RINGER HUT Co., Ltd. (8200): A Real Franchise, a Borrowed Record, and No Price

Verdicts

Lens	Verdict	Buy below	Most load-bearing items
Buffett	watch	¥600	B42 owner-earnings yield ~2% — strip the one-off DTA F296 and the business earns ~¥1.1–1.3bn against a ~¥60bn cap D2; B26/B60 pre-tax return on tangible capital ~7% F279F261D12, adequate not wonderful; B46 the "+78% record" D6 is a tax artifact over a −16.3% operating decline D5
Munger	pass	null	M88/M44 no margin of safety / pari-mutuel — ~50× forward on the guided ¥1.2bn F163D15, a 3.15% operating-margin chain D4 at ~4× book D1; M40 raisins-and-turds — tonkatsu ¥141M profit, −51.5% F351, Hamakatsu in 債務超過 F133; M37 great-at-a-fair-price, not fair-at-a-great-price
Pabrai	pass	null	P1 max permanent loss — 3.96× book D1, net debt D12, only ~¥7bn of hard resale value vs ~¥60.4bn cap D2; no floor; P53/P50 fifty-cent dollar fails — ~35× tax-flattered / ~50× guided D15F163; P13 the market fears nothing — low-uncertainty, priced for perfection
Li Lu	watch	¥900	L18 dollar-at-fifty-cents fails — conservative owner value ~¥640/sh (15× normalized ~¥1.1bn F293F327) vs ¥2,330; L35/L4 non-compounding value behind a tax mirage (NI +78% D6 / OP −16% D5); price is +3.6% not traffic D11; L24 payout 20.7% vs the firm's own 30% benchmark F116E99
Claude	pass	null (implied ¥430)	C38 earning power below asset value — EPV(equity) ~¥6.8bn < book ¥15.2bn F261, so ~¥60bn D2 pays ~9× no-growth earning power; C35/H1+H2 normalized owner-yield ~2.6% (record is a ¥597M tax event F296 over a −16.3% line D5, guided down F163)

No lens buys at the ¥2,330 stamp, and — unlike the last quality name on the record — every buy-below price named sits 74–82% below it (Buffett ¥600, Li Lu ¥900, Claude's implied floor ¥430). The panel does not dispute the brand; it disputes the price, unanimously and by a wide margin.

The business

RINGER HUT (リンガーハット) cooks and sells two things: a big bowl of Nagasaki champon — wheat noodles piled with stir-fried vegetables — under the RINGER HUT brand, and breaded pork cutlets (tonkatsu) under the Hamakatsu name E4E5E14E15. It does this from 641 stores, 627 of them in Japan and 14 overseas F154F152F153, and it grows and makes much of its own product — noodles, gyoza, fried rice — in its own factories, buying 100% domestic vegetables under contract farming as a brand promise since 2009 E78E34E35E45. A shopkeeper would understand the whole thing in a sentence: it is a regional noodle chain that has sold essentially the same bowl since 1974 E5, a bit dearer each year, to roughly the same number of people.

The group reports three segments, and the concentration is the first fact to hold. Champon is the franchise: 81.8% of revenue D7 and 76.2% of segment profit D10, across 556 stores F172. Tonkatsu is the weaker second brand — 85 stores F178, 17.8% of revenue D8 but only ~9% of segment profit D10, and its profit fell −51.5% this year F351F175. A tiny facility-maintenance arm (¥195,200千 of external revenue F346) earns an outsized margin but cannot move the whole. The network is mostly directly operated — 488 of the stores, with 153 franchised E29E83 — and management wants to push franchising toward ~30% of stores to shift capex to franchisees E33F149. The store base actually shrank net −5 this year (7 opened, 12 closed F156); the "overseas engine" is one first directly-managed store in Vietnam E56 on ¥9.5M of equipment E84 — optionality, not yet a business.

Two structural facts sit under the tidy picture. Both mainstay operating subsidiaries are in 債務超過 (negative net worth) — RINGER HUT Japan at −¥41,774千 and Hamakatsu at −¥261,374千 F128F133E19 — sustained by the parent's intra-group loans and a large affiliate doubtful-account allowance F430, even as the consolidated group looks healthy. And the company's real estate is leased, with the operating leases off the balance sheet: the new right-of-use standard is 未適用 (not yet applied) E128, so only ¥203,758千 of finance-lease liability is booked F235F243 against ¥4,347,737千 of annual rent running through SG&A F274. The register is foundation-and-relationship: the founding Yonezu family's wealth sits in the 米濵・リンガーハット財団 (its two vehicles hold ~4.6% F393F394), alongside trust banks and reciprocal holders like Asahi and Kirin E90E91, with no Yonezu-family individual among the top holders E92.

The numbers

FY2026/2 was, on the marquee, a record: net income attributable to owners of ¥1,727,752千, up +78.4% F15D6 — the best in the company's history, on revenue that grew +2.9% to ¥45,084,681千 (the five-year-table basis, which includes ¥819,558千 of その他営業収入; the P&L revenue line is ¥44,265,122千) F5. ROE was 12.0%, the best in five years F51; the equity ratio was 48.9% F46; book value per share ¥587.75 F35.

Do not let the word "record" do any work. Operating profit — what the restaurants actually earned selling food — FELL −16.3%, to ¥1,418,176千 from ¥1,694,051千 F279F280D5, as SG&A rose ~¥1,035M on labour and cashless-payment fees E74. Ordinary profit was essentially flat at ¥1,598,313千 (+1.0%) F10. The entire gap between the +78% bottom line and the −16% operating line is one item: a one-off deferred-tax benefit of ¥597,166千 (法人税等調整額), the recognition of deferred-tax assets on COVID-era loss carryforwards, which turned the whole tax line into a net credit of ¥249,229千 — an effective tax rate of −16.9% F296F298F328E75. Management concede the point directly: they guide FY2027 net income DOWN to ¥1.2bn (against forecast revenue ¥47,300百万 and operating profit ¥2,200百万) precisely because the benefit does not repeat F163F164F165E62.

At the ¥2,330 stamp the shares are 3.96× book D1 and ~34.9× earnings D15 on reported EPS ¥66.67 F40; normalized for the tax gift, the multiple is nearer ~59×, and on the guided ¥1.2bn it is roughly 50× forward F163. Market capitalization is ¥60.4bn on the book-consistent float D2 (¥60.73bn on the legal issued-less-treasury float D3). Same-store sales rose +3.3% F150 — but same-store customers fell −0.3% F151: every yen of the gain, and then some, came from per-customer spend (+3.6%), not traffic D11, off three menu-price revisions in two years E63. The balance sheet carries net debt of ¥5,574,078千 D12 — not net cash — with long-term borrowings termed out to 2027–2030 at an average 0.378% F334E174; a conservative finance policy E49, but the "cushion" is an equity ratio, not an asset floor. Store impairment took ¥74,950千 across 9 stores F289; the sole listed cross-holdings are ¥1,294,057千 (six names, a ¥994,006千 unrealized gain inside) F418E173 — a rounding error against the cap. The dividend is DPS ¥13.00 (¥6 interim F404 + ¥7 final F405), a 20.7% payout F116 against the company's own 30%-consolidated benchmark E99F407; there is no buyback E98, and the share count has been flat at 26,067,972 for five years F95. Over that window the total shareholder return was 99% against the 配当込みTOPIX's 238% F118F119. Governance is orderly: a 6-director board, 3 outside F408; a clean five-year audit from Grant Thornton Taiyo whose sole Key Audit Matter is store fixed-asset impairment — store PP&E is 32.6% of total assets — with no separate KAM for the large deferred-tax benefit E181E184E153; related-party transactions nil E169; no controlling parent E178; directors' pay a modest ¥109M, ¥20M of it tied to the ordinary-profit margin F415F416E130.

The five lenses

Buffett — watch, buy below ¥600

Let me tell you what this company is before I say a word about the stock, because that is the only honest order. Ringer Hut cooks two things — a big bowl of champon and breaded pork cutlets E4E5E14 — out of 641 stores F154, and grows its own noodles and vegetables so the product tastes like nobody else's E78E34. Champon is four-fifths of revenue and three-quarters of profit D7D10; the tonkatsu side is smaller and had a rough year F351. This is squarely inside the circle — a plain restaurant chain, no technology I have to guess about — and I can estimate its earnings a decade out: a well-run noodle chain that has sold the same bowl since 1974 E5 will still be selling it, a bit dearer, to a few more people. So we don't stop here — and that is rare; most of what I look at, I stop.

Now the part where I earn my keep. The headline is a "record" net profit, up 78% F15D6. Do not let that word work. Operating profit — what the ovens earned — FELL 16% F279D5. The whole of that "record" is a one-time gift from the tax man: ¥597 million of deferred-tax assets booked on old COVID losses F296E75, and management then guide next year's profit DOWN to ¥1.2 billion because the gift does not come twice F163. Strip it and the business earned about ¥1.1 billion — which is roughly what they say it will earn next year. That is the true earning power, not the ¥1.7 billion on the marquee.

Is there a franchise underneath? Partly. Same-store sales rose 3.3%, but the customer count fell 0.3% F150F151 — every yen came from charging more, not from more people walking in D11. That is genuine pricing power, the thing I prize most — but pricing power with flat traffic is a narrower moat than pricing power with a growing line at the door. And here is the number that fixes the character of the business: on operating profit of ¥1.4 billion F279 against about ¥20.8 billion of equity-plus-net-debt F261D12, the pre-tax return on the capital the restaurants require is under 7%. That is not a wonderful business earning 20% on tangible capital that I'd pay up for [B26]; it is an adequate one that must keep pouring money into store fit-outs — capex ¥1.9 billion against depreciation ¥2.1 billion F311F306 — just to stand still. Nearly every yen the ovens throw off gets plowed back at that same middling return. That is the good savings account, not the great one.

So to price, and here I put my hat back on and walk out. Owner earnings — normalized profit ~¥1.1 billion, add back depreciation, subtract the capex it truly takes to hold the line — land near ¥1.3 billion. At ¥2,330 the whole company is priced at about ¥60 billion D2: an owner-earnings yield of a bit over 2%, on earnings flattered by a windfall, at 35 times this year and something like 50 times next year's guided profit D15F163. You are paying almost four times book D1 with no asset floor to catch you — current assets of ¥6.2 billion sit under total liabilities of ¥15.9 billion F207F251, so there is no net-net, no net cash, nothing but the going concern. The people are fine — conservatively financed F46, candid about the tax quirk E75 — which is why this is a watch and not a walk. But they retain most of the earnings at ~7% while the stock has trailed a simple index 99% to 238% over five years F118F119, and the register has no owner-operator eating the same cooking I would E92. I'd own this decent little noodle chain — but only near ¥600, where the owner yield finally clears about 8% and I'm paying roughly tangible book. At ¥2,330 I tip my cap to the champon and keep my wallet shut.

What a student should take from this: when a "record" profit sits on top of a falling operating line, find out why, and value the business on what the ovens earn — not on what the tax accountant handed you once. A good business at four times book D1 and a 2% owner yield D2 is still a pass; price is not a detail you add after you admire the moat, it is the whole decision.

Munger — pass

Invert first, as Jacob always said: how does this business die? A champon chain earning 82% of revenue D7 from one regional format, in a country with a shrinking population and flat foot traffic, faces a per-customer-spend strategy that runs out of runway once you've taken the price rise twice; a second format — tonkatsu — that cannot support itself; and a cost ratchet in labour and utilities that a 3% margin cannot outrun. And note the balance sheet is net debt ¥5.57 billion D12, with ¥4.35 billion of annual operating-lease rent running entirely off the page F274 because the ROU standard is not yet applied E128. The comfortable-looking 48.9% equity ratio F46 is comfortable only because those lease obligations are invisible.

Now what management handed the market: net income up 78% to a record F15, while operating profit fell 16.3% F279D5 — the record resting entirely on a ¥597 million one-off deferred-tax benefit F296 that the company itself tells you will not repeat, guiding FY2027 net income down to ¥1.2 billion F163. So the 35× multiple you pay D15 is on tax-flattered earnings, and the forward multiple on the guided number is about 50×. That is not a fair price for a low-single-digit-growth noodle chain.

Is there a moat? I'll give it a genuine, modest one: fifty years of the champon brand E5, all-domestic sourcing since 2009 E9, self-manufacture of noodles and gyoza E78, and the scale to advertise it — a challenger would need a decade and real capital to copy those supply relationships. Three price increases in 24 months without a volume collapse E63 is a pricing-power fingerprint. But look at what the moat produces: a 3.15% operating margin D4 against the company's own 10% target F147, missed for fifteen years and missed wider this year. A moat that only holds a 3% margin after half a century of brand-building is a habit franchise on a treadmill, not a durable edge. And the raisins-and-turds problem is acute: tonkatsu earns ¥141 million on ¥8 billion of revenue — a 1.8% margin, down 51.5% F351F344 — with Hamakatsu in standalone 債務超過 F133E19. At ~4× book D1 you buy the turd with the raisin.

The pari-mutuel check decides it. At ¥2,330 you pay ~¥60 billion D2 for a business earning ~¥1.4 billion of operating profit F279, plus ¥5.6 billion of net debt D12 and ¥4 billion a year of off-book rent. The quality is real — but the market has known these things for a decade; the stock has gone essentially nowhere against a market that doubled F118F119. This is not a bet at mispriced odds; it is the popular horse at 3-to-2 where the trainer has just told you earnings fall next year. Pass — not too-hard, because it is simple, and not a watch, because the terminal issue is price, not uncertainty. I'd revisit below ¥1,400, roughly 20× the guided normalized earnings; there is no falsifiable re-entry trigger to offer because there is no buy case here to falsify.

What a student should take from this: always bridge from operating profit to net income before you conclude — when the bridge needs a one-time item the size of the whole operating profit, and management themselves remove it next year, the headline is not earnings. And a moat with fifty years of history is not a reason to buy; it is a reason to understand what the price already contains.

Pabrai — pass

I run everything through one door: heads I win, tails I don't lose much. Before any thought of upside I want the realistic worst case per share and what fraction of ¥2,330 it destroys — so I look for the hard floor first: net cash, listed securities, land at conservative marks. On Ringer Hut, that floor is not there. Book is ¥587.75 F35 but the price is 3.96× book D1, and an equity cushion is not an asset floor. The group is net debt — ¥7.87 billion of interest-bearing debt including leases against ¥2.3 billion of cash, ¥5.57 billion net F331F76D12. The genuinely saleable assets are land at ¥5.81 billion F212 and listed cross-holdings at ¥1.29 billion E173 — maybe ¥7 billion against a ~¥60.4 billion market cap D2, an eighth. There is no fleet of ships to sell three at a time. Pay ~4× book for a restaurant chain and the whole case rests on the champon story staying true, with nothing catching you if it doesn't.

Now the earnings, because the warning is real. A record net income of ¥1,727,752千, +78.4% F15D6 — a mirage. Operating profit fell −16.3% F279D5; the gap is one line, a ¥597 million one-off deferred-tax benefit F296 that management guide away next year to ¥1.2 billion F163. So the ~35× multiple D15 is on tax-flattered earnings, and on the guided number it is roughly 50×. I buy fifty-cent dollars; this is a dollar-fifty for a dollar of thin, cyclically-soft operating earnings — operating margin 3.15% D4, a 10% target missed at 3.6% actual F147E134, same-store sales carried entirely by price while customers fell F150F151D11, tonkatsu profit down 51.5% F351. And I'm not paid to wait: a 0.56% dividend yield F111, no buyback E98, a ~1.7% free-cash-flow yield F310F311. I am paying a premium to wait while operating profit shrinks.

So name what the market is afraid of. Honestly — nothing. This is not distressed, not a cyclical trough, not too-small-and-ugly for institutions; it is a healthy, fully-priced compounder with no analyst fear to exploit. That is the tell. Dhandho hunts low-risk, high-uncertainty — where the crowd mistook a wide range of outcomes for permanent-loss risk. This is the opposite: low uncertainty, priced for perfection, with no floor. The asymmetry runs against me. Few bets, big bets — this does not clear the bar to be one of ten. Pass.

What a student should take from this: a record earnings headline is not earnings — read to the operating line and the tax line before you believe it. And at ~4× book with net debt, an "equity cushion" is not a floor: when there is nothing hard beneath the price, a good business becomes a bad bet, because the whole case rests on the story staying true and there is no margin of safety if it doesn't.

Li Lu — watch, buy below ¥900

Begin where I always begin, before any thought of price: could I predict this business ten years out? Honestly, yes — and that is worth saying, because it is what separates this from the puzzles I set aside as too-hard. The variables that decide 2036 are few and each is in the filings: whether same-store growth is volume or merely price, whether the FL-cost line holds against Japanese labour and food inflation E32E74, and whether the yen reinvested in stores earns a real return. So the knowledge bar is cleared [L1]. The discipline now is refusing to let a knowable business seduce me into paying a compounder's price for economics quietly going the wrong way.

Start with the number the market celebrates and I distrust: net income a record ¥1,727,752千, +78.4% F15D6 — while operating profit fell 16.3% F279D5. The entire gap is a one-off deferred-tax benefit of ¥597,166千 F296E75 that management's own FY2027 guidance concedes by guiding net income down F163. In the ledger of know / assume / pretend, "the company earned ¥1.7bn" is a pretend; the honest figure is the ~¥1.0–1.2bn the operations threw off, and stripping the pretend collapses the celebrated 12.0% ROE F51 back toward the ~7% it earned before the gift F47F50. Then the quality of the growth, which is the heart of it: same-store sales 103.3% but customers 99.7% F150F151 — the franchise defending margins with price, not pulling more people through the door, and the operating decline says the price did not even cover the cost inflation. The brand is real and the 100%-domestic sourcing is a genuine, defended differentiation E35E45 — value is not melting — but neither is it visibly compounding on the axis that matters, and the company's own 10% margin target F147 has stood unmet at ~3% for fifteen years. I credit stated ambition little and the trail of results much, and the trail says thin, competitive, hard-won restaurant economics.

The balance sheet deepens the caution: a net-debt business D12, with the true fixed-charge burden hidden off the page under the not-yet-applied lease standard — ¥4.35bn of rent through SG&A F274E128. On capital allocation, the test that defines management for me, the verdict is "careful, not shareholder-minded": shares dead flat for five years F95, no buyback at any price E98, and a payout of 20.7% against the company's own 30% benchmark F116E99 — cash retained by habit while incremental returns fall, the classic pattern I distrust. To the credit column: no listed parent E178, no family individual extracting through the register E92, related-party transactions genuinely nil E169, a clean five-year audit E180. The accounting I trust; the capital discipline toward me, the owner, I cannot yet.

So to price, and the dollar-at-fifty-cents standard. My conservative owner value normalizes the tax away — pre-tax income ¥1,478,522千 F293 at the 30.5% statutory rate F327 is ~¥1.0bn of real earnings — and fifteen times ~¥1.1bn is about ¥640 a share, barely above book F35. At ¥2,330 this is not a dollar at fifty cents; it is a dollar at three-and-a-half dollars [L18]. The unknowns are not the problem — I understand this business. The price is. This is a knowable, honestly-run, non-compounding business at an unpayable price — a watch, not a pass and never a buy here. I keep it on the list and write down the price at which the arithmetic works: I would begin to accumulate near ¥900, a modest premium to book, and only if the traffic turns.

What a student should take from this: understanding a business and being able to own it are different verdicts, and a "record" year is where the difference bites. The knowledge bar is cleared and the accounting is honest — but net income up 78% D6 while operating profit fell 16% D5 is a tax mirage, and once you strip it, a thin-margin chain whose growth is all price and no traffic D11 does not earn its four-times-book price D1. A knowable, decent business at an unpayable price is a watch: keep it on the list, name the price, and wait.

Claude — pass (implied buy-below ¥430)

My figures-blind priors survive the ledger — confirmed, and in the direction I feared. I registered RINGER HUT as a genuinely good, fully-recovered champon brand whose "record" headline was a tax artifact over a softening operating line, priced full, behind a quiescent register — modal outcome "compound slowly without re-rating." Two priors were the verdict-bearing gates: is it actually cheap on normalized owner-earnings, and is the operating engine genuinely strong. The ledger overturns neither; it resolves both against the constructive case, harder than I expected — which moves the honest verdict from the watch I anticipated to pass. The one thing I got wrong figures-blind: I hedged toward "a good brand I'd own lower." The magnitudes say the price is not merely full — it is demanding, and the earning power sits below the asset base.

The business is real. Champon is 81.8% of revenue and 76.2% of segment profit D7D10; the own-factory, 100%-domestic-sourcing supply chain E78E35 is a genuine differentiation mechanism, not a slogan. But the question a whole-company owner asks is not "is the brand good" (it is) — it is "does the differentiation earn a return above the assets, for a minority, at this price." Here the ledger is unkind. The record is a tax event: operating profit fell −16.3% D5, ordinary profit was flat F10, and net income set a record +78.4% D6 solely because a ¥597,166千 deferred-tax benefit F296 turned total tax to a credit at a −16.9% effective rate F328. Normalize the tax at the 30.5% statutory rate F327 on pre-tax income F293 and normalized net income is ≈¥1,027,573千 — the reported ¥1.7bn overstates sustainable earnings by ~¥700M, and the "cheap-ish" 34.9× headline D15 is really ~59× on normalized earnings, ~50× on the guide F163.

There is no margin of safety on normalized earnings; the price is demanding. After-tax operating profit capitalized as a no-growth stream at 8% gives an earnings-power value of the firm of ≈¥12.3bn, and after net debt D12, EPV of the equity ≈¥6.75bn — about ¥260 a share. The market cap at ¥2,330 is ¥60.4bn D2: the price pays ~9× the no-growth equity earning power, which requires you to believe normalized earnings compound at 6.3% in perpetuity — a large, unproven growth assumption dressed as a quality stock. And the decisive cross-check: **EPV of the equity (¥6.8bn) sits far below net assets book (¥15.2bn** F261) — the signature that earning power is below the asset base. The moat does not show up as earning power above the assets; that reads as a structure/return problem, not a franchise premium. There is no asset floor either — P/B 3.96× D1, net debt not net cash D12 — and the lease-adjusted EV (capitalizing the ¥4.35bn of off-book rent F274) runs to ~¥100bn, ~71× operating profit. The runway does not rescue it: same-store growth is price, not traffic D11; the store base is net −5 F156; the Vietnam store is optionality E56. Governance is orderly but offers no offsetting unlock — a quiescent foundation register E90E92, no buyback E98, no controlling parent E178, related-party transactions nil E169; a ceiling, not a trap-door, but no agent to convert over-earned capital or force a re-rate. A whole-company owner does not buy the business's modest return at ~9× that return's capitalized value. I would own this brand — at a fraction of this price. At ¥2,330 it is a pass; my downside arithmetic implies a re-engage floor near ¥430, a ceiling not a target.

What a student should take from this: when a "record" bottom line is a one-off tax event on a declining operating line, value the operating line, then ask the one question that settles it — does the moat show up as earning power above the asset base? Here it does not: EPV of the equity (~¥6.8bn) is less than half book (¥15.2bn F261), so the price pays ~9× no-growth earning power for a brand whose differentiation, however real, has never out-earned its own assets. A good brand and an excess return for the buyer who pays full price are different claims; the arithmetic, not the story, tells them apart.

Synthesis

Where the five lenses agree

There is no bear among the five on the business — and no buyer among the five at the price. Every lens grants the champon franchise its due: fifty years of a differentiated brand E5, 100%-domestic sourcing defended as a promise E35, real pricing power (three increases in two years without a volume collapse E63), a clean audit E180, and honest management that flagged the tax quirk and guided earnings down rather than dressing them up E75E62. And every lens reaches the same verdict on the headline: the "+78% record" D6 is a one-off ¥597 million deferred-tax benefit F296 sitting on top of an operating profit that fell −16.3% D5, which management themselves remove next year F163. Buffett calls it "a one-time gift from the tax man"; Munger, "cosmetic"; Pabrai, "a mirage"; Li Lu, "a pretend"; Claude, "a tax event." Five independent reads, one conclusion: value the operating line, not the marquee — and the operating line is softening, not accelerating.

The second agreement is the one that matters for the decision: the price has no margin of safety and no floor. At 3.96× book D1 with net debt D12, there is no asset cushion — the saleable land and securities are ~⅛ of the market cap F212E173D2. On normalized earnings the whole panel lands in the same place: an owner-earnings yield of ~2% (Buffett), a ~2.6% normalized yield (Claude), ~50× forward earnings (Munger, Pabrai), a conservative owner value near book (Li Lu, ~¥640). The dispersion is not in the diagnosis; it is only in what each lens does with a great brand at an ungreat price.

Where the lenses diverge

The split is watch (Buffett, Li Lu) vs pass (Munger, Pabrai, Claude) — and it is not a disagreement about the facts. It is a disagreement about whether a good brand deserves a standing bid at some far-lower price (the watchers) or no bid at all until the whole proposition changes (the passers), and about the one number none of them can settle from a single year: whether the champon franchise earns enough on its capital to ever justify a premium.

The watchers keep it on the list. Buffett would own "this decent little noodle chain — but only near ¥600, where the owner yield finally clears about 8% and I'm paying roughly tangible book" F35. Li Lu, who clears the knowledge bar he usually fails names on, would "begin to accumulate near ¥900, a modest premium to book, and only if the traffic turns" — his falsifier is explicit: two years of volume-led same-store growth (客数 ≥ 100%) plus operating-profit growth D11D5. Both are saying the same thing in different registers: this is a knowable, ownable brand whose only defect is its price, so you write down the price and wait.

The passers will not hold a bid at all, and for two distinct reasons. Munger and Pabrai pass on price-and-structure: "I do not buy 3%-operating-margin chains at 34× current and 50× normalized earnings," says Munger D4D15, and the second segment makes it worse — "tonkatsu is the turd," ¥141 million of profit down 51.5%, Hamakatsu in 債務超過 F351F133, bought at 4× book alongside the raisin [M40]. Pabrai inverts it through the Dhandho door: "at 4× book with net debt, an equity cushion is not a floor — nothing hard sits beneath the price [P1], and the market fears nothing, so this is low-uncertainty priced for perfection, the opposite of what I hunt [P13]." Claude passes on the deepest cut of all — earning power below asset value: "EPV of the equity (¥6.8bn) is less than half book (¥15.2bn) F261, so ¥60bn pays ~9× no-growth earning power for a brand that has never out-earned its own assets [C38]." Notice the passers are not more bearish on the brand than the watchers; they simply refuse to name a buy price for a business whose moat, on this year's arithmetic, does not convert into a return above its assets. Buffett and Li Lu answer the same doubt by demanding a price so low (near or below book) that the question becomes moot.

Two threads run through all five. (a) The borrowed record — the ¥597M deferred-tax benefit F296 — is the single fact that turns a superficially great year into a candidly mediocre one, and every lens strips it before valuing. (b) The off-balance-sheet leases F274E128 and the two 債務超過 subsidiaries F133E19 are the recurring reminder that the tidy 48.9% equity ratio F46 understates the real fixed-charge burden — Munger, Pabrai, Li Lu and Claude each reconstruct it by hand and each finds the leverage worse than the page shows. The disagreement, in the end, is temperamental as much as analytical: whether "a great brand I'd own much cheaper" earns a watch (Buffett, Li Lu) or whether "a great brand whose economics don't justify a premium and whose price offers no floor" earns a flat pass (Munger, Pabrai, Claude). On this name the practitioner should note that the watch thresholds (¥600, ¥900) sit at or below book — so even the constructive lenses are, in substance, saying the same thing as the passers: not at anything like this price.

Self-distance note. The Claude lens holds one of the five verdicts compared above (pass) and wrote this synthesis; it also built the dual-blind reconciled figure table and evidence ledger (two independent passes each, reconciled against page-delimited source text) that all five lenses consumed. That is an unusual concentration of authorship in one model — the answerer, the ledger-builder, and one of the five voters are the same system. Read the synthesis with that in mind.

Prediction-vs-actual: VOID. This was an autonomous headless cycle; the human blind prediction is voided (void: no-human-prediction, never forged). No prediction-vs-actual scoring applies to this study.

Verdict accounting (fixed ex-ante)

• A buy-below-¥X verdict is price-falsifiable against the unadjusted stamp. The two watch thresholds recorded here — Buffett ¥600 and Li Lu ¥900 — are far below the ¥2,330 stamp; they are the prices at which each watcher would re-engage. Claude's pass publishes an implied downside floor of ¥430 (C44), a ceiling not a target, distinct from a buy-below.
• pass / watch / too-hard are recorded but unscored in any future review. Munger's pass, Pabrai's pass, and Claude's pass carry no buy-below by construction.
• The original verdict counts at its original stamp regardless of later corrections.
• On a stock split, reverse split, or consolidation, the buy-below threshold restates mechanically by the announced ratio (corporate-action disclosure cited); the stamp itself never restates.

What would change our minds

Pre-registered falsifiers, per lens issuing a buy-below or watch. Future review notes score against these, not hindsight.

• Buffett (watch ¥600). Flips to a pass (not ownable at any sensible price) if normalized owner earnings — reported net income stripped of the one-off deferred-tax benefit F296 plus D&A less maintenance capex — fall below ~¥1.0bn for two consecutive years while the store count keeps shrinking F156, i.e. the champon franchise stops out-pricing its own traffic loss D11F151. It flips toward buy-below only under ~¥600, where the owner-earnings yield clears ~8% at roughly tangible book F35.
• Li Lu (watch ¥900). Two consecutive fiscal years in which same-store growth is carried by customer count rather than per-customer spend (既存店客数 ≥ 100%, with SSS ≥ customers, reversing the FY2026 gap F150F151D11) and consolidated operating profit grows year-on-year — evidence champon has regained volume-driven reinvestment economics — would move this from watch toward buy-below at a higher price. Conversely, a third straight year of operating-profit decline (after the −16.3% of FY2026 D5) with SSS still all-price would retire it to pass.
• Munger (pass), Pabrai (pass) and Claude (pass) carry no price falsifier by construction — they issued neither a buy-below nor a watch, so there is nothing of that kind to pre-register. We say so plainly rather than manufacture one. Munger would revisit below ~¥1,400 (≈20× guided normalized earnings F163); Claude's implied re-engage floor is ~¥430 — both are contexts for a future look, not pre-registered buy triggers.

The single observable most lenses converge on is whether same-store growth reverts to volume — resolved by the FY2027 same-store disclosure (客数 vs 売上高 F150F151) and whether operating profit grows off the −16.3% base D5.

What this taught the checklists

Queued for the next study (F2 revision proposals; see docs/process/evolution.md), attributed per lens:

• Buffett — B42 (owner earnings) needs a "one-off tax item" sub-prompt. The item normalizes maintenance capex and depreciation well but does not instruct the analyst to strip non-recurring tax effects (DTA recognition, valuation-allowance release) before computing the owner-earnings yield — and here the entire "record" F296 is exactly that. The mistake gets made in the yield computation itself, so the fix belongs there, not only in B46. Minor companion: B26/B60 could name a restaurant/asset-heavy sub-bar (capex-to-depreciation persistently ≥ ~1.0 with flat unit count F311F306F156 as the fingerprint of restricted earnings).
• Munger — two ledger-shaped notes. M19 (volume cost curve) should require the analyst to compute or flag the FL-cost ratio (food + labour as % of revenue) when management discloses it as a KPI F148E32, treating its absence as data-insufficient rather than pass. And M34 (moat direction) wants a three-year same-store series (customers, realized pricing, operating margin, net store count); the ledger gave one year F150F151, enough to diagnose but not to score the trend at full confidence.
• Pabrai — add a "record-earnings tax-mask" check adjacent to P56: when net income diverges sharply from operating profit, decompose the tax line — a one-off DTA recognition can invert a declining operating result into a record headline; normalize on the operating trend and management's own guidance, never the flattered PER. Ringer Hut is the clean prompting case: −16.3% operating D5 wearing +78.4% net D6 via ¥597M F296, guided back down F163. Minor: P25 could name the Japanese-specific 未適用-ROU case where operating rent F274 materially understates economic leverage.
• Li Lu — three. (1) L15/L42 test "clears the hurdle" and "works on earnings alone" without a stated numeric owner-hurdle — a bracketed default (e.g. "≥ ~8% IRR from yield + realized per-share growth, no re-rating") would make pass/fail reproducible across lenses. (2) An explicit off-balance-sheet-lease sub-prompt under L20/L47 directing the analyst to capitalize disclosed operating rent when the ROU standard is 未適用 F274E128 — a recurring feature of Japanese consumer names. (3) A named "reported metric flattered by a one-off" flag under L4 — does the verdict survive removing the non-recurring tax/gain/subsidy item F296 that flips ROE from ~7% to 12% F51?
• Claude — two, plus a library seed. (1) C38 should be sharpened for the premium-priced case, not only the cheap-asset case: when P/B is high, state the ratio of price to EPV(equity) explicitly — a price far above an EPV that is itself below book (¥60bn ≫ ¥15.2bn book ≫ ~¥6.8bn EPV D2F261) is a growth-priced quality trap, the mirror of the below-book value case. (2) C9 should distinguish a recovered margin percentile from a structural one when the peak year is tax- or recovery-flattered — apply the margin-persistence base rate to the operating margin, not the ROE the record inflates F51. Library (Class-level): RINGER HUT seeds the flagged-gap class "healthy, premium-valued, slow-growing, vertically-integrated Japanese branded-restaurant / consumer-franchise mid-cap" (N=1), outcome on the key axis = earning power below asset value at a premium price — a health-and-quality counterweight to the distress-heavy library, and distinct from the ICT-SI health class (TSUZUKI).

---

Corrections