more or less, with one exception we’ll get to. It pays a dividend that yields about three and a half percent. It just announced they’re raising it. The stock trades for about one-third of the book value of the underlying assets. Things, on paper, look very nice.

Things look so nice, in fact, that the natural question is: what’s wrong with it? Because nothing trades at a third of book in a country where the entire stock market has been rallying for three years unless something is wrong, or appears wrong, or used to be wrong, or the people who would normally buy it have all moved on to something flashier.

I have spent a lot of hours over the last few weeks reading their annual reports, which in Japan are called yūkashōken hōkokusho, and which run to about ninety-six pages of solid pharmacist’s-handwriting Japanese, dense as a brick of frozen tofu. I’m going to tell you what I found, but I’m going to tell it the long way, because the short way doesn’t do it justice.

Read more

In 1988, the company was targeted by a stock manipulation group called Koshin, led by a man named Mitsuhiro Kotani. Kotani built up roughly a 20% stake using borrowed money, became the largest shareholder, demanded the company buy his shares at inflated prices, and threatened to sell the position to organized crime if management refused. Through a series of round-trip financings between the comapny’s banks and Koshin-affiliated entities, the company paid out approximately ¥30 billion in extortion proceeds and took on ¥187.5 billion of Kotani’s debts as guarantor. Kotani was eventually convicted of extortion in 2003. Shareholders sued the former executives for ¥93.9 billion in damages, and the Supreme Court in 2008 affirmed personal liability of ¥58.4 billion against five former directors. The case is taught in Japanese corporate law classes as the leading example of how not to handle a stock raider.

In 1997, the company’s second main business — a 24-hour bath circulation product called Yumeijin — was implicated in a series of Legionnaires’ disease outbreaks. The business never recovered. The product was discontinued in 2022, by which point the market had shrunk by 97% from its peak.

In 2009, after Lehman, revenue hit ¥35.7 billion, less than 40% of the 1990 peak. The company sold its Kyobashi headquarters building to pay down debt and moved corporate operations to the factory site in Hachioji.

In September 2022, the company terminated its entire 300-person door-to-door sales force via mass layoff notices, effectively exiting the consumer direct sales channel that had defined the sewing machine industry in Japan since the 1950s.

In October 2021, on the 100-year anniversary, the company dropped “Sewing Machine Industries” from its corporate name.

And throughout all of this, the company sat on a fortress balance sheet, paid token dividends, ignored shareholder returns, and treated capital allocation as something that happened to other companies. Return on equity was negative in FY3/23 and barely above 5% in the years since. The stock has traded below tangible book value continuously for over a decade.

Then, in November 2024, an investor called MM Investments crossed the 5% threshold. Then crossed 7% two weeks later. Then 8% in March 2025. Then 9% in mid-2025. Then 10.34% in November 2025. And on the same November 14, 2025 day that this 10% disclosure was filed, the company — which had literally spent a century not doing this — announced a ¥1.5 billion buyback program for up to 10.96% of its outstanding shares.

For the first time in its 105-year history, it is being forced to behave like a public company.

But it still trades at 0.5x tangible book and about 3x EV/EBITDA.

Read more

Sansha Electric Manufacturing, ticker 6882 on the Tokyo Standard exchange, is a stock I have spent the last week working through in detail, and it is a useful case study in the disposition I am describing, because every surface fact about it argues for purchase and almost every substantive fact about it argues against. The shares trade at approximately ¥1,200, having closed at ¥1,099 on the day of the FY2026 results announcement on May 8 and having bounced fourteen and three quarters percent to ¥1,261 by the close of business on the eleventh. The market capitalization at the midpoint is approximately ¥17 billion, or roughly $108 million at the prevailing exchange rate of ¥157.6 to the dollar. The book value per share, as of the third-quarter balance sheet dated December 31, 2025, is ¥1,834. The price-to-book is therefore 0.65x. The company carries ¥6.04 billion of cash against ¥2.20 billion of short-term debt, having reduced the debt by ¥800 million from the ¥3.00 billion peak reached during fiscal 2025. The net cash position is ¥3.84 billion, or approximately ¥289 per share, which is roughly twenty-two percent of the market capitalization. The current ratio, twenty-four point seven eight billion of current assets against nine point zero four billion of current liabilities, is two point seven four. The equity ratio is seventy-one point seven percent. The company has paid a dividend every year for over a decade. The current rate of ¥40 produces a yield of three point three percent against the midpoint price. There is a tangentially related silicon carbide product line, a market that the entire power electronics industry has spent the better part of a decade insisting will be one of the secular growth stories of the energy transition. The company was founded in 1933 and is one of the older continuously operating power electronics manufacturers in Japan. It holds the number-one market share position in domestic plating power supplies for surface treatment, a niche but real industrial vertical. A reasonable Graham-tradition screener could be forgiven for stopping at any of these facts and deciding to add the position to a basket. A screener that does so, in this case, will be wrong, and the way in which it will be wrong is instructive for the rest of the deep value universe, because the failure mode in Sansha is the failure mode in roughly one out of every three Japanese small-caps that screen this way, and the discipline of recognizing the failure mode in advance is the part of the craft that compounds.

Read more

Then Marelli blew up. Civil rehabilitation in July 2022. A Delaware Chapter 11 in June 2025. KKR is gone, the senior lenders are taking the company over, and somewhere in the chaos this company quietly stopped being a Marelli subsidiary. The deposits all got returned. And the operating P&L — freed from forty years of intra-group exploitation — started doing what it had always been capable of doing.

In FY3/22, the last full year as a sub, this company earned ¥106 million on ¥27 billion of revenue. A 0.4% net margin. Four years later, it’s now guiding to ¥1.85 billion of net income on ¥34.8 billion of revenue — a 17.5x improvement in net profit on 29% more revenue. Operating margin has gone from sub-2% to 12.5%. Through nine months of FY3/26, operating profit is up 57.7% year-over-year. The dividend just got raised 30%. The share count is shrinking through actual cancellations, not just buyback-and-park.

And the stock trades at 1.2x EV/EBITDA.

Read more

It wasn’t a funeral hall. It was an office building, a relic from a different era of Japanese corporate strategy when conglomerates accumulated trophy real estate as a hedge against everything from inflation to land reform. The funeral company had owned it for decades. It sat on the books for a fraction of its market value, generating modest rental income, doing nothing strategic.

In late 2024, they sold it. The proceeds: about ¥3.8 billion in cash. The accounting gain: ¥3.4 billion. The implied book-to-market markup: roughly 9-10x.

And here’s the thing that should grab a value investor’s attention: this was not their main asset. This was a non-core building they happened to own. The company’s real assets — the ones it actually uses every day to run a business that has been profitable for 93 consecutive years — are still on the balance sheet. At cost. Marked at numbers that, judging by the Kitahama transaction, may understate true market value by a factor of several.

The company is SAN Holdings (TSE: 9628). It is the largest publicly listed funeral operator in Japan. Its core subsidiary is one of the most profitable hospitality operations I have ever seen disclosed in a public filing. And as I write this, you can buy the entire business — including all the embedded real estate — for less than its stated book value, at roughly 12-14x normalized earnings, with a 3% dividend yield and a progressive dividend policy.

There are reasons it’s cheap. Some of them are good reasons. Some of them are misleading. The job of this article is to walk you through both, so you can decide whether the asymmetry is what it appears to be.

This is going to be long. The story has too many layers — a 93-year operating history, a recent transformative acquisition, a brand-new second acquisition that just closed, a real estate portfolio that almost nobody is paying attention to, and a Japanese demographic backdrop that nobody knows quite how to price — to compress meaningfully.

A standard disclaimer up front: I am not a financial advisor, and none of what follows is investment advice. It is a structured walk-through of public disclosures with my interpretation of what they mean, intended to help engaged readers do their own work. With that out of the way, let’s start with the headline numbers.

The trade

Read more

We thought it would be useful to pick ten of them, lock in entry prices today, and follow the basket in public for the next twelve months.

Most of them are slow-growth, structurally unloved small-caps that the global investment community gave up on a decade ago. Sewing machines. Plywood. Asphalt. Industrial paint distribution. Kitchen cabinets. The kind of names where a sell-side analyst would lose his job for writing a research report. There is no AI angle, no GLP-1 angle, no nuclear renaissance angle. Just net cash, tangible assets, and dividends.

Methodology

The TDV 10 Index is a rules-based portfolio drawn from the Tokyo Deep Value screener universe. The criteria are deliberately strict:

• P/TBV below 0.6x, material discount to tangible book

• EV/EBITDA below 6x, cheap on cash earnings

• Positive LTM EBITDA, operationally profitable on a cash basis

• More cash than debt, net cash balance sheet

• Market cap above ¥5 billion, minimum liquidity floor

• Excludes financials, REITs, and businesses with captive finance arms

The surviving names are force-ranked by trailing dividend yield, with the top ten constituting the index. Equal-weighted at inception. Held without rebalancing for twelve months.

The aggregate stats are striking. The equal-weighted basket trades at roughly 0.6x tangible book on a weighted average basis, in the low single digits on EV/EBITDA, with dividend yields clustering around 4%. Every name has more cash than debt. None are followed meaningfully by sell-side analysts.

We’ll be tracking this basket of names closely over the next 12 months, with periodic updates.

The Ten

Entry prices below reflect approximate recent market levels (early 2026). Specific share prices and market caps fluctuate; the basket will be locked in at the close on inception date.

1. Sansha Electric Manufacturing (6882.T) — ~¥890, ~¥11.9B mkt cap, ~6% yield

Osaka semiconductor and power supply maker, founded 1933, serving industrial markets nobody pays attention to. The product list reads like an electrical engineer’s stockroom: TRIACs, thyristors, rectifier diodes, IGBTs, arc welders, dental equipment, dimmer switches. I do not know what any of these things actually do.

Buried in the catalog is a SiC MOSFET line. Same power devices going into EV inverters, solar inverters, and grid storage. The company is too small to matter in that conversation today. The SiC ramp could change that, or it could pass them by entirely.

This is one of the lower-quality names on the list by operating metrics. Recent ROE has been thin and trailing FCF has been weak. The book value, however, is real, and net cash is meaningful relative to the market cap. The dividend has a long runway because the cash position is large. You collect the yield while either the SiC business shows up in numbers or someone larger takes them out.

2. Janome Corporation (6445.T) — ~¥1,180, ~¥21B mkt cap, ~4% yield

The household-name sewing machine business diversified into industrial servo presses, Cartesian robots, screw presenters, and an IT services arm. Nobody outside Japan pays attention to it. Sewing machines globally have been written off as a slow-decline business for thirty years.

The interesting development is that the activists have shown up. MM Investments raised its stake from 9.34% to 10.34% in November 2025. On November 14, 2025, the company authorized a buyback program for up to 1,960,000 shares, representing 10.96% of outstanding (excluding treasury shares), capped at ¥1.5 billion, running through November 16, 2026. They have been executing monthly. By the end of February 2026, cumulative repurchases were 812,300 shares for approximately ¥1.03 billion, leaving room for roughly ¥470 million more before expiry.

A 10.96% buyback authorization is large by any measure. For a stock at well below tangible book with substantial net cash, it is enormous. ROE is low but the balance sheet is swimming in working capital. The bet is on activist pressure, continued buybacks, and the eventual emergence of either a private equity carve-out of the industrial automation business or a take-private of the whole thing.

3. Kawata Manufacturing (6292.T) — ~¥800, ~¥5.8B mkt cap, ~5% yield

Founded in Osaka in 1935. Makes the auxiliary equipment around plastic injection molding presses: autoloaders, dryers, mixers, mold temperature controllers, granulators. Customers are anyone running an injection molder, which means auto parts, packaging, electronics, medical devices.

Cyclical capital equipment business. Sells in waves. Right now the market is treating it like a melting ice cube. Trades at a steep discount to tangible book, very low EV/EBITDA, smallest market cap on this list. Trailing FCF has been strong, partly working capital recovery from a cycle bottom, but even normalizing for that, the company generates cash.

Liquidity is the obvious risk; this is a name where you size correctly and wait for either a cycle turn or a takeout. Conair Group in the U.S. is its long-time technical partner and the most logical strategic buyer.

Read more

Not at half book. Not at a third. At roughly twenty-one cents on the dollar of stated net asset value, against a balance sheet that is over a trillion yen of mostly Japanese small-business loans, mostly Japanese government and corporate bonds, and a recently-cleaned-up capital structure for the first time in twenty years.

It is the cheapest listed bank in Japan. There are roughly a hundred listed banks in the country, and none of them trade as cheaply as this one. The next cheapest sits about forty percent more expensive. The median tier-two regional bank trades at nearly double this multiple. The cohort it sits in does not exist — it is a sample size of one.

That alone would be interesting. What makes it actionable is the rest of the setup.

A known investor is accumulating

The largest single shareholder of the bank is a holding company controlled by a man who has spent over thirty years buying small Japanese companies that trade below book and pushing their managements to act like they care about shareholders. He has run hostile takeover bids. He has proposed dismissing CEOs at annual general meetings. He runs a network of more than fifty companies under one umbrella, with a long career of taking concentrated stakes in cheap, asset-rich, weakly-managed Japanese small-caps.

Over the last three years he has built his position in this bank from under three percent to over fifteen percent. Each of his most recent four large-shareholder filings has reported a higher percentage than the one before. He has spent over a billion yen of his vehicle’s net assets — more than ten percent of the entire net asset value of the holding company he uses for this position — on a single bank’s common stock. He is not flipping. He is not hedged. He is buying with conviction, against a multi-decade career of doing exactly this kind of trade.

Crucially, he has just crossed a regulatory threshold under the Japanese Banking Act that requires regulatory permission for any further large stake increases. He cannot quietly raise his position to thirty or forty percent without applying for and receiving permission from the financial regulator. So the path is constrained: he either holds at his current level and waits for management to act, or he applies to go further and crosses into the territory where the bank itself becomes a meaningful chunk of his economic exposure.

The stated purpose on his filings is the standard Japanese formulation that keeps optionality open between passive financial holding and strategic action. His track record at his other portfolio companies says the optionality is real.

The CEO just changed

In the middle of last year, this bank installed a new president whose CV is unusually international for a regional Japanese bank. She started her career at a major American bank in Tokyo working on derivatives and capital markets. Then a stint at a venture think-tank she helped found. Then nearly two decades at the Japanese central bank, with senior overseas postings in Hong Kong and London. Then several years running a digital design and tech R&D subsidiary of one of Japan’s three megabank groups. Then crossed back into commercial banking via this name as deputy president, and was promoted to the top job a year later.

She is also notable for being the first woman ever to run this bank — and one of a very small number of women running any Japanese regional bank. Most regional Japanese bank presidents are lifers — recruited out of university, promoted up the same hierarchy, ascending to the top job in their late fifties without ever working outside the firm. This is not that.

Nobody installs a credentialed external operator at the top of a sleepy regional bank in order to maintain the status quo. The combination — the investor on the share register, the recapitalised balance sheet, the new external CEO — is not coincidental.

The numbers are absurd

The bank earns around a billion yen of net income in a typical year, against book equity of roughly fifty billion yen and total assets above a trillion yen. The return on equity has bounced between one and two percent for most of the past decade. The cost-to-income ratio has been deteriorating, not improving. The net interest margin — the spread the bank earns between what it pays for deposits and what it earns on loans, after expenses — has compressed to under ten basis points. A blade of grass.

This is not a great bank. To be clear, it is not even a good bank. It is a structurally low-return, slowly-shrinking franchise in a structurally low-growth Japanese prefecture, where the population is falling faster than the national average and the working-age cohort is in steeper decline.

And yet.

The bank has just emerged from a twenty-year shadow that no longer exists on the balance sheet. The Japanese government’s bad-bank vehicle, which had held over forty percent of this bank’s stock as the rump of an old post-financial-crisis recapitalisation, was completely bought out and exited two years ago. The clean-up of that legacy ate into book equity at the time. The book equity has since stabilised. The new president has inherited the cleanest balance sheet the bank has had since the late 1990s.

There is one preferred-stock instrument left to redeem, scheduled by management for redemption within the next several years. After that, the capital structure is fully normalised, and any meaningful return of capital to common shareholders becomes available for the first time since the recapitalisation era began.

The bond portfolio is currently sitting on a substantial unrealised loss because Japanese yen interest rates have risen and the bank holds a large book of corporate and government bonds at roughly five-year duration. That mark moves with rates. If yields retrace even modestly — and the Bank of Japan’s policy trajectory is no longer one-directional — that mark unwinds, and book value pops mechanically upward without management lifting a finger.

What it could be worth

The medium-term plan that the bank has published targets a roughly threefold increase in core operating profit and a near-quadrupling of net income by the end of the decade. Management has been candid that closing the gap between their cost of equity and their actual return on equity will take roughly ten years.

Set the plan aside. The numerical exercise is straightforward.

If the bank closes only half of its discount to its tier-two peer set — meaning it re-rates from one-fifth of book to a little under one-half of book, still cheap by any global standard — the share price doubles. That is not a heroic assumption. The peer-group median trades there today.

If the largest shareholder forces a corporate event — a friendly transaction, a tender offer, a strategic combination — the floor under the share price becomes liquidation-style book value, which is several multiples of the current price. Banking-sector consolidation is happening across Japan; tier-two regionals with cleaned-up capital structures and no government overhang are exactly the kinds of names that get picked up.

If both happen — re-rating plus event — the upside is multiples of the current price.

Against this, the bear case is that the underlying franchise economics are weak enough that the bank simply continues to grind at one-fifth of book for years, while the investor gets bored and the CEO’s plan misses its targets. In that scenario you collect a two-and-a-half percent dividend, you wait, and you take a small loss in real terms over a five-year holding period.

The asymmetry is what makes the trade interesting. You are buying a balance sheet at a fraction of stated value, supported by tangible loans against real Japanese collateral, while a known investor is buying alongside you, while a credentialed CEO is running the operating turnaround, in a sector where consolidation is the explicit policy stance of the Japanese government.

This is a real trade. It is not without risk. But it is the kind of setup that rarely shows up in liquid markets, and almost never with this many moving pieces aligning at once.

* * *

In the paid version of this post I name the bank and the ticker, walk through the specific shareholder dynamics including who the investor is and what his other portfolio companies look like, lay out the bond stress test that shows the equity is more rate-resilient than the headline mark suggests, name the new CEO and walk through her track record, do the peer-comparison table that establishes the cheapness ranking, and put numbers around what the upside scenarios actually look like at specific share-price targets.

If you want to actually trade this, that is what you need.

As always, I found this company using the screener I built on Tokyo Deep Value. I screened for the lowest price to book I could find and saw that there was a very cheap bank on this list. After screening I did the analysis. Check out the screener if you want to find cheap stocks too.

Read more

I came to it the way one comes to most things worth finding: through someone else’s enthusiasm. A colleague had passed along a write-up framing the company as a deep-value situation — fifty-four cents on the dollar against book equity, a four-and-some-percent dividend yield, a fortress balance sheet, and a near-term catalyst in the form of an FDA decision on a chemonucleolytic drug for lumbar disc herniation. The kind of setup that, on a single page, makes the heart move faster than the head. I read it twice and decided to look closer.

What follows is what I found after reading the last five annual reports, the regulatory filings, the partner press releases, the trade-press coverage, and the Japanese-language pharmacovigilance notices that nobody had bothered to translate. The thesis survives, in altered form. But the picture is more interesting, and harder, than the page suggested.

Read more