Scores are algorithm-generated research tools, not investment recommendations.
Buffett-Style Quality Assessment
By Maksym Lytvynov · Rules as of (v23)
Warren Buffett's own tests from his Berkshire Hathaway letters — returns on equity and tangible capital, owner earnings, debt, retained earnings and price — computed from ten years of annual filings.
Score output: Strong / Attractive / Neutral / Reject / Insufficient data, with a 0–10 score
What is the Buffett Score?
The Buffett score checks a company against criteria Warren Buffett wrote down in his shareholder letters between 1977 and 1992, and the quality leg of the Buffett's Alpha paper (Frazzini, Kabiller & Pedersen, 2018). Each criterion passes or fails against a stated number, and the score is the share passed out of the criteria the filings let us evaluate. One of them is "moat signals in the numbers": not a judgement of brand or competitive position, which no filing records, but whether returns on tangible capital and margins have held up for a decade. It is one of two models on the Quality axis.
How does the Buffett Score work?
Returns. The 1987 letter judges a business by its return on equity: ROE on average equity ≥ 15% in every year and ≥ 20% on average over 10 years (judged on at least 5). The 1983 letter adds the return on the tangible capital a business actually needs: after-tax operating profit ÷ tangible capital (equity + debt − cash − goodwill − intangibles) ≥ 15% on average.
Moat signals in the numbers: return on tangible capital ≥ 15% in at least 80% of the years, and gross (else operating) margin with a coefficient of variation ≤ 0.15 across the window. This measures durability that shows up in filings. Brand, customer loyalty and competitive position are not in any filing, and the model does not score them.
Owner earnings and debt. Owner earnings follow the 1986 letter — net income plus depreciation and amortization, less capital spending; stock-based pay stays the expense it is. The criteria: owner earnings (net income + D&A − capex; stock pay stays an expense) positive every year, and the latest 3-year average above the first; and total debt ≤ 3 × owner earnings, and equity positive. Negative equity is never read as low debt.
Capital allocation and price: market value added ≥ $1 for each $1 of earnings retained over 5 years; diluted share count not higher than 5 years earlier; and price ≤ 75% of intrinsic value (owner earnings per share, 10 years at the company's own growth held to -5%…10%, then 3% forever, discounted at 10%).
Banks run on the 1990 letter's adaptation: owner earnings are reported earnings, return on average assets ≥ 1.25% on average and total assets ≤ 20 × equity replace the debt test, and the tangible-capital and margin criteria do not apply. Insurers run on an adaptation from the 2007 and 2009 letters, which judge an insurer by its underwriting and its per-share book value: total benefits, losses and expenses ÷ net premiums earned ≤ 100% summed over 10 years (judged on at least 5) — an upper bound on the combined ratio; not applicable when premiums are below 75% of total revenues; ROE on average equity ≥ 10% on average over 10 years and no year below 0% (judged on at least 5); book value per share, with dividends paid added back, compounded ≥ 10% a year over up to 10 years (at least 5); and equity positive and net premiums earned ≤ 3 × equity, alongside the retained-dollar, share-count and margin-of-safety tests.
What does Buffett look for in a stock?
Each criterion as the scoring code evaluates it, with the threshold it uses and where that threshold comes from.
Return on equity ≥ 15% every year and ≥ 20% on average
- Passes when
- ROE on average equity ≥ 15% in every year and ≥ 20% on average over 10 years (judged on at least 5)
- Applies to
- General, Bank, Asset manager, Holding company, Utility, REIT
- Source
- Berkshire letter 1987 ("an average return on equity over 20% … and no year worse than 15%")
Return on tangible capital ≥ 15% on average
- Passes when
- after-tax operating profit ÷ tangible capital (equity + debt − cash − goodwill − intangibles) ≥ 15% on average
- Applies to
- General, Asset manager, Holding company, Utility, REIT
- Source
- Berkshire letter 1983 (economic goodwill: excess return on net tangible assets)
Moat signals in the numbers: durable returns and stable margins
- Passes when
- return on tangible capital ≥ 15% in at least 80% of the years, and gross (else operating) margin with a coefficient of variation ≤ 0.15 across the window
- Applies to
- General, Asset manager, Holding company, Utility, REIT
- Source
- Frazzini, Kabiller & Pedersen (2018), Financial Analysts Journal 74(4); Asness, Frazzini & Pedersen (2019), Quality Minus Junk. The 0.15 cut is ours
Owner earnings positive every year and growing
- Passes when
- owner earnings (net income + D&A − capex; stock pay stays an expense) positive every year, and the latest 3-year average above the first
- Applies to
- General, Bank, Asset manager, Holding company, Utility, REIT
- Source
- Berkshire letter 1986 (owner earnings)
Debt payable from three years of owner earnings, equity positive
- Passes when
- total debt ≤ 3 × owner earnings, and equity positive
- Applies to
- General, Asset manager, Holding company, Utility, REIT
- Source
- Berkshire letter 1987 ("Really good businesses usually don't need to borrow"); the 3-year limit is the strict end of Buffettology (1997)
Each $1 retained added at least $1 of market value (5 years)
- Passes when
- market value added ≥ $1 for each $1 of earnings retained over 5 years
- Applies to
- Every profile the model rates
- Source
- Berkshire letter 1983 (the $1 test, "on a five-year rolling basis")
Diluted share count not higher than five years ago
- Passes when
- diluted share count not higher than 5 years earlier
- Applies to
- Every profile the model rates
- Source
- Berkshire letters 1984 (buybacks) and 1992 (options are compensation)
Price at or below 75% of intrinsic value
- Passes when
- price ≤ 75% of intrinsic value (owner earnings per share, 10 years at the company's own growth held to -5%…10%, then 3% forever, discounted at 10%)
- Applies to
- Every profile the model rates
- Source
- Berkshire letter 1992 (value = discounted cash flows); margin of safety: Graham, The Intelligent Investor, ch. 20
Return on assets ≥ 1.25% on average (1990 letter)
- Passes when
- return on average assets ≥ 1.25% on average
- Applies to
- Bank
- Source
- Berkshire letter 1990 (Wells Fargo: "1.25% on assets")
Assets no more than twenty times equity (1990 letter)
- Passes when
- total assets ≤ 20 × equity
- Applies to
- Bank
- Source
- Berkshire letter 1990 (banks carrying "assets twenty times equity")
Premiums covered losses and expenses (combined ratio ≤ 100%)
- Passes when
- total benefits, losses and expenses ÷ net premiums earned ≤ 100% summed over 10 years (judged on at least 5) — an upper bound on the combined ratio; not applicable when premiums are below 75% of total revenues
- Applies to
- Insurer
- Source
- Berkshire letters 2007 (float is free when "premiums we receive equal the losses and expenses we incur") and 2009 (underwriting profit); the premium-share cut is ours
Return on equity ≥ 10% on average, no loss year
- Passes when
- ROE on average equity ≥ 10% on average over 10 years and no year below 0% (judged on at least 5)
- Applies to
- Insurer
- Source
- Berkshire letter 2009 (insurers' returns on tangible equity fall short of the S&P 500's); the floor follows the 1987 letter's "no year worse than" form
Book value per share (plus dividends) compounded ≥ 10% a year
- Passes when
- book value per share, with dividends paid added back, compounded ≥ 10% a year over up to 10 years (at least 5)
- Applies to
- Insurer
- Source
- Berkshire letters 2007 and 2009 (per-share book value, the yardstick opening each letter)
Equity positive, premiums ≤ 3× equity (NAIC IRIS ratio 2)
- Passes when
- equity positive and net premiums earned ≤ 3 × equity
- Applies to
- Insurer
- Source
- NAIC IRIS ratio 2 (net premiums written to policyholders' surplus, usual range under 300%), proxied by premiums earned over GAAP equity
- Insurers — adaptation — Berkshire letters (insurance): 2007 and 2009 letters on underwriting profit and float; per-share book value, the yardstick opening each letter (2009: "a crude proxy for it: per-share book value"); NAIC IRIS ratio 2. Buffett judges an insurer by whether underwriting is profitable, so the float costs nothing, and by per-share book value. Filings carry no combined ratio, so total benefits, losses and expenses over net premiums earned stands in: it counts every expense, so it can miss an underwriting profit, never invent one. Retained-dollar, share-count and margin-of-safety criteria are kept, the intrinsic value discounting reported earnings.
- "Moat signals in the numbers" measures what filings can show — how durable the return on tangible capital and the margins have been over ten years. Brand, switching costs, network effects and management are not in any filing, and the model does not score them.
How the Buffett score maps to a status
| Status | Rule |
|---|---|
| Strong | ≥ 85% of evaluable criteria passed |
| Attractive | ≥ 65% |
| Neutral | ≥ 40% |
| Reject | below 40% |
| Insufficient data | fewer than 5 criteria evaluable |
The bands are ours: with eight criteria, Strong allows one miss and Attractive two.
Which companies Buffett rates
| Sector profile | Rated | Note |
|---|---|---|
| General | Yes | — |
| Bank | Yes | Bank adaptation (1990 letter): owner earnings are reported earnings, leverage replaces the debt test, and the tangible-capital and margin criteria do not apply. |
| Insurer | Yes | Insurer adaptation (2007 and 2009 letters): judged on underwriting, returns and per-share book value; reported earnings stand in for owner earnings. |
| Asset manager | Yes | — |
| Holding company | Yes | — |
| Utility | Yes | — |
| REIT | Yes | — |
“Not applicable” means the model is left out of that company's axis — never scored as zero or as neutral.
How AlphaStocks uses Buffett
Inside the Quality axis the weights are Piotroski F-Score 35%, Buffett Quality Criteria 65% (general profile; the sector profiles shift them — see the methodology page). Only the models that apply to a company and have a score are averaged, re-weighted among themselves; a model that does not apply is left out, never counted as zero. The Quality axis carries 40% of the composite.
Buffett looks across ten years; Piotroski, the other Quality model, compares the latest fiscal year with the one before. A company can pass most of Buffett's tests and still have a weak year on Piotroski, or the reverse.
Quality = weighted mean of the applicable Quality models (Buffett 65%); composite weight of Quality: 40%Current top stocks by Buffett
The Buffett-Style Quality Assessment model is evaluated for every stock in our coverage it applies to. You can filter the full rankings table to see which stocks currently score highest on this model and compare them across all five investment frameworks.
Frequently Asked Questions
What does the Buffett score measure on AlphaStocks?
Criteria from Warren Buffett's shareholder letters, each a pass or a fail: ROE on average equity ≥ 15% in every year and ≥ 20% on average over 10 years (judged on at least 5); after-tax operating profit ÷ tangible capital (equity + debt − cash − goodwill − intangibles) ≥ 15% on average; return on tangible capital ≥ 15% in at least 80% of the years, and gross (else operating) margin with a coefficient of variation ≤ 0.15 across the window; owner earnings (net income + D&A − capex; stock pay stays an expense) positive every year, and the latest 3-year average above the first; total debt ≤ 3 × owner earnings, and equity positive; market value added ≥ $1 for each $1 of earnings retained over 5 years; diluted share count not higher than 5 years earlier; price ≤ 75% of intrinsic value (owner earnings per share, 10 years at the company's own growth held to -5%…10%, then 3% forever, discounted at 10%). The score is the share passed out of the criteria that can be evaluated.
What are "moat signals in the numbers"?
A measurable stand-in for a moat: return on tangible capital ≥ 15% in at least 80% of the years, and gross (else operating) margin with a coefficient of variation ≤ 0.15 across the window. It shows whether high returns and margins have lasted, which is what a moat produces. It does not score brand, switching costs or network effects — nothing in a filing measures them.
How does the Buffett model treat banks and insurers?
Banks use the 1990 letter's tests: return on average assets ≥ 1.25% on average and total assets ≤ 20 × equity, alongside the return-on-equity test. Insurers are an adaptation labelled as one, from the 2007 and 2009 letters: underwriting (total benefits, losses and expenses ÷ net premiums earned ≤ 100% summed over 10 years (judged on at least 5) — an upper bound on the combined ratio; not applicable when premiums are below 75% of total revenues), ROE on average equity ≥ 10% on average over 10 years and no year below 0% (judged on at least 5), book value per share (book value per share, with dividends paid added back, compounded ≥ 10% a year over up to 10 years (at least 5)) and equity positive and net premiums earned ≤ 3 × equity.
Why does a company with negative equity fail the debt test?
A company that has bought back more stock than its book equity (McDonald's is one) has negative shareholders' equity. A debt-to-equity or debt-to-earnings ratio over a negative base is undefined, and the model treats undefined as a fail — never as low debt.
What is the difference between the Buffett score and the Piotroski F-Score?
Buffett looks at ten years of returns, owner earnings, debt and capital allocation. Piotroski compares nine accounting signals in the latest fiscal year with the year before. Together they form the Quality axis.
See all 1,500+ stocks scored
Every stock evaluated by Buffett-Style Quality Assessment and four other investment models, from SEC filings. Changed in v23? See what changed and why.
Explore the rankingsExplore Other Models
Model names reference the published investment methodologies of their respective authors. AlphaStocks is not affiliated with, endorsed by, or sponsored by Warren Buffett or any related entities. Our implementation follows the cited sources; where it adapts them, the adaptation is named above.
AlphaStocks provides investment research and analysis, not investment advice. Past performance does not guarantee future results. Always do your own due diligence before making investment decisions. Data sourced from SEC EDGAR and Alpaca Markets.