Graham Number Calculation: What It Can Prove

A stock can look inexpensive at $40 until its earnings prove temporary or its stated book value rests largely on goodwill. A Graham number calculation does not resolve those questions. It does something narrower and useful: it establishes a conservative price reference from two reported measures, earnings per share and book value per share.

That restraint is the point. Benjamin Graham's framework was designed to reduce dependence on forecasts, stories, and market sentiment. Used carefully, the Graham Number can help an investor identify where market price has moved far from a company's reported earning power and net asset base. Used carelessly, it can turn a damaged balance sheet or a cyclical earnings peak into a false signal.

What the Graham Number Measures

The Graham Number is a maximum price estimate under a conservative set of assumptions. Its familiar formula is:

Graham Number = √(22.5 × EPS × BVPS)

EPS is earnings per share. BVPS is book value per share. The constant 22.5 comes from multiplying two limits Graham used for a defensive investor: a price-to-earnings ratio of 15 and a price-to-book ratio of 1.5.

The formula is not an intrinsic value model in the modern discounted-cash-flow sense. It does not forecast revenue, margins, reinvestment needs, interest rates, or terminal value. It asks whether the price is reasonable relative to recent profits and accounting equity, with neither component allowed to dominate the result.

That structure creates a useful discipline. A company with high earnings but little net asset support will not receive the full benefit of its earnings. A company with substantial book value but weak profits will not receive the full benefit of its assets. The geometric relationship penalizes imbalance.

Graham Number Calculation, Step by Step

Start with the company’s annual filing, not a finance portal headline. The two inputs appear simple, but definitions matter.

First, identify earnings attributable to common shareholders. Divide that figure by an appropriate weighted-average common share count to calculate EPS, or use the reported diluted EPS when it reflects the economic dilution an owner would reasonably expect. For a stable company, trailing 12-month earnings can be a practical starting point. For a cyclical business, a single strong year may be a poor representation of normal earning power.

Second, calculate BVPS by dividing common shareholders’ equity by common shares outstanding at period end. If preferred equity sits ahead of common holders, it should not be treated as common book value. Nor should a share count from a different reporting date be paired casually with balance-sheet equity. Small mismatches can distort a screen, particularly after large buybacks or stock issuance.

Consider a hypothetical company reporting diluted EPS of $4.00 and book value per share of $20.00. The calculation is:

√(22.5 × $4.00 × $20.00) = √$1,800 = $42.43

If the shares trade at $30, the market price is below this reference point. If they trade at $55, the price is above it. Neither result is a buy or sell instruction. It is an invitation to inspect the inputs and understand why the market is assigning that price.

The formula stops working when either EPS or BVPS is zero or negative. That is not a technical inconvenience to work around with an adjusted number. It is information. A company with losses or negative common equity does not fit the conditions the Graham Number was built to evaluate.

Use reported figures, then test their quality

Reported EPS and equity are starting points, not unquestioned facts. A filing-first review should test whether earnings include unusual gains, whether a major impairment has changed the economics of the business, and whether the capital structure has shifted materially since year-end.

For EPS, examine discontinued operations, asset-sale gains, litigation recoveries, tax benefits, restructuring charges, and pension-related items. Some adjustments may be economically justified; others merely make a weak period look cleaner. The correct response is not to accept management's preferred adjusted EPS automatically. Reconcile it to the income statement and decide whether the item is recurring, nonrecurring, or simply inconvenient.

For book value, inspect the composition of equity. Goodwill and acquired intangible assets may be valid accounting assets, but they often provide less protection in a liquidation or stressed operating environment than cash, receivables, inventory, or productive property. Accumulated share repurchases can also reduce reported equity without necessarily reducing business quality. The implication depends on the price paid for those repurchases and the return earned on the remaining capital.

When the Formula Is Most Useful

The Graham Number is most informative for established companies with positive earnings, positive common equity, understandable operations, and a financial history long enough to evaluate across more than one reporting period. Asset-heavy industrial businesses, insurers, banks, and mature enterprises can sometimes fit the framework better than fast-growing software companies or firms whose value rests primarily on internally developed intangibles.

Even then, industry context matters. For a bank, book value can be central, but its reliability depends on loan-loss reserves, securities marks, credit concentration, and capital quality. For an insurer, equity must be read alongside loss reserves and underwriting performance. For a manufacturer, inventory valuation, pension obligations, and cyclicality deserve attention. The same formula does not eliminate the need to read the notes to the financial statements.

The measure is especially useful as a screening constraint. It can reduce a large universe to companies worth investigating, then force a more demanding question: are earnings and book value durable enough for the comparison to mean anything?

A price below the Graham Number may reflect neglect. It may also reflect deteriorating margins, weak governance, excess leverage, litigation risk, or assets carried above their economic value. The screen identifies a discrepancy. Filings explain whether the discrepancy is opportunity, impairment, or something unresolved.

Where Investors Misuse It

The most common mistake is treating the formula as a price target. It is not. Graham’s original criteria came from a different market structure, accounting environment, and interest-rate regime. The 15 times earnings and 1.5 times book value limits are conservative anchors, not universal laws.

A second mistake is using one year of EPS for a cyclical company. Commodity producers, homebuilders, freight businesses, and some industrial companies can appear cheap near a profit peak and expensive near a trough. A multi-year earnings record, normalized cautiously and tied back to reported results, is often more informative than the latest annual number.

A third mistake is ignoring debt. Two companies can have the same EPS and BVPS while carrying radically different obligations. The Graham Number does not directly account for net debt, lease commitments, refinancing pressure, or off-balance-sheet exposures. A balance sheet with heavy debt can make both reported earnings and reported book value less protective than they first appear.

Management credibility also sits outside the formula. An executive can describe a temporary setback on an earnings call while the filing discloses customer concentration, declining backlog, covenant amendments, or a growing contingent liability. The calculation cannot judge that gap. A careful owner should.

Build a Repeatable Review Around the Number

A disciplined process treats the Graham Number as one checkpoint among several. Start with a consistent reporting period and retain the source figures. Compare the result with the current share price, but also compare EPS, BVPS, revenue, free cash flow, and debt across several years. Then read the annual report and recent quarterly filings for the events that may have changed the business after the measurement date.

Document the judgment calls. If you exclude an asset-sale gain, state why. If goodwill makes stated book value less meaningful, state that too. If normalized earnings differ from reported EPS, preserve both figures rather than replacing one with a cleaner narrative. Evidence is more useful when an investor can retrace the calculation and disagree with an assumption without losing the underlying record.

This is where a filing-first research workflow earns its keep. Tools such as Hety can screen Graham-style quality and valuation measures across a broad stock universe, while keeping the reported figures and disclosure context available for verification. The screen saves time; it does not outsource judgment.

Past results do not predict future performance, and this material is not investment advice. The value of the Graham Number is not that it predicts the next quote. Its value is that it makes an investor ask a better question before acting: what, exactly, do these earnings and assets represent?