How Mr. Market values a business.
Eight steps. Owner Earnings discounted at the long Treasury, faded to GDP, with a 25% margin of safety. The same protocol is applied identically to every business.
A business is worth the cash it will produce for its owners over the rest of its life, discounted back to today at an appropriate rate. That sentence — written in essence by John Burr Williams in 1938, repeated by Buffett in 1992 — is the entire theory of value. Everything that follows is mechanics. Mr. Market's job is to mechanize the theory consistently, the same way for every business, so that the resulting numbers can be compared honestly to the prices Mr. Market quotes each day.
Data prep & validation.
Garbage in, garbage out. Before any calculation runs, we confirm we have what the protocol requires: operating cash flow greater than zero in the latest annual filing, depreciation & amortization greater than zero (or a sector exception that legitimately skips it — banks, insurance carriers, and REITs), shares outstanding greater than zero, and at least three years of forward analyst EPS estimates on file. A business that fails any of these gets a warning banner on its Quote page naming the specific gap, but the math still runs on the inputs we do have so subscribers can see the partial picture rather than a blank tile. A reckoning built on incomplete data is labelled as such — never silently inflated to hide the gap.
Owner Earnings = OCF − D&A.
Owner Earnings is the foundation. It represents the cash a business generates after spending what is necessary to maintain its current productive capacity — not grow it, just sustain it. This is different from Free Cash Flow, which subtracts all capital expenditure including growth investments and therefore penalizes companies that are actively reinvesting in expansion.
We use D&A as the proxy for maintenance CapEx. It is an approximation, but it is consistent and automatable at scale. For Banks, Insurance and REITs the protocol uses Operating Cash Flow only — the economics of D&A are structurally different in those businesses and subtracting it produces a distorted picture.
G1 (analyst consensus, capped) and G2 (GDP), faded linearly.
The valuation uses a two-stage growth structure. The first stage — G1 — reflects how fast the business is expected to grow in the near term, based on a consensus of analyst EPS forecasts over a five-year forward window. We average the five year-over-year growth rates and clip the result to [0%, 20%]. Using a multi-year consensus rather than a single year-over-year figure produces a more stable input — one outlier year cannot skew the entire valuation.
The second stage — G2 — reflects the long-run sustainable growth rate the business can maintain in perpetuity, anchored to U.S. real GDP growth, since no company can grow faster than the overall economy forever. G2 typically lands between 0% and 3%.
Between the two stages, growth fades linearly across years 1 to 11. Growth Rate for Year n = G1 + (G2 − G1) × (n − 1) / 10.
Project Owner Earnings, year 1 through year 11.
Apply the linear-fade growth schedule to Owner Earnings. Year 1 = OE × (1 + G1). Year n = Year (n−1) × (1 + the growth rate for year n). Continue through year 11, which is calculated only as the input to the terminal value formula in step V — it does not enter the discounting sum directly.
Discount rate r = 10-year U.S. Treasury yield + sector adjustment.
The discount rate is the required rate of return — the minimum return an investor demands to justify owning this asset instead of a risk-free alternative. Buffett anchors here for a reason: if a stock can't generate returns above the risk-free rate after accounting for its uncertainty, it is not an attractive investment.
The protocol's baseline is the long Treasury yield. The protocol also notes a discretionary adjustment: “add 1–2 percentage points if rates are extremely low (which would otherwise produce artificially high valuations) or if the business carries above-average risk.” We codify both adjustments rather than apply them by hand, so the discount rate for any given business comes out the same way every time we run it.
| Adjustment | Premium |
|---|---|
| 10Y Treasury below 2% (auto) | +1.0pp |
| Technology / Communication Services | +2.0pp |
| Energy / Basic Materials (cyclical) | +1.0pp |
| Cyclical consumer discretionary (autos, homebuilders) | +1.0pp |
| Pharma / biotech (patent-cliff risk) | +1.0pp |
Banks, insurers, and REITs receive a separate sector exception in step I (Owner Earnings = OCF only) and a structural-caveat banner on the page; we do not pile a discount-rate premium on top.
Terminal value — perpetuity beyond year 10.
No business stops generating cash after a decade. The terminal value captures the value of all cash flows from year 11 forward, modeled as a perpetuity growing at G2: FV of TV = Year 11 OE / (r − G2). Then discount to present: PV of TV = FV / (1 + r)¹⁰. This is typically the largest single component of intrinsic value, which is why getting G2 right matters.
A guard: if G2 ever rises to meet or exceed r, the formula breaks mathematically. We cap G2 at r − 0.5% to prevent that.
Intrinsic value per share.
Sum the present value of all projected Owner Earnings (years 1 through 10) plus the present value of the terminal value: Total IV = Σ Year n OE / (1 + r)ⁿ + PV of TV. Divide by fully diluted shares outstanding to get a per-share figure that can be compared directly to Mr. Market's asking price.
Margin of safety — 25% by default.
Intrinsic value is an estimate, not a fact. Even a well-constructed model carries uncertainty in its growth assumptions, its discount rate, its maintenance CapEx proxy, and the business itself. The margin of safety is the buffer that protects against being wrong. Buffett's principle: only buy when the market price is significantly below intrinsic value.
We default to 25% — a buy is signalled only when the current price is at least 25% below the calculated intrinsic value per share. The margin is exposed as a slider on every Quote page so a subscriber who wants a harsher hurdle for, say, a cyclical or a leveraged business can dial it up to 33% or 50% (Graham's historical defaults for lower-quality names). The 25% applies uniformly across the universe; the subscriber chooses where to demand more.
The output table.
For each business, we publish: ticker, Mr. Market's asking price, intrinsic value with margin of safety, the Δ% gap between the two, the Owner Earnings input, G1 and G2. The Δ% column is the headline signal: green means Mr. Market is offering the business below our margin-of-safety price; red means he is asking too much.
Sector variants.
The eight steps above are the standard protocol — applied identically to roughly ninety of the hundred businesses we cover. For deposit-and-loan banks, the math is structurally different and the standard protocol is replaced wholesale by a sector-specific variant.
Banks — Gordon residual.GAAP operating cash flow at a bank is dominated by deposit and securities-portfolio movement that has very little to do with cash an owner could actually withdraw. Running the Owner Earnings DCF on JPMorgan produces an intrinsic value north of $2,000 per share, which is absurd. For names whose FMP industry classification contains “Bank,” we replace the OE-DCF entirely with the banking-textbook Gordon residual model:
IV per share = TBV per share × (ROE − g) / (COE − g)
Tangible book value per share is total stockholders' equity less goodwill and other intangibles, divided by diluted shares. The intangibles subtraction matters because goodwill is the price a bank once paid above book for an acquired institution; in a liquidation scenario it wouldn't recover. Return on equity is the trailing average of (net income ÷ average stockholders' equity) across each fiscal year we have on file, typically five — a longer window than Stage I uses for non-banks because credit cycles span multiple years and a single year's ROE doesn't represent normal economics. Sustainable growth g is computed as ROE × an assumed 30% retention ratio (banks pay out roughly 30-40% of earnings as dividends), capped at 5%; this is the textbook sustainable-growth formula rather than an arbitrary terminal-growth constant. Cost of equityis the 10-year U.S. Treasury yield plus a 5pp bank-specific equity risk premium, floored at 10% — Buffett's historical hurdle rate. The floor prevents the model from producing inflated multiples in low-rate regimes.
If ROE is non-positive, the bank is destroying book value and the intrinsic value is capped at zero. If ROE is positive but at or below the cost of equity, the bank earns its capital cost but creates no value above book — intrinsic value floors at tangible book per share. If ROE exceeds COE, the bank trades at a multiple of book given by (ROE − g) / (COE − g), the standard residual-income result. Every input is visible on every bank's Quote page.
Insurance carriers and REITsstill receive a sector exception in Step I (Owner Earnings = OCF only — D&A on real estate is non-economic; insurance float and reserve dynamics distort the standard cash-flow framework) plus a structural-caveat banner on the Quote page. Sector-specific variants for these — float-plus-investments for insurers, AFFO-based for REITs — are the next two on the protocol roadmap. Until they ship, the intrinsic-value figure on insurance and REIT pages is labelled informational only.
Known limitations.
The protocol is internally consistent and reproducible — every number above can be derived from the documented sources by anyone who cares to check. It is not, however, a finished product, and there are three known limitations worth naming plainly so subscribers know where the model's edges are.
EPS-consensus as an Owner Earnings growth proxy.G1 is built from analyst forward EPS estimates. Owner Earnings and GAAP EPS diverge — sometimes materially — for businesses that buy back a lot of stock (EPS rises faster than OE), book heavy stock-based compensation (EPS understates the real OE drain), or invest heavily in capex (D&A lags actual maintenance spend). For the names where this gap is largest — large-cap technology with significant buybacks and SBC, capital-intensive cyclicals — read G1 as an analyst-consensus growth proxy, not a literal forecast of how fast Owner Earnings will compound. A future protocol revision may cross-check G1 against trailing OE-CAGR; for now, the limitation is documented rather than mitigated.
3-year median smoothing. Owner Earnings is smoothed over the last three annual filings rather than the seven-to-ten-year window Buffett tended to use for cyclically normalized estimates. Three years is long enough to trim a single working-capital swing or one-off charge but not long enough to fully normalize a deep cyclical (energy producer in a price slump, a homebuilder mid-cycle). We made the trade deliberately — three years is automatable across the universe daily; ten years is not — and we flag the trade explicitly here.
Sector risk premiums are house heuristics, not calibrated equity risk premiums.The +2pp for technology, +1pp for energy / materials / cyclicals / biotech are conservative defaults rooted in Buffett's stated guidance and decades of practitioner literature (Hagstrom, Damodaran, Greenwald, Pabrai), not a daily-recalibrated implied ERP. Damodaran's implied ERP for U.S. technology in aggressive markets has historically been higher than our +2pp; in defensive markets, lower. The Quote page's manual risk-premium slider exists so a subscriber who disagrees with our default can dial it.
We believe a publication that is honest about its limitations is more trustworthy than one that pretends to have solved them. These three are the limitations we'd be asked about by any disciplined value investor reading the methodology.
A note on consistency.
Within each variant, the protocol is applied identically to every business. The same sources, the same formulas, the same decision rules. Inconsistent inputs — using different sources for one ticker versus another, or applying judgment selectively — introduce subjective bias and make comparisons across stocks meaningless. The value of the model is as much in its consistency as in its methodology. When in doubt, follow the protocol exactly as written.
Sources
- · Berkshire 1986 letter — Owner Earnings, the “Cash Flow Fallacy” appendix
- · Berkshire 1992 letter — John Burr Williams' theory of value
- · Benjamin Graham, The Intelligent Investor, 1949 (esp. ch. 8 — Mr. Market)
- · Benjamin Graham & David Dodd, Security Analysis, 1934
- · John Burr Williams, The Theory of Investment Value, 1938
- · Robert Hagstrom, The Warren Buffett Way, 3rd ed. (Wiley, 2014)
- · Mohnish Pabrai, The Dhandho Investor (Wiley, 2007)
Want to argue with the math? Open today's quote on Coca-Cola — every input is visible.