The Buffett-Fit Score is a 0 to 100 number invest-like produces for every US-listed stock to grade how closely it fits Warren Buffett's documented value-investing criteria. Five pillars, equal weight, each pillar carries its own AI-generated reasoning anchored in current fundamentals.
Last reviewed:
The Buffett-Fit Score grades a stock from 0 to 100 across five pillars Warren Buffett has discussed across 60+ years of Berkshire Hathaway shareholder letters: economic moat, durability of demand, owner-friendly management, financial health, and valuation. An LLM rubric produces a sub-score and a paragraph of reasoning for each pillar using current fundamentals from Financial Modeling Prep and retrieved passages from Berkshire letters. The five sub-scores blend (equal weight) into the headline 0 to 100 number, which then maps to an A+ to D grade. Educational fit assessment only, not a forecast or recommendation.
Each pillar is independent. A stock cannot get past 70 on the headline score by hitting one pillar alone, and a strong moat does not paper over a stretched valuation.
Whether the business has a structural advantage that lets it earn above-average returns over time without being competed away. Quantitative signals: ROIC above 15% over a multi-year window, gross margin meaningfully above peers, evidence of brand, regulatory, network-effect, or switching-cost protection.
Whether the product or service is the kind of thing people will still need in 10 or 20 years. Tested with revenue trajectory across at least one full downturn, recurring versus discretionary demand, and customer concentration that does not depend on a single counterparty.
Whether the people running the business behave like long-term owners or like compensation-maximising agents. Signals: insider ownership, buybacks executed at low prices rather than high, plain-English shareholder communication, executive compensation not skewed to one-year stock targets.
Whether the balance sheet is robust enough that a bad year does not force the business to issue equity or sell assets at the wrong time. Net debt to EBITDA below sector median (typically under 2x), free-cash-flow margin above 10%, interest coverage that holds up at peak rates.
Whether the current price gives a margin of safety relative to a defensible estimate of intrinsic value. P/E and EV/EBITDA in a reasonable range for the quality on offer, owner-earnings yield comparable to the 10y Treasury plus an equity-risk premium, DCF and reverse-DCF producing overlapping intrinsic-value bands.
The 0 to 100 score maps to five letter grades. Same thresholds across the verdict pages, the track-record log, the weekly newsletter, and the public API.
85 to 100
Textbook Buffett-fit. Strong moat, clean balance sheet, fair price, all five pillars passed convincingly.
70 to 84
Strong fit. Multiple pillars pass and valuation is either fair or only mildly stretched.
55 to 69
Partial fit. The business is interesting; usually one pillar (often valuation or management) holds it back.
40 to 54
Weak fit. Several pillars fail, but the failures are specific and worth understanding rather than a simple sell signal.
0 to 39
Not a Buffett-style business at this price. Pillar failures are systemic; the stock may still be a fine pick under another framework.
Each pillar gets its own paragraph of reasoning, not just a number. The LLM rubric ingests three inputs: the company's current fundamentals (refreshed daily from Financial Modeling Prep), the latest 10-K filing, and retrieved passages from the Berkshire Hathaway shareholder-letter corpus spanning the last quarter century. The rubric instructs the model to test the company against the documented criterion for that pillar, score on the 0 to 100 scale, and explain in plain English which signals supported or contradicted the score.
The exact prompt and rubric version used to produce any given verdict are stored alongside the verdict in the database. A future regeneration with a different prompt produces a separately-stored row instead of overwriting the original. This is what makes the public accuracy log meaningful: you can trace any number back to the rubric that produced it.
Every Monday: the stocks that newly pass 5 or more of the 7 legendary-investor frameworks (Buffett, Graham, Lynch, Greenblatt, Munger, Fisher, Smith), plus every grade change that matters. The 5-year track record of this screen is published openly. One email a week, unsubscribe anytime.
The fastest way to understand the score is to read one real verdict, with all five pillars graded against live data.
Educational only. invest-like is not a registered investment adviser; nothing here is personalised investment advice. Always do your own research and consider your individual circumstances.