Graham Number Strategy

The Graham Number is a formula—the square root of 22.5 times earnings per share times book value per share—that yields a single price threshold Benjamin Graham proposed as an upper limit for margin-of-safety buying. It is not a target price but a screen: stocks trading below the Graham Number are theoretically safer from permanent capital loss.

The formula and its logic

Graham’s number emerges from a blend of two common valuation approaches: price-to-earnings (P/E) multiples and price-to-book (P/B) ratios. He believed a prudent investor should not pay more than a moderate P/E for a stable business and simultaneously should respect the company’s balance-sheet backing.

The formula—√(22.5 × EPS × Book Value Per Share)—reflects Graham’s judgment that safe purchases sit at no more than 15 times earnings and 1.5 times book value. Multiplying 15 × 1.5 yields 22.5. The square root normalizes the result to a single price, making stocks of different sizes comparable on the same screen.

A company with USD 2 earnings per share and USD 20 book value per share would have a Graham Number of √(22.5 × 2 × 20) = √900 = USD 30. Any stock of this quality trading below USD 30 passes the filter; above it, the margin of safety shrinks.

Historical context and evolution

Graham developed this framework in the mid-20th century, a period of lower corporate debt and more conservative accounting. The formula gained wider use after The Intelligent Investor (1949) and was popularized further by Graham disciples including Warren Buffett and value-oriented practitioners. It became a canonical first screen in fundamental analysis education.

Over decades, the formula’s rigidity became apparent. It treats a supermarket and a software firm identically, ignoring growth rates, competitive advantage, and industry cyclicality. A stock yielding rapid earnings growth could genuinely merit a higher price despite the Graham Number’s ceiling. Yet as a mechanical starting point—a way to exclude obvious overvaluations—it retained pedagogical and practical value.

Applying the screen

Most practitioners use the Graham Number as a preliminary filter, not an investment decision. A stock trading at USD 18 with a Graham Number of USD 30 enters the candidate list. The analyst then scrutinizes the business: Is earnings quality genuine? Are assets truly worth book value, or are they inflated? Does the company have a durable competitive edge, or is it vulnerable to disruption?

The formula works best on mature, cyclical stocks where earnings are volatile but fundamentals are stable: utilities, consumer staples, regional banks, industrial manufacturers. It performs poorly on high-growth firms, which can sustain elevated valuations, or asset-light technology companies, where book value is negligible relative to earning power.

Investors must also adjust the inputs. Book value can be distorted by one-time gains, goodwill write-downs, or depreciation choices. Earnings per share varies depending on share buybacks and capital structure. Using normalized or trailing-twelve-month figures rather than quarterly snapshots improves reliability.

Strengths of the approach

The Graham Number is admirably simple. It requires only two numbers from any company’s most recent financial statements—earnings and book value per share—both of which are public and audited. No proprietary data, no models, no guesswork.

The formula also enforces discipline. Instead of chasing performance or following sector momentum, the Graham Number investor systematically moves through screened lists, limiting downside through methodical price discipline. This mechanical approach removes emotion and has, historically, protected investors from obvious value traps during bubbles.

Stocks meeting the Graham Number test, on average, have underperformed the broader market during bull runs but outperformed during drawdowns—a classic defensive posture consistent with Graham’s emphasis on risk minimization.

Limitations and pitfalls

The greatest weakness is that the formula ignores quality. A USD 10 stock with USD 1 earnings and USD 5 book value has the same Graham Number as a equally-priced rival with higher return on equity or stronger competitive moats. The screen does not distinguish between a compounder and a value trap.

The formula also assumes earnings persistence. A company with cyclically-depressed earnings will show an artificially low Graham Number; once earnings recover, the stock may jump above the threshold, forcing the investor to sell winners. Conversely, a firm in secular decline will have a Graham Number that overstates safety.

Book value itself is a brittle metric. Intangible assets, goodwill, and hidden liabilities can render it meaningless. A technology company with minimal tangible assets will have a tiny Graham Number despite genuine earning power. A manufacturer with obsolete inventory or stranded real estate will have an inflated book value suggesting false bargains.

Complementary screens

Few practitioners today rely on the Graham Number in isolation. Most combine it with additional filters: minimum return on equity thresholds, debt-to-equity caps, current ratio minimums, and insider ownership checks. Some layer in relative dividend yield or require earnings growth above inflation.

The Graham Number pairs well with price-to-sales screening, which avoids earnings distortions, and negative enterprise value checks, which ensure a cash floor. It also complements fundamental analysis of competitive position and management quality, which the formula cannot capture.

Modern relevance

The Graham Number survives in professional and amateur portfolios because its simplicity remains a virtue in an age of data overload. It is a teaching tool that embeds Graham’s conservatism into a single number and acts as a circuit-breaker against irrational valuations. Whether applied as a hard rule or a soft guideline, it disciplines the buying process.

Yet few serious investors would construct an entire portfolio from Graham Number stocks without deeper assessment. The formula’s utility lies in narrowing a universe of thousands of candidates down to a handful worth examining in detail—a sieve, not a decision engine.

See also