Buffett checklist — v0.3.0

This profile deliberately carries EARLY Buffett — the Graham-trained bargain hunter of the 1957–1969 partnership letters plus the owner-earnings calculator of the Berkshire letters — because the Munger profile owns the quality/mental-models territory. The core move is simple: could you explain the business, would you buy the whole thing at today's market cap, and is the price a bargain against owner earnings or hard assets? It looks for cheapness you can verify from filings, not greatness you have to predict. Later-Berkshire franchise and moat tests appear only where Buffett himself made them checkable from filings (franchise test, economic goodwill, incremental returns); Munger's psychology and Pabrai's asymmetry stay in their own profiles.

Items tagged [PRACTITIONER] are answered by the practitioner, never by the profile agent — the agent marks them practitioner-pending and moves on; where an Agent proxy is given, the agent evaluates the proxy instead.

Evidence grades: quote-backed = verbatim-verified primary source; teaching-derived = faithful to published teaching, paraphrased; grades signal how much weight the citation itself can bear.

Items run B1–B104 as one flat list (ID gaps mark merged items; IDs are never renumbered); section headers are organizational only and every item has equal standing.

Circle of competence & process

B1. Circle of competence

• Asks: Can you explain in one plain paragraph how this company makes money — what it sells, who pays, and why they keep paying — and write a dated inside/outside-the-circle call with the reason? Heavy technology or dynamics you can't independently judge means outside.
• Good: A paragraph a non-specialist would follow, with no hand-waving over any segment above ~20% of revenue or profit; "outside" routes to too-hard regardless of how cheap the figures look — "the size of that circle is not very important; knowing its boundaries, however, is vital."
• Source: 1996 Berkshire letter; 1982 letter (acquisition criteria: simple businesses). (quote-backed)

B2. Can't-estimate-it test

• Asks: Can you make a rough, defensible estimate of the company's earnings ten years out?
• Good: A one-page future-earnings estimate a skeptic could audit; if key drivers are unknowable, the verdict is too-hard — "If you don't feel comfortable making a rough estimate of the asset's future earnings, just forget it and move on."
• Source: 2013 Berkshire letter. (quote-backed)

B3. High-probability qualitative insight

• Asks: State in one falsifiable sentence a qualitative insight about this business with a very high probability of being right, plus the primary evidence behind it.
• Good: The sentence and its evidence written down (AmEx model: "cardholders kept using the card through the scandal"); if no such insight exists, the idea must pass on the figures alone.
• Source: Partnership letter, October 9, 1967 — the "high-probability insight" that "causes the cash register to really sing." (quote-backed)

B4. Facts, not forecasts

• Asks: List the 3–5 load-bearing facts of the thesis. Is each verified from a primary filing — and is any hockey-stick projection doing load-bearing work?
• Good: Each fact cited to a document in the figure table; where the future is assumed better than the past, the case is re-run without that assumption. "You will be right ... if your hypotheses are correct, your facts are correct, and your reasoning is correct."
• Source: Partnership letter, January 24, 1962; 1984 Berkshire letter (historical evidence over projections). (quote-backed)

B5. What should happen, not when

• Asks: Strip every timing and market-level assumption. Does the thesis specify WHAT should happen (earnings accrual, asset conversion, corporate event) and still work if the index goes nowhere for three years?
• Good: Realization mechanics named with no timing prediction — "we tend to concentrate on what should happen, not when it should happen."
• Source: Partnership letter, July 12, 1966. (quote-backed)

B6. No macro forecast in the thesis

• Asks: Does the case require a correct call on rates, GDP, elections, commodities, or FX to work?
• Good: The write-up survives deletion of all macro sentences; any macro variable that can kill the thesis is a disqualifier, not a forecast to be made — "We will continue to ignore political and economic forecasts."
• Source: 1994 Berkshire letter; 2013 letter. (quote-backed)

B7. Mr. Market is servant: price, don't time

• Asks: Does the intrinsic-value derivation (the valuation work of B89–B104) reference the current market price, the price chart, or the 52-week range at any point before the value estimate is stated — and is the buy trigger a standing buy-below price rather than a view on where the market is headed?
• Good: The derivation in the output builds value from filed figures alone, with the current price appearing only after the value estimate, for comparison; a buy-below number valid regardless of market direction; no anchoring to 52-week ranges or what it "used to trade at" — "We try to price, rather than time, purchases."
• Source: 1987 Berkshire letter (Mr. Market); 1994 letter; 2008 letter ("Price is what you pay; value is what you get"). (quote-backed)

B8. Why does the bargain exist

• Asks: Name the specific, non-fundamental reason for the mispricing — obscurity, illiquidity, payout policy, forced sellers, scandal headlines — and list which figure-table rows actually deteriorated versus which merely reflect mood.
• Good: A documented transient cause (Commonwealth: ~300 holders, ~two trades a month, no dividend despite ~$10/share of earnings) with permanent-impairment alternatives checked and ruled out; for large covered names, an explanation of why the crowd is wrong. If no transient cause can be named, treat the discount as the market knowing something.
• Source: Graham, The Intelligent Investor ch. 8; Partnership letter, February 1959 (Commonwealth); 1994 Berkshire letter ("Fear is ... the friend of the fundamentalist"). (quote-backed)

B9. Return from the asset, not the next buyer

• Asks: Does the base case clear the hurdle from what the asset itself produces (earnings, dividends, buybacks at value) with the exit multiple at or below entry?
• Good: Return decomposition with zero contribution from multiple expansion — "there is nothing at all conservative ... about speculating as to just how high a multiplier a greedy and capricious public will put on earnings."
• Source: Partnership letter, January 24, 1962; 2000 and 2013 Berkshire letters (speculation test). (quote-backed)

B10. Approach discipline

• Asks: Does the idea sit inside a playbook you understand and have practiced (net-net, hidden assets, workout, franchise-under-a-cloud), or is it drift toward a fashionable game?
• Good: The idea maps to a named playbook; "others are making easy money in this" appears nowhere — "I will not abandon a previous approach whose logic I understand ... to embrace an approach which I don't fully understand."
• Source: Partnership letter, October 9, 1967. (quote-backed)

B11. Category declaration

• Asks: Before sizing, classify the idea — general (private-owner basis), general (relatively undervalued), workout, or control — and name the return driver that category implies.
• Good: One category, one driver, written down; a "workout" with no timetable is a general, and a "general" needing board seats is a control.
• Source: Partnership letters, January 24, 1962 and January 18, 1965 (category definitions). (teaching-derived)

B12. Fat-pitch selectivity

• Asks: Ranked against everything else studied and against adding to existing holdings, is this in the best cell of the strike zone — good enough to be one of a dozen lifetime decisions?
• Good: A written ranking versus current alternatives; "barely in the strike zone" is a documented pass, not a small buy — there is no called strike for not swinging.
• Source: 1997 Berkshire letter (Ted Williams); 2022 letter ("about a dozen truly good decisions – that would be about one every five years"). (quote-backed)

B13. Concentration test

• Asks: Does the idea meet the two-part bar for a large position — extremely high probability the facts and reasoning are correct, AND very low probability that anything drastically changes the underlying value? Would you put 10%+ of net worth in?
• Good: Both halves argued in writing with the specific facts listed; inability to justify concentration is evidence the work or the business is inadequate.
• Source: Ground rule added November 1965 / Partnership letter, January 20, 1966 (up to 40% under those conditions); 1993 Berkshire letter (concentration comfort). (quote-backed)

B14. Risk is permanent loss, not beta

• Asks: What is the reasoned probability of a permanent loss of purchasing power over the holding period — assessed on business economics, management, price, taxes, and inflation, never on volatility?
• Good: A written permanent-loss inventory (leverage, obsolescence, fraud, regulatory kill) with each pathway bounded or disqualifying; beta appears nowhere — "Never risk permanent loss of capital."
• Source: 2011 Berkshire letter (risk definition); 1993 letter (five business risk factors); 2023 letter. (quote-backed)

B15. Yardstick pre-registration

• Asks: Before purchase, write down the benchmark, the minimum evaluation window (three years or more, ideally spanning weak and strong markets), and the relative result that would count as thesis failure.
• Good: Named benchmark, dated window, kill criterion — recorded in the study before the buy, judged relatively over the window, not by any single year's absolute sign.
• Source: BPL Ground Rules (1957–58, restated January 18, 1963); January 30, 1961 letter. (teaching-derived)

B16. Down-market behavior expectation

• Asks: In a 20–30% market decline, what fraction of the price is covered by hard assets or dated deal proceeds — is relative resilience plausible?
• Good: A downside bridge (NCAV or workout proceeds vs price) plus the expectation stated: "a year in which we declined 15% and the Dow declined 30%" is "much superior to a year when both we and the Dow advanced 20%."
• Source: BPL Ground Rules / results discussions, 1960–63 letters. (quote-backed)

B17. Opportunity-cost and friction hurdle

• Asks: Does the expected annualized return to realization — after transaction costs, taxes, and realistic timing slippage — beat the passive alternative and the best existing holding?
• Good: A written after-cost, after-tax comparison; ideas below the passive alternative rejected regardless of narrative, because modest compounding differences become staggering outcome differences.
• Source: "The Joys of Compounding," Partnership letters 1963–65; 2005 Berkshire letter (Gotrocks frictional costs). (quote-backed)

B18. Tax-tail check

• Asks: Is any buy, hold, or sell judgment driven by tax outcomes rather than comparative value?
• Good: Every decision justified on pre-tax comparative value first, tax noted afterward as a cost — "More investment sins are probably committed ... because of 'tax considerations' than from any other cause."
• Source: Partnership letter, January 18, 1965. (quote-backed)

B19. Ten-year ownership test

• Asks: Would you buy at today's price if the market closed for ten years — and does the study contain the ten-year earnings path justifying a yes?
• Good: A written "willing to own with no quote for ten years because ..." sentence, with the only sell triggers being thesis-breaking deterioration or gross overvaluation — "If you aren't willing to own a stock for ten years, don't even think about owning it for ten minutes."
• Source: 1996 Berkshire letter; 2023 letter (patience with wonderful businesses). (quote-backed)

B20. Playing field, not scoreboard

• Asks: During the holding period, are you tracking operating results against the original thesis rather than the stock quote?
• Good: Position reviews compare revenue, margins, and per-share earnings to thesis projections; price appears only as opportunity, never as evidence the thesis is right or wrong.
• Source: 2013 Berkshire letter ("Games are won by players who focus on the playing field"). (quote-backed)

B21. Assumption audit

• Asks: For every model output used (valuation, company risk models, actuarial figures), list the assumptions behind the symbols and test whether the conclusion survives adverse changes.
• Good: An assumptions table with sensitivity checks; conclusions that flip on one soft assumption labeled fragile — "Beware of geeks bearing formulas."
• Source: 2008 Berkshire letter. (quote-backed)

B22. Exit rule and replacement test

• Asks: Before buying, define the exit: at what fraction of conservatively appraised value does the position get sold, and what documented comparison justifies a swap?
• Good: A stated exit threshold (Commonwealth sold at $80 against a $135 value) and a written value-vs-value comparison for any replacement — capital moves only because the replacement is cheaper relative to value, not because the market said so.
• Source: Partnership letter, February 1959 (Commonwealth sale). (teaching-derived)

Business economics

B23. Economic franchise test

• Asks: Does the product meet all three franchise conditions: (1) needed or desired, (2) thought by customers to have no close substitute, (3) not subject to price regulation?
• Good: A written yes/no on each with evidence — repeat or mission-critical demand, sustained price premium versus alternatives, no rate-setting regulator over core pricing.
• Source: 1991 Berkshire letter (franchise definition). (quote-backed)

B24. Pricing power with capital-light growth

• Asks: Can the company raise prices "even when product demand is flat and capacity is not fully utilized" without losing share or unit volume — and grow dollar volume "with only minor additional investment of capital"?
• Good: Price/mix-led revenue growth with flat-to-rising units over 5–10 years; incremental capex plus working capital per incremental revenue dollar small (e.g., under ~$0.15) with stable-or-rising asset turns.
• Source: 1981 Berkshire letter (the two prized business characteristics); 2011 FCIC teaching on pricing power. (quote-backed)

B25. Nameable barrier: low-cost producer or dominant brand

• Asks: Which recognized moat type applies — demonstrable low-cost production or a powerful brand — and what figure evidences it?
• Good: A unit-cost/expense-ratio advantage versus named competitors, or a persistent realized-price premium with stable-or-growing volume; "good management" alone fails the item.
• Source: 2007 Berkshire letter (GEICO/Costco vs Coca-Cola/Gillette/AmEx); 1991 letter (low-cost operator). (quote-backed)

B26. Returns on net tangible capital behind a moat

• Asks: Does the company earn good returns on the net tangible capital its operations actually require, and can you name the structural barrier protecting those returns from competitive assault?
• Good: After-tax operating income / (equity + net debt − goodwill − intangibles − excess cash) at or above ~15% in most of the last 10 years without leverage doing the work, plus a one-sentence barrier statement — "a truly great business must have an enduring 'moat.'" Where those returns run considerably above market rates, the excess is economic goodwill — earning power residing in brand, habit, or distribution rather than plant that must be re-bought at inflated prices — and it doubles as inflation resilience: "businesses needing little in the way of tangible assets simply are hurt the least" by inflation, with low tangible assets-to-sales versus industry as the fingerprint.
• Source: 2019 Berkshire letter (acquisition criteria); 2007 letter (moat); 1987 letter (20% ROE screen); 1983 letter, goodwill appendix (economic goodwill and inflation). (quote-backed)

B28. Moat tested under assault

• Asks: Has the moat actually been attacked and held? Identify at least one past episode — new entrant, price war, technology shift, recession — and show margins and share through it.
• Good: Operating margin and market share held or recovered within two years through a named assault; an untested moat, or one that failed a test, is flagged — "business history is filled with 'Roman Candles.'"
• Source: 2007 Berkshire letter. (quote-backed)

B29. Severe change and rebuilt moats

• Asks: Is the industry facing severe technological, regulatory, or behavioral change — or one where the advantage must be re-earned each product cycle?
• Good: The advantage rests on assets that don't expire (distribution, brand, cost position, switching costs); "severe change and exceptional returns usually don't mix," and a moat that must be continuously rebuilt is eventually no moat at all.
• Source: 1987 Berkshire letter; 2007 letter. (quote-backed)

B30. Moat trajectory: time is the friend

• Asks: Over the past decade, has the competitive position strengthened without heroic intervention?
• Good: A 10-year trend of market share, gross margin, and return on tangible capital that is flat-to-improving through at least one recession — "Time is the friend of the wonderful business, the enemy of the mediocre."
• Source: 1989 Berkshire letter ("Mistakes of the First Twenty-five Years"); annual-meeting widen-the-moat teachings. (quote-backed)

B31. Survives mediocre management

• Asks: Would the economics survive a stretch of inept management or the departure of the current CEO/founder?
• Good: Returns predate the current era and are explained by structural position, not one operator's brilliance; succession visible — "When a management with a reputation for brilliance tackles a business with a reputation for bad economics, it is the reputation of the business that remains intact."
• Source: 1989 Berkshire letter; 1991 letter (franchises tolerate mismanagement); 2007 letter (superstar test). (quote-backed)

B32. Ten-year sameness and longevity

• Asks: Is the company selling essentially the same product, the same way, as a decade ago — and is there a specific reason it will still be needed in 20 years?
• Good: The current business description matches the decade-old one in core product, customer, and economics, and the survives-two-more-decades argument names the demand driver without a technology guess — "the best business returns are usually achieved by companies that are doing something quite similar today to what they were doing five or ten years ago."
• Source: 1987 Berkshire letter; 2023 letter ("Berkshire is not big on newcomers"). (quote-backed)

B33. Incremental returns: three savings accounts

• Asks: What did each additional dollar of retained/invested capital earn over the last 5–10 years (Δ operating earnings ÷ Δ invested capital) — and which account is this: great, good, or gruesome?
• Good: A reconciled incremental-return figure at or above ~12–15% with identifiable reinvestment runway at similar returns; See's is the "great" benchmark, versus the gruesome account that "requires you to keep adding money at those disappointing returns."
• Source: 2007 Berkshire letter (See's / three savings accounts); 1992 and 2019 letters (reinvestment runway). (quote-backed)

B34. Gruesome screen: growth that eats capital

• Asks: Is revenue growing rapidly while requiring significant capital and earning little — cumulative capex plus acquisitions versus cumulative free cash flow over 10 years?
• Good: Growth self-funded at good returns, or explicitly reclassified as a cost; a decade of negative cumulative FCF alongside rising revenue fails — "The worst sort of business is one that grows rapidly, requires significant capital to engender the growth, and then earns little or no money. Think airlines."
• Source: 2007 Berkshire letter. (quote-backed)

B35. Earnings materially higher in 5/10/20 years

• Asks: From units, price, and reinvestment, why are earnings virtually certain to be materially higher in five, ten, and twenty years — not merely hoped to be?
• Good: A written causal chain grounded in figure-table rows (volume trend, pricing history, reinvestment runway); if the honest answer is "can't know," the item fails toward too-hard rather than being fudged.
• Source: 1996 Berkshire letter. (quote-backed)

B36. Commodity economics check

• Asks: Is the product undifferentiated in any customer-important way, in an industry carrying overcapacity — and does the 10-year ROE/ROIC table show a business earning below its cost of capital across the cycle?
• Good: Differentiation customers pay for (price premium, switching costs) or documented capacity discipline; a chronic sub-par earner is admissible only as an asset play with a catalyst and a short clock, never a hold-forever — the Berkshire textile lesson.
• Source: 1982 Berkshire letter (commodity/overcapacity); Berkshire textile operations (partnership-era control). (teaching-derived)

B37. No turnaround dependency

• Asks: Does the case require a turnaround rather than resting on demonstrated results?
• Good: Thesis stands on trailing, reported earning power with no dependency on new management, restructuring success, or a cyclical rebound — "future projections are of little interest to us, nor are 'turn-around' situations."
• Source: 1982 Berkshire letter (acquisition criteria); 1979 letter (turnarounds). (quote-backed)

B38. Franchise survival under scandal (AmEx test)

• Asks: If the price collapse follows a scandal or one-time loss, do transaction volumes, revenue, and customer counts hold through the crisis quarters?
• Good: Segment/volume data through the event showing customers kept transacting; the loss quantified as one-time and separable from the ongoing franchise, as with American Express through the salad-oil scandal.
• Source: American Express position 1964–66; Buffett's November 1964 letter to AmEx CEO Howard Clark. (teaching-derived)

B39. Melting-business catalyst requirement

• Asks: If the industry is in secular decline, does the thesis rely on asset conversion with a named catalyst (control party, activist, announced liquidation or divestiture) rather than operations recovering?
• Good: Either the industry is sound, or a named conversion mechanism with a timetable exists; "cheap and the cycle will turn" is not accepted for a secularly declining business — the Dempster and Berkshire-textile lesson.
• Source: Partnership letters 1962–63 (Dempster); Berkshire Hathaway control 1965–69. (teaching-derived)

B40. Run-of-the-mill business at the right price

• Asks: Which claim is the write-up making — "adequate business at a wide discount" or "superior business at a fair price" — and is the evidence bar matched to the claim?
• Good: The claim labeled explicitly: adequacy needs a wider discount plus management and industry decent enough that waiting isn't destructive; superiority needs return-on-capital proof, not adjectives.
• Source: Partnership letter, January 18, 1964 (private-owner generals). (teaching-derived)

B41. Host-economy tailwind

• Asks: Does the thesis ride, rather than fight, the long-term direction of the company's home economy and institutional regime?
• Good: A base-rate note on the industry's 10–20-year real growth versus GDP, plus any documented structural shifts in the company's favor (e.g., TSE governance reforms); no leg of the thesis requires the home economy to stagnate — "Never bet against America," generalized to the host economy.
• Source: 2018 Berkshire letter (The American Tailwind); 2020 letter; 2023 letter (Japan). (quote-backed)

Earnings quality

B42. Owner earnings and the whole-business yield

• Asks: Compute owner earnings per the 1986 definition — reported earnings plus depreciation/amortization and other non-cash charges, minus the average capitalized spending required to fully maintain competitive position and unit volume (triangulate maintenance capex at least two ways; it "must be a guess") — and divide by market cap.
• Good: Owner earnings positive and reconciled to the figure table, with divergence from net income traceable in the cash-flow statement; the yield comfortably in double digits or clearly superior to the passive alternative, judged as if buying the entire company at today's price. When maintenance capex AND depreciation are both undisclosed, state the fallback used: triangulate from industry peers with the method shown, or use reported net income flagged as an upper bound — silently proceeding is a fail; if neither fallback is defensible, the item is data-insufficient.
• Source: 1986 Berkshire letter, appendix (owner-earnings definition). (quote-backed)

B43. Ten-year earnings-power stability

• Asks: Over the last 10 fiscal years, has the company shown stable or gently rising earnings power — operating income positive every year, at most one net loss year — with the valuation built on the cycle average rather than the latest or peak year?
• Good: A 10-row table of revenue, operating income, net income, and ROE with no unexplained collapses and the average computed across at least one bad year; "demonstrated consistent earning power" — a business that needed a rescue or reinvention mid-decade fails.
• Source: 1982 Berkshire letter (acquisition criteria); Graham's average-earnings doctrine; 1987 letter (Fortune ROE study). (quote-backed)

B44. EBITDA and "cash flow" rejection

• Asks: Does the bull case (or management's own headline metric) lean on EBITDA or pre-capex "cash flow" in a business where required capex is significant?
• Good: Valuation uses owner earnings, never EBITDA, for manufacturing, retailing, extractive, or utility-like businesses; any EBITDA-led disclosure reconciled back to GAAP with exclusions quantified — "'Cash Flow' is meaningless in such businesses."
• Source: 1986 Berkshire letter, appendix; 2002 letter ("Trumpeting EBITDA ... is a particularly pernicious practice"). (quote-backed)

B45. Depreciation vs reality (tooth-fairy test)

• Asks: Over 10 years, how does depreciation compare to capex alongside unit volume — does reported depreciation understate the true cost of staying in business?
• Good: Cumulative D&A versus cumulative capex with the maintenance portion estimated; earnings marked down where capex persistently exceeds D&A just to hold volume flat — "does management think the tooth fairy pays for capital expenditures?"
• Source: 2000 Berkshire letter; 2002 letter (depreciation). (quote-backed)

B46. Operating vs non-operating earnings split

• Asks: Split reported earnings into operating earnings versus investment income, mark-to-market swings, and one-off gains for each of the last 5 years — is a portfolio or a run of gains masking core decline?
• Good: A 5-year two-column split with the trend judged on the operating line alone; the Sanborn pattern — investment income cushioning a deteriorating core — named if present.
• Source: 2019 Berkshire letter (focus on operating earnings); Partnership letter, January 30, 1961 (Sanborn). (quote-backed)

B47. Retained-earnings reconciliation

• Asks: Over 5–10 years, does book value growth plus dividends roughly equal cumulative reported earnings — and what return did the retained dollars actually earn?
• Good: Reconciliation within ~10–15% with gaps explained by named items (write-offs, buybacks above book, currency); ΔEPS over the window ÷ cumulative retained earnings per share at or above a ~10% hurdle, with retained cash traceable to reinvestment, buybacks, or debt reduction.
• Source: Early-Buffett control practice (Dempster/Berkshire per-share asset value); 2019 Berkshire letter (retained earnings working). (teaching-derived)

B48. Restricted vs unrestricted earnings

• Asks: What portion of reported earnings is "restricted" — required to be plowed back merely to maintain the company's economic position — and thus unavailable to owners?
• Good: An earnings split driven by the asset/profit ratio and replacement-cost needs; high asset-intensity businesses shown to have materially lower distributable earnings than the reported "ersatz" figures.
• Source: 1984 Berkshire letter (dividend policy). (quote-backed)

B49. Look-through earnings

• Asks: Including the company's share of retained earnings in significant minority investees (less an allowance for tax on hypothetical distribution), what are its look-through earnings?
• Good: A table: reported operating earnings + proportional retained earnings of major investees − dividend-tax allowance, used whenever equity stakes are material.
• Source: 1991 Berkshire letter. (quote-backed)

B50. Amortization vs economic goodwill

• Asks: Do goodwill/intangible amortization charges distort earnings relative to what is actually happening to the company's economic goodwill?
• Good: Earnings analyzed before purchased-intangible amortization, paired with an evidenced judgment on whether economic goodwill is stable, rising, or genuinely eroding — and marked down when erosion is real.
• Source: 1983 Berkshire letter, goodwill appendix. (teaching-derived)

B51. EPS growth vs retention arithmetic

• Asks: Is per-share earnings growth anything more than the automatic result of retaining earnings — is the return on equity capital actually high?
• Good: Actual EPS growth compared against what mere retention at a static ROE would mechanically produce (a dormant savings account compounds too); management judged on ROE without undue leverage or gimmickry, not on the EPS series.
• Source: 1977 and 1979 Berkshire letters. (teaching-derived)

B52. Serial "one-time" charges

• Asks: How many restructuring, impairment, or "special" charges in the last 10 years, and what do earnings look like with them treated as ordinary costs?
• Good: Cumulative special charges under ~5% of cumulative operating income or tied to one identifiable episode; more than two "non-recurring" charges in a decade means they are recurring — "The distortion du jour is the 'restructuring charge.'"
• Source: 1998 Berkshire letter; Owner's Manual principle 12 (no "big bath" maneuvers). (quote-backed)

B53. Managed numbers: guidance games and period-shifting

• Asks: Does management trumpet earnings targets and reliably "hit the number" — and is there evidence of earnings moved between periods (reserve timing, pulled-forward revenue, deferred expense)?
• Good: No standing growth promise; the last 20 quarters show natural lumpiness rather than penny-beat streaks; accruals, deferred revenue, and reserve movements show no period-shifting pattern — fudging that snowballs "can turn ... into fraud."
• Source: 2000 Berkshire letter (accounting shenanigans); 2002 letter (guidance suspicion); 2019 letter. (quote-backed)

B54. Footnote intelligibility

• Asks: Take the three longest footnotes in the latest annual report: can you restate each in plain language after one careful reading?
• Good: Every material footnote explainable in a sentence or two; one that resists honest explanation is a deliberate signal — "if you can't understand a footnote ... it's usually because the CEO doesn't want you to."
• Source: 2002 Berkshire letter. (quote-backed)

B55. Pension assumptions and funded status

• Asks: What return does the company assume on pension assets, what does the actual bond/cash allocation make mathematically achievable, and what is the net funded status?
• Good: Assumed return justified by the actual asset mix after fees; any deficit added to debt in the figure table; an aggressive assumption reduces the analyst's earnings figure.
• Source: 2007 Berkshire letter ("Fanciful Figures"). (quote-backed)

B56. Stock compensation is an expense

• Asks: Does the company's preferred earnings measure exclude stock-based compensation, and how large is the SBC drag with expected dilution counted?
• Good: Analyst earnings include SBC at fair value with 5-year share-count creep reconciled; any non-GAAP measure adding SBC back is rejected — if options aren't compensation, what are they, and if compensation isn't an expense, what is it?
• Source: 1998 Berkshire letter (option accounting); annual-meeting restatements. (quote-backed)

B57. Mark-to-model exposure

• Asks: What fraction of assets and equity is valued by model rather than market (Level 3 or equivalent), and are the people doing the marking paid on the marks?
• Good: Model-marked assets small relative to tangible equity (under ~10–15%) with the marks-to-bonus link checked; large model-marked books with mark-linked pay fail — "mark-to-model degenerates into ... mark-to-myth."
• Source: 2002 Berkshire letter (derivatives). (quote-backed)

B58. Reserve development honesty

• Asks: For insurers and estimate-heavy books: has prior-year reserve development been favorable or adverse over the last 5–10 years — were past earnings real?
• Good: Development triangles neutral-to-favorable; persistent adverse development means past earnings and book value were overstated, and the analyst restates them.
• Source: 2007 Berkshire letter (loss-reserve "honor system"). (quote-backed)

B59. Cockroach rule

• Asks: Any restatements, regulator accounting actions, or documented integrity lapses (backdating, related-party favors, aggressive tax games) in the last decade?
• Good: A clean search of filings and enforcement records; one confirmed cockroach shifts the burden of proof onto the study or forces exit — "There is seldom just one cockroach in the kitchen."
• Source: 2002 Berkshire letter; 1998 letter (disrespect for accurate reporting). (quote-backed)

Balance sheet

B60. Good returns without leverage

• Asks: Are the good returns on equity earned "while employing little or no debt" — what is the return on unleveraged tangible equity?
• Good: ROE >15–20% with conservatively low debt, or a computed unlevered return on capital that remains good after stripping leverage; a fine ROE built mostly on borrowing fails. Owner's Manual standard: reject interesting opportunities rather than over-leverage.
• Source: 1982 Berkshire letter (acquisition criteria); Owner's Manual principle 7 ("We use debt sparingly"). (quote-backed)

B61. Survive-the-wait leverage test

• Asks: Can the company survive an indefinite wait — check debt maturities, covenants, and fixed-charge coverage at trough earnings for any path to forced dilution or fire sale before value is realized?
• Good: No significant maturities inside ~3 years, trough coverage comfortably above 2x, no covenant cliff; Buffett borrowed only against workouts with near-certain timetables, never against generals, because leverage against uncertain timing converts a temporary quotation into a permanent loss.
• Source: Graham margin-of-safety doctrine in BPL practice; BPL borrowing rule (workouts only, ~25% cap). (teaching-derived)

B62. Tide-goes-out funding test

• Asks: If credit markets closed for two years, would the company survive on its own resources — and would it remain solvent through a shock worse than any it has yet experienced?
• Good: A maturity ladder covered by cash, committed facilities, and stressed internal generation with no required refinancing at any point — "you only find out who is swimming naked when the tide goes out."
• Source: 2001 Berkshire letter; 2009–10 letters (kindness of strangers); 2023 letter (beyond-precedent stress). (quote-backed)

B63. Full prior-claims deduction

• Asks: In any NCAV or liquidation computation, are ALL claims senior to the common deducted — debt, preferred at liquidation value plus arrears, minority interests, tax and lease obligations?
• Good: An itemized claim stack above the common, each at settlement value, all netted before the per-share bargain arithmetic; nothing senior omitted.
• Source: Graham, Security Analysis (net current assets "after deducting all prior claims"). (teaching-derived)

B64. Receivables, inventory, and tested book values

• Asks: Compute receivable days and inventory turns against peers and the company's own history — and has stated book been tested by actual realizations (disposal proceeds versus carrying values, write-down history, reserve adequacy)?
• Good: Turns near industry norms, or explicitly deeper haircuts fed back into the valuation table; untested book (no disposals, thin reserves, slowing turns) gets deeper haircuts — Dempster's book was $75 against ~$35 realizable.
• Source: Partnership letters, January 1962 / January 1963 (Dempster). (teaching-derived)

B65. Melting bargain test

• Asks: Compute NCAV (or tangible book) per share for each of the last 3–5 years alongside operating cash flow — is the asset value stable or growing while you wait?
• Good: Flat-to-growing NCAV per share and non-negative cash generation; a shrinking NCAV plus cash burn requires a named catalyst on a short, dated clock, or the item fails — a discount that shrinks each year through operating losses is not a margin of safety.
• Source: BPL net-net practice; Dempster/Berkshire lesson on eroding discounts. (teaching-derived)

B66. Off-balance-sheet and contingent claims

• Asks: Are there claims outside the liabilities section — pension deficits, litigation, guarantees of subsidiaries, preferred arrears? Size the worst case and deduct it.
• Good: Each contingency named, quantified, and deducted in the valuation table; the AmEx salad-oil episode is the model — a subsidiary's contingent liability can land on the parent's common.
• Source: American Express 1963–64 (warehouse subsidiary liability); Graham's prior-claims doctrine. (teaching-derived)

B67. Float and its cost

• Asks: Does the business hold money it doesn't own — insurance float, customer prepayments, negative working capital — and at what cost over the cycle?
• Good: Float quantified over 10 years with its effective cost computed (insurers: underwriting result ÷ average float) below the long-term government bond rate, ideally negative — "float is money we hold but don't own."
• Source: 2002 Berkshire letter; float framework of the 1977–93 and 2018–22 letters. (quote-backed)

B68. Deferred taxes as interest-free loan

• Asks: Does the company carry a large, durable deferred-tax liability that functions as an interest-free loan from the government?
• Good: Deferred tax balances quantified with durability assessed (persists or grows as long as assets are held); treated as low-cost funding rather than near-term debt in the figure table.
• Source: 1989 Berkshire letter (deferred taxes on unrealized gains). (teaching-derived)

B69. Derivatives book and downgrade triggers

• Asks: What is the size of the derivatives book (notional, receivables, collateral posted), and do contracts require collateral posting on a credit downgrade?
• Good: Gross derivative receivables small versus equity and no material rating-trigger collateral clauses; a large uncollateralized receivables pile or downgrade triggers is a hard fail — "financial weapons of mass destruction."
• Source: 2002 Berkshire letter (derivatives). (quote-backed)

B70. Currency-matched financing

• Asks: Are foreign-currency assets and earnings matched by same-currency liabilities, so the position embeds no implicit FX forecast?
• Good: Debt by currency versus earnings/assets by currency shows material matching (Berkshire funded its Japanese stakes with yen bonds); the thesis requires no view on exchange-rate direction.
• Source: 2023 Berkshire letter (Japan trading houses; yen financing); 2020 letter. (quote-backed)

Management

B71. Management candor

• Asks: Reading the last 3–5 years of shareholder letters, management discussion, and mid-term plans: does management report failures as candidly as wins, in plain language, with consistent metrics year to year?
• Good: At least one plainly acknowledged, costed mistake (the "confession time" standard) and targets that persist across years rather than quietly resetting; boilerplate that never admits anything, or a mid-term plan that vanishes when missed, is the failing pattern. Owner's Manual bar: "tell you the business facts that we would want to know if our positions were reversed."
• Source: Owner's Manual principle 12; 1983 letter (owner-related principles); 2007 letter (unforced errors). (quote-backed)

B72. One-dollar test on retained earnings

• Asks: For every ¥1 of earnings retained over the past 5–10 years (rolling windows; cumulative net income minus dividends and buybacks), has at least ¥1 of market value been created — and can you trace where the retained and freed cash actually went?
• Good: Change in market cap (smoothed to dampen market noise, adjusted for buybacks/issuance) ≥ cumulative retained earnings, plus a sources-and-uses of freed capital; the Sanborn failure mode — a growing idle hoard while the core declines and owners get nothing — named if present. A cash-hoarding balance sheet with sideways market value is a documented failure.
• Source: 1984 Berkshire letter ("for every dollar retained ... at least one dollar of market value"); Owner's Manual principle 9; Partnership letter, January 30, 1961 (Sanborn). (quote-backed)

B73. Buyback discipline: the two conditions

• Asks: Do repurchases meet both conditions — funds genuinely surplus to operating and liquidity needs, and a price below conservatively calculated intrinsic value — with the accretion math computed rather than asserted?
• Good: Buyback history mapped against your own value range: repurchases cluster in discount periods (the Sanborn tender is the model), pause when the stock is expensive, and are funded from surplus rather than debt; disclosures let continuing holders estimate that value.
• Source: 1999 Berkshire letter ("only one combination of facts"); 2011 letter; 1984 letter; Partnership letter, January 30, 1961 (Sanborn resolution). (quote-backed)

B74. Buyback motives and share-count trajectory

• Asks: Are buybacks justified as "supporting the stock" or offsetting option issuance — and has the fully diluted share count actually fallen over the decade, net of stock compensation?
• Good: Stated rationale is price-versus-value; 10-year diluted count CAGR ≤ 0%; buying high while issuing low fails — "Buying dollar bills for $1.10 is not good business for those who stick around."
• Source: 1999 Berkshire letter (share repurchases); 2022–23 letters (value-accretive repurchases; Japan share reduction). (quote-backed)

B75. Share issuance discipline

• Asks: Has management issued shares only when receiving as much intrinsic business value as it gave — and is "non-dilutive to EPS" ever offered as the test?
• Good: Share count flat or shrinking over a decade with every issuance showing demonstrably fair value received; accretive-but-value-destroying deals flagged — "we will not issue shares unless we receive as much intrinsic business value as we give."
• Source: 1982 Berkshire letter. (quote-backed)

B76. Acquisition record: toads and kisses

• Asks: What premiums has management paid in past acquisitions, and did the realized results justify them?
• Good: A deal table — price paid, premium over standalone value, subsequent segment earnings and impairments; a record of full-priced deals followed by write-downs or divestitures fails — "those kisses had better pack some real dynamite."
• Source: 1981 Berkshire letter; 2019 letter ("just be sure you buy good businesses"). (quote-backed)

B77. Institutional imperative resistance

• Asks: Do capital decisions show independent reasoning — or imitation of peers, projects materializing to soak up available funds, and staff studies produced to justify the leader's cravings?
• Good: Documented counter-examples: cancelled projects, capital returned when opportunities were poor, cash allowed to build at cycle tops ("painful ... but not as painful as doing something stupid"), no me-too diversification into fashionable sectors.
• Source: 1989 Berkshire letter (the four imperative symptoms); 2003 letter (cash patience). (quote-backed)

B78. Owners eat their own cooking

• Asks: Do executives and directors hold stock that is large relative to their pay and bought or held rather than merely granted — and does any board group represent interests adverse to the common shareholder?
• Good: Insider ownership worth multiple years' compensation with open-market purchases on record and no chronic grant-and-sell pattern; the Sanborn board — customer-industry representatives holding token shares — is the automatic red flag. "We eat our own cooking."
• Source: Owner's Manual principle 2; 1983 letter; Partnership letter, January 30, 1961 (Sanborn board). (quote-backed)

B79. Board that can do its one job

• Asks: Is the board capable of its bedrock task — finding and retaining a talented CEO with integrity — with directors economically independent enough to challenge management, and an audit committee that puts the auditor on the spot?
• Good: Directors with real operating or ownership credentials whose fees are not life-changing income; evidence of action (CEO changes, blocked deals) rather than a cocker-spaniel record; non-audit fees small, no auditor-shopping, and none of the period-moving behavior Buffett's four audit-committee questions target.
• Source: 2019 Berkshire letter (Boards of Directors); 2002 letter (The Audit Committee). (quote-backed)

B80. Hurdle-based incentive alignment

• Asks: Does compensation resemble Buffett's own fee deal — meaningful pay only above a performance hurdle, with substantial personal capital invested — or fixed rewards regardless of results?
• Good: Comp tied to per-share results above a hurdle with real downside and large co-investment; the BPL model — paid only above a 6% threshold, deficiencies made up before taking any share, virtually his entire net worth in the partnership.
• Source: BPL ground rules and fee structure, 1957–62 letters; 1985 Berkshire letter (compensation). (teaching-derived)

B81. Options and the ratchet

• Asks: Do fixed-price, long-dated option grants let executives profit merely from retaining earnings — and is pay set by peer-percentile benchmarking that ratchets regardless of per-share results?
• Good: Equity awards carry strike escalation, cost-of-capital hurdles, or genuine performance conditions, with no repricing; a proxy built on consultant-driven peer tables fails — the "Ratchet, Ratchet and Bingo" pattern where yesterday's excess becomes today's baseline, and the Fred Futile pattern where options capture retained earnings, not created value.
• Source: 2005 Berkshire letter (CEO pay; Fred Futile); 2006 letter. (quote-backed)

B82. Per-share yardstick and pay modesty

• Asks: Does management measure itself — in targets, comp metrics, and reporting — on per-share value growth rather than company size, and is total executive pay modest relative to earnings?
• Good: Incentive metrics per-share or return-based, with top-executive comp small versus operating income (the Japanese trading-house standard); when owners took dividend cuts, insider pay took cuts too — Sanborn's five dividend cuts in eight years with fees untouched fails. "We measure by per-share progress."
• Source: Owner's Manual principle 3; 2023 letter (sōsha compensation); Partnership letter, January 30, 1961 (Sanborn fees). (quote-backed)

B83. Asset-conversion operator evidence (Harry Bottle test)

• Asks: Is there filed evidence that management or a control party actually converts underperforming assets to cash — completed divestitures, inventory reductions, plant closures with dates and proceeds — rather than announcements?
• Good: Sources-and-uses showing realized proceeds, with inventory and receivables visibly reduced in subsequent balance sheets, as Harry Bottle did at Dempster within a year — the conversion, not the cheapness, produced the profit.
• Source: Partnership letter, January 18, 1963 (Dempster). (teaching-derived)

B84. Blocker map

• Asks: Who can block value realization — controlling holder, family faction, board group — and what are their personal incentives? Is there evidence the blockage is ending (age, estate, stake sale, board turnover)?
• Good: Named parties with stakes and motives from the proxy, plus a dated judgment on whether the blocker persists; Commonwealth's obviously sensible merger had been blocked "for personal reasons" with evidence the situation would not last.
• Source: Partnership letter, February 1959 (Commonwealth Trust). (teaching-derived)

B85. Management in place

• Asks: Is capable management already in place, so the thesis does not depend on importing new talent?
• Good: Operating leadership tenured through the record being relied upon, no key-person vacancy, succession evidenced rather than promised — "management in place (we can't supply it)."
• Source: 1982 Berkshire letter (acquisition criteria). (quote-backed)

B86. Rational dividend-retention split

• Asks: Is the split between dividends and retained earnings driven by reinvestment returns — retaining only what compounds, paying out what cannot?
• Good: Payout ratio linked to demonstrated incremental returns (the sōsha model: ~1/3 payout with retained funds building businesses and funding buybacks); no borrowing to sustain a dividend.
• Source: 2023 Berkshire letter (Japan trading houses); 1984 letter (dividend policy). (quote-backed)

B87. Widening the moat vs hitting the quarter

• Asks: Is there concrete evidence management sacrifices short-term earnings to widen the moat — pricing, service, brand, or capacity investments — rather than cutting them to hit targets?
• Good: At least one documented episode of knowingly depressed margins to strengthen cost position, customer satisfaction, or brand; cuts into customer-facing spend during soft quarters fail — "when short-term and long-term conflict, widening the moat must take precedence."
• Source: 2005 Berkshire letter. (quote-backed)

B88. Underwriting and lending discipline (financials)

• Asks: For insurers and lenders: does management walk away from mispriced business — premium volume shrinking in soft markets, loans requiring real down payments and sized against verified actual (not hoped-for) income?
• Good: Volume voluntarily declining in industry soft years with combined ratios or credit losses better than industry across the cycle; the Clayton pattern — standards, not credit scores, driving performance. Growth precisely when industry pricing weakens fails.
• Source: 2001 Berkshire letter (underwriting principles); 2008 letter (Clayton Homes); 1987 letter. (quote-backed)

Valuation

B89. Graham bargain check (net-net and net cash)

• Asks: Compute net current asset value (current assets minus all liabilities and prior claims) and net cash (cash plus marketable securities minus total liabilities) per share, and compare each to the share price. Is the company a net-net, or trading near hard asset value?
• Good: Line-by-line NCAV from the reconciled balance sheet with preferreds, minorities, and tax liabilities deducted; price at or below ~two-thirds of NCAV is the classic bargain (these still exist in Japan), and price below net cash for a business not burning it is stronger still. Failing is not disqualifying — the case must then rest entirely on the owner-earnings yield (B42).
• Source: Graham's bargain-issue rule (Security Analysis / The Intelligent Investor), as practiced in BPL generals 1957–66. (teaching-derived)

B90. Haircut liquidation value (Dempster method)

• Asks: Rebuild liquidation value with per-line haircuts — cash ~100%, receivables ~85%, inventory ~50–65%, prepaids and fixed assets at prompt-sale values. Is the price below the haircut total per share?
• Good: A per-line haircut table shown next to stated book, with the conclusion drawn from the haircut total, never book alone; Dempster's book was $75 per share against ~$35 of prompt-realization value.
• Source: Partnership letters, January 24, 1962 (Dempster appendix) and January 18, 1963 (valuation tables). (teaching-derived)

B91. Private-owner whole-business valuation

• Asks: What would an informed private buyer pay in cash for the whole company (asset values plus sustainable earnings power) — and is the stock available at a large discount to that appraisal? Write the memo as a business-picker, with the ticker never mentioned.
• Good: An explicit whole-company number with its build-up shown and a discount of one-third or more, sanity-checked against private-market transactions for comparable businesses. "Investment is most intelligent when it is most businesslike."
• Source: Partnership letter, January 18, 1964 (private-owner basis); Graham, The Intelligent Investor ch. 20; 2022 Berkshire letter ("we are business-pickers"). (quote-backed)

B92. Hidden separable assets (Sanborn test)

• Asks: List and value every separable non-operating asset per share from the filings — excess cash beyond a stated operating requirement, securities at market, real estate at appraisal, overfunded pension. After subtracting them from the price, what are you paying for the operating business?
• Good: An itemized per-share table with a source note for each line and the implied price of the operating business stated explicitly; Sanborn-class answer: the operating business at a negative implied price ($45 stock against a ~$65/share portfolio).
• Source: Partnership letter, January 30, 1961 (Sanborn Map). (teaching-derived)

B93. Margin-of-safety width

• Asks: State intrinsic value computed on a conservative basis — trough assumptions, with "conservative" defined line by line — and the current price. Is the discount at least one-third of that conservative value?
• Good: Commonwealth-scale gap (bought ~$50 against a $125 conservative value); the base case still exceeds price meaningfully with terminal assumptions at or below historical averages. "Confronted with a challenge to distill the secret of sound investment into three words, we venture the motto, MARGIN OF SAFETY."
• Source: Graham, The Intelligent Investor ch. 20; Partnership letter, February 1959 (Commonwealth); 2018 Berkshire letter (sensible prices). (quote-backed)

B94. Range of values, never pseudo-precision

• Asks: Does the valuation produce a conservatively derived range — with the buy price below the bottom of it — rather than a single point estimate with decimals?
• Good: Explicit low/high intrinsic-value figures with the drivers of each end stated; a single precise target price is itself a process fail — intrinsic value is estimable "only when we employ a range of values, rather than some pseudo-precise figure."
• Source: 1999 Berkshire letter (share repurchases). (quote-backed)

B95. Relative undervaluation vs matched comparables

• Asks: If the claim is "relatively undervalued," compute the multiple gap against 3–5 named peers of the same general quality.
• Good: A peer table with quality-matching justified (returns, balance sheet, growth) and a quantified gap — not "cheaper than the market."
• Source: Partnership letter, January 18, 1965 ("Generals — Relatively Undervalued"). (teaching-derived)

B96. Workout arithmetic

• Asks: If the position depends on a corporate event (merger, tender, liquidation, spin-off), compute expected proceeds, the timetable, the annualized return, and enumerate what could break the deal.
• Good: Gross spread, expected closing date, annualized net return, and named applecart-upsetters each with a probability judgment — "securities with a timetable where we can predict, within reasonable error limits, when we will get how much and what might upset the applecart."
• Source: Partnership letter, January 24, 1962 (work-outs definition). (quote-backed)

B97. Purchase price does the work

• Asks: Does the return still clear the hurdle if the exit happens at a mediocre price — say 75–80% of appraised value — rather than at full intrinsic value?
• Good: IRR recomputed at a mediocre-sale exit and still acceptable; no thesis that only works if a strategic buyer pays full value. "Never count on making a good sale. Have the purchase price be so attractive that even a mediocre sale gives good results."
• Source: Partnership letter, February 1959 (Commonwealth). (quote-backed)

B98. Baseball-bat obviousness

• Asks: Is the quantitative cheapness visible in ≤5 lines of arithmetic from filed numbers, or does it need a multi-stage model with optimistic inputs to appear?
• Good: Cheapness demonstrable directly from the reconciled figure table; if a DCF with growth assumptions is needed to make it look cheap, it is not this kind of idea — "the figures should hit you over the head with a baseball bat."
• Source: Partnership letter, October 9, 1967. (quote-backed)

B99. Intrinsic value as discounted cash (Aesop's three questions)

• Asks: Value the business as "the discounted value of the cash that can be taken out of a business during its remaining life": how certain are you the birds are in the bush, when will they emerge and how many, discounted at the long-term government bond rate?
• Good: A written owner-earnings-based valuation with explicit certainty grading, timing, and arithmetic shown, comps used only as a sanity check; where certainty can't honestly be graded, too-hard rather than a hopeful multiple.
• Source: Owner's Manual (Intrinsic Value); 2000 Berkshire letter (Aesop); 1992 letter (John Burr Williams). (quote-backed)

B100. Growth only counts at high returns

• Asks: Does the growth in the valuation come from capital deployed at returns above the discount rate — or is a premium being paid for growth that consumes value?
• Good: Growth assumptions tied to the computed incremental-return figure (B33); growth funded at sub-threshold returns valued at zero or negative — "Growth is simply a component — usually a plus, sometimes a minus — in the value equation."
• Source: 2000 Berkshire letter (growth vs value); 1992 letter. (quote-backed)

B101. Price paid for economic goodwill

• Asks: How much is being paid above net tangible assets, and is that premium justified by the size and durability of above-market returns on those tangible assets?
• Good: A reconciled table — market price vs net tangible assets vs capitalized excess earnings — with the implied payback period on the goodwill premium computed and defensible given the moat evidence (B26, B28).
• Source: 1983 Berkshire letter, goodwill appendix. (teaching-derived)

B102. Wonderful at fair beats fair at wonderful

• Asks: Is this a wonderful business at a fair price, rather than a fair business being rationalized because the price looks wonderful?
• Good: The quality gates pass before price is considered, and price is then merely fair — owner-earnings yield at or better than the long bond with growth as the margin; the discount is not the thesis. "It's far better to buy a wonderful company at a fair price than a fair company at a wonderful price."
• Source: 1989 Berkshire letter; 2023 letter (Munger as architect). (quote-backed)

B103. One-puff check: cigar butts need a catalyst

• Asks: Is the only attraction statistical cheapness — a one-puff bargain in a poor business?
• Good: Either durable-economics reasons independent of the discount, or a hard near-term realization catalyst (B39); within this bargain-first profile, cheapness alone — with no catalyst and no adequately earning business — fails.
• Source: 1989 Berkshire letter ("Mistakes of the First Twenty-five Years"). (quote-backed)

B104. Sum-of-parts: groves and two columns

• Asks: For holdcos, insurers, and conglomerates: value the investments per share and the operating earnings ex-investment income as two separate columns — and can each major segment be appraised with reasonable accuracy?
• Good: A reconciled two-column (or per-grove) table that sums, with corporate costs and net debt, to a whole-company appraisal compared against price; no black-box segment material to the total, and no double-counting of float-funded assets.
• Source: 2008 Berkshire letter (Yardsticks); 2018 letter (five groves). (quote-backed)

Verdict guidance

The circle-of-competence gate comes first: if B1 fails (or B2's ten-year estimate can't honestly be written), the verdict is too-hard and nothing else is computed — this exit is a feature, not a failure. If the business is understandable, the earnings-quality core — B42 (owner earnings), B43 (ten-year stability), and B47/B72 (retention actually working) — establishes whether earnings power is real, stable, and not being destroyed by retention; B71 (candor) and the rest of the management section check whether the people reporting the numbers can be believed. A business that clears B1 but stumbles on stability or candor is at best a watch.

Price decides between watch and buy-below-¥X. The buy-below level is derived one of two ways: from owner earnings — the price at which the owner-earnings yield (B42) reaches the required threshold, with a margin of safety on the maintenance-capex estimate and the discount width of B93 — or from assets, at two-thirds of NCAV per share (B89, cross-checked by the haircut and hidden-asset tables of B90 and B92), whichever the case actually rests on. An asset case must also survive the wait (B61, B65) and name its catalyst where the business is melting (B39, B103). Pass is reserved for companies inside the circle where both the earnings case and the asset case fall short at any plausible price.

Archive-depth rule. When the archived sources are shallower than an item requires (fewer years, missing competitor/customer filings, absent disclosures), the item is answered data-insufficient with the missing source named — never estimated from general knowledge without an explicit [general-knowledge: not-from-archive] flag. Load-bearing items stuck at data-insufficient push the verdict toward too-hard or watch, not toward a guess.

A too-hard verdict names the specific item ID(s) that triggered it. Item-level answers use pass / fail / data-insufficient / practitioner-pending; items after a too-hard early exit are marked not-evaluated, never silently skipped.