The Graham Number: Formula, Worked Example and Real Limitations

Estimating what a stock "should" be worth can absorb weeks. Benjamin Graham offered a shortcut: a one-line formula giving a reasonable ceiling price for a defensive stock.

The formula

The Graham Number is the square root of 22.5 × earnings per share × book value per share.

The 22.5 is not arbitrary: it is 15 × 1.5. Graham held that a defensive investor should pay no more than 15 times earnings, and no more than 1.5 times book value. Multiplying the two ceilings gives 22.5, and the square root brings the result back to a price per share.

A worked example

Take a company earning $3 per share, with shareholders' equity of $20 per share:

• 22.5 × 3 × 20 = 1,350
• √1,350 ≈ $36.74

If the stock trades at $25, it sits roughly 32% below its Graham Number. At $45 it sits above: the formula does not say sell, it says Graham's defensive criterion is not met.

Illustrative figures, shown to demonstrate the calculation.

Net asset discount: the other angle

The Graham Number starts from earnings. The net asset discount starts from property: what would be left if everything stopped today?

You look at net tangible assets — equity minus goodwill and intangibles, whose resale value is doubtful. Paying less than that means buying the assets at a discount.

The radical version is NCAV (net current asset value): current assets minus total liabilities, ignoring fixed assets entirely. A company trading below its NCAV sells for less than its cash and receivables net of debt. That is Graham's celebrated "net-net".

Three cases where the Graham Number misleads

This is where most articles stop, and where the trouble starts.

1. Asset-light businesses

The formula depends on book value. A software, audit or services firm has almost no balance-sheet assets: its assets are its customers, its code, its people. The Graham Number will declare it permanently overvalued, which teaches you nothing.

2. Unrepresentative earnings

A single year's EPS can be inflated by a one-off disposal, or crushed by a non-recurring charge. The formula takes that figure at face value. Graham himself recommended averaging earnings over several years — advice routinely ignored.

3. Stale book value

A balance sheet records historical cost. Land bought thirty years ago sits there at its original price; obsolete equipment sits there at a value it no longer has. Book value is not market value — sometimes not remotely.

What Graham also required (and almost everyone forgets)

The Graham Number was never meant to work alone. In The Intelligent Investor it is the last of a list of criteria the defensive investor had to satisfy in full:

• adequate size, to rule out the most fragile companies;
• a strong financial condition: current assets at least twice current liabilities;
• positive earnings for ten consecutive years, without a single loss;
• uninterrupted dividends for twenty years;
• earnings growth of at least one third over ten years, measured on three-year averages;
• a moderate P/E, below 15 on the average of the last three years;
• a moderate P/B, below 1.5 — or the product of the two below 22.5.

You will recognize the last two: that is precisely where the famous 22.5 comes from. In other words, the Graham Number condenses two criteria out of seven. The other five all concern the soundness of the business — the subject of seven signs a cheap stock is a trap.

Isolating the final line of that list and applying it alone amounts to keeping the thermometer and discarding the diagnosis. Yet that is what most free screeners do — and it is the most common source of disappointment with this formula.

Should you apply all seven today?

Not literally. The twenty-year unbroken dividend rules out excellent companies that reinvest everything, which was rare in 1949 and is commonplace now. The ten-loss-free-years rule excludes cyclical businesses by construction, however well run.

The spirit still holds: demand evidence of soundness before looking at the price. It is the letter that has aged, not the intention.

Where it sits in what we do

Nowhere inside the engine, and that is deliberate. The Graham Number is not a component of any of our readings: it produces no discount that we publish, and it moves nothing in the ranking.

We teach it because it is one of the very few valuation benchmarks a reader can rebuild by hand, from two figures anyone can find, and then hold up against whatever anybody claims. That property is rare, and it is worth more as a habit of mind than as a cog in a machine.

Our own readings are built differently. Health asks whether the business can hold up. Value asks whether the stock is cheap. The two are read separately, never folded into one score. What we publish is the rule — what puts a stock on a ranking, what takes it off, and on which data — and you can read it in full on the method page.

The Graham Number keeps a narrower job, and does it well. It tells you how much you are paying. It never tells you what you are buying.

This article is information and education. It is not personalized investment advice. Investing carries a risk of capital loss.