What the Graham Number Tells You
Benjamin Graham spent his career at Columbia and on Wall Street teaching that the market oscillates between optimism and despair, and that a disciplined buyer should extract a price from fundamentals rather than mood. The Graham Number is the compressed version of that philosophy: one ceiling price, computed from two accounting figures, that a defensive investor should not exceed. Warren Buffett absorbed the idea as a student in 1950-51 and still describes the margin-of-safety logic as the three most important words in investing.
The number works as a geometric mean of two separate ceilings. Graham told defensive buyers to pay no more than 15 times earnings and no more than 1.5 times book value, and multiplying those limits gives the 22.5 inside the formula. A stock priced exactly at the number satisfies both constraints simultaneously — a price slightly above one limit can still pass if the other ratio clears with room. That flexibility is why the square-root construction survived when a stricter minimum-of-two-ratios rule would have been simpler.
Run the defaults to see the whole idea in one line: $3.50 of diluted earnings on $28.00 of book value produces a ceiling of $46.96, and a $42.00 market quote sits 11.8% under it. The verdict band calls that below fair value with a thin cushion, not a buying signal — the cushion is the decision, and the number is only its measuring stick. For sizing up an entire acquisition or private stake rather than a listed share, a business valuation calculator extends the same discipline with multiples and discounted cash flow.
The Formula and the 22.5 Constant
The algebra is short: Graham Number = sqrt(22.5 x EPS x book value per share). The constant is the product of the two ratio ceilings from The Intelligent Investor's defensive-buyer rules — 15 for price to earnings, 1.5 for price to book. Because both inputs multiply under one square root, the output moves with the square root of each: a 20% jump in EPS lifts the ceiling by 9.5%, not 20%, and a 20% jump in book value does the same. Doubling EPS alone lifts it 41.4%.
The EPS input deserves care. Use trailing-twelve-month diluted earnings, since basic EPS overstates the ceiling for companies with options or convertibles outstanding, and strip one-off items — a single litigation reserve reversal or asset sale gain can inflate the number by a tenth or more. An EPS calculator works out the weighted-average share count and the diluted figure from the raw financials, which keeps the input honest when a company issued shares mid-year.
Book value per share is total common equity — assets minus liabilities minus preferred equity — divided by shares outstanding. The quality question is what sits inside that equity. A balance sheet carrying 40% goodwill and intangibles produces a softer anchor than one carrying cash, plant, and receivables, which is why value investors re-run the math on tangible book. A goodwill calculator isolates the intangible layer so you can judge how much of the floor is accounting convention rather than recoverable value.
Reading the Verdict and Sizing the Margin
The calculator sorts the price gap into four bands: undervalued with a margin above 25%, below fair value with a thinner cushion, roughly fair value within 10% of the number, and overvalued beyond that. On the defaults, the band edges sit at $35.22 (25% below the $46.96 number) and $51.65 (10% above it). Band membership is a triage decision — deep-cushion names earn research time, fair-value names go on the watchlist, and overvalued names get ignored until something changes.
Graham's own doctrine was that the margin of safety exists to make accurate forecasting unnecessary. Estimates of earnings and book value carry error, markets overshoot, and business quality decays quietly; a 25% discount absorbs a quarter of that damage before capital is lost. He suggested pushing toward 33% on cyclicals and weaker franchise businesses, and accepting thinner cushions only on high-grade names with decades of earnings stability behind them.
The practical output is a maximum buy price. Applying the standard 25% cushion to the default number gives $35.22 — the level at which a defensive buyer acts without needing a story about why this time is different. Note that the cushion cuts both ways: a stock that rallies from $42 to $46 is not suddenly expensive, it has simply used up the cushion, and the discipline is to let it go rather than chase the last few points of a re-rating.
The Revised 1962 Formula with Growth
Graham revisited his own work in a 1962 interview and offered a growth-adjusted version: V = EPS x (8.5 + 2g), where g is the expected annual growth rate over the next seven to ten years. The 8.5 is the multiple he assigned a company with no growth at all, and every point of expected annual growth adds two points to the multiple. It prices the trajectory rather than the balance sheet, which makes it the natural companion when book value is small relative to earnings power.
The ladder is steep. On $3.50 of EPS: minus 2% growth gives $15.75, zero growth gives $29.75, 5% gives $64.75, 10% gives $99.75, and 15% gives $134.75. The calculator clamps growth between -5% and 15% because Graham considered double-digit decade-long projections speculative, and above 15% the multiple grows faster than any reasonable estimate of certainty. A company whose bull case requires g = 20% produces the same $134.75 as one at 15% — the cap is the honesty mechanism.
The two formulas disagree by design. The defaults produce $46.96 from the classic number and $64.75 from the revised one, a 38% gap that reflects their anchors: book value on one side, growth on the other. When they diverge sharply, the stock's economics live in the intangibles — brand, network effects, reinvestment runway — and the sensible response is to triangulate with an expected return calculator rather than average the two numbers into false precision.
Where the Formula Breaks Down
Asset-light businesses wreck the classic number. A software firm earning $12.00 per share on $8.00 of book value gets a ceiling of $46.48 while the market pays $190 for the earnings stream — a 75% gap that signals the formula's blind spot, not a short sale. Two decades of buybacks shrink book equity on purpose at exactly these firms, so the input the formula depends on is engineered downward by the very companies most likely to compound.
Negative inputs fail outright. Companies posting losses at the bottom of a cycle, or carrying negative equity after buybacks and impairments, return an undefined result by design — the guard in the calculator says so in plain language instead of printing a nonsense number. That failure is information: the business has left the territory where Graham's defensive rules operate, and the honest valuation path runs through normalized earnings or a discounted cash flow model with explicit recovery assumptions.
The sweet spot is balance-sheet-heavy businesses: banks, insurers, utilities, industrials, and commodity producers with real assets generating the earnings. The bank example later in this page prices at a 63.5% upside with the market paying 0.95 times book — exactly the habitat the formula was built for. Even there, pair the pricing verdict with a leverage and solvency check via a debt to equity ratio calculator, because an overleveraged balance sheet can make a bank cheap on paper and radioactive in a downturn.
Graham Number Versus DCF and the Dividend Discount Model
A DCF calculator discounts explicit cash-flow forecasts into a present value and demands assumptions about growth, margins, and discount rates for a decade or more. The Graham Number asks for two accounting figures and returns a ceiling. The trade is precision for speed: the DCF captures the actual economics of a business but inherits every error in its forecasts, while the Graham Number ignores economics entirely except as frozen in the last balance sheet and income statement.
The dividend discount model calculator prices the cash a company hands shareholders, which suits utilities, telecoms, and consumer staples with decades of payout history. Where the DDM and the Graham Number disagree on the same stock, the disagreement is diagnostic. A high DDM value with a low Graham number points to a franchise earning returns far above its book; the reverse points to asset value the market refuses to pay for, often for a reason buried in the footnotes.
In practice the tools form a sequence rather than a competition. Screen a universe with the Graham Number in seconds, since it needs two inputs per name, then push survivors through a full discounted cash flow and payout analysis before committing capital. Graham himself recommended exactly this division of labor — cheap mechanical filters first, human judgment on the short list — and it remains the cheapest way to cover a thousand tickers without burying the interesting ones.
Worked Examples Across Three Sectors
A regional bank earns $6.20 per share on $58.00 of book value, and the market quotes $55.00. The number is $89.95, the market price sits 63.5% below it, and at the quote the bank trades at 8.9 times earnings and 0.95 times book — squarely in the deep-value band. The follow-up work is credit quality: charge-off trends, commercial real estate exposure, and unrealized bond losses, since the formula prices the balance sheet without reading it.
An industrial company earns $8.40 per share on $65.00 of book and trades at $120.00, which puts the number at $110.84 and the stock 7.6% above it — inside the roughly-fair band. At the quote the market pays 14.3 times earnings, close to Graham's 15x ceiling on its own. This is the formula working as intended: not screaming buy, not screaming sell, and leaving the decision to business quality and price discipline.
The tech example earns $12.00 on $8.00 of book at a $190.00 price, and the number of $46.48 sits 75% under the market. Sensitivity matters here: move either input 20% and the ceiling shifts only 9.5%, both together 20% — the square root dampens everything, so no reasonable revision rescues a formula whose anchor the company has deliberately shrunk. For sizing a position once a valuation case is made, a stock beta calculator measures how much market exposure the holding adds.
The Full Defensive Screen Behind the Number
The Intelligent Investor's chapter on defensive buying set seven screens, and the ratio ceilings behind 22.5 are only the last of them. Graham wanted adequate size, a current ratio of at least 2, long-term debt below net current assets, ten consecutive years of profits, twenty uninterrupted years of dividends, and at least one-third cumulative earnings growth over a decade — plus the pricing limit itself. The number prices the survivors; it was never meant to stand alone.
The quality legs still translate decades later. Earnings stability filters out cyclicals mid-peak, the dividend record forces a real capital-return history, and the leverage limits do the balance-sheet work the formula skips. A modern practitioner can run distress scoring through an Altman Z-Score calculator to catch bankruptcy risk that a cheap P/B hides — the 1990s value traps and the 2007-09 banks all looked statistically cheap while solvency quietly deteriorated.
Treat the whole apparatus as a quarterly routine rather than a one-off trade signal. Refresh both inputs when each report lands, re-grade the band, and let the portfolio math — a ROI calculator covers the realized side — keep score of whether the discipline is actually paying you. Graham's own returns came from consistency over decades, and the number is simply the piece of that system small enough to carry in your head.