Skip to main content
The Value InvestorsLesson 1 of 6 · Price first

Investor frameworks

The Graham Number and NCAV

Benjamin Graham's two defensive value tests — a ceiling price built from earnings and book value, and the liquidation-value floor he called a net-net.

5 min readIntermediateLast checked against the product on

The Graham Number

Graham wanted a single number a defensive investor could compute from a filing and compare to a price. His formula multiplies two of the criteria he insisted on — a modest multiple of earnings and a modest multiple of book value — and takes the square root:

Graham Number = √(22.5 × EPS × book value per share)

The 22.5 is not magic. It is 15 × 1.5: Graham's ceiling of 15× earnings multiplied by his ceiling of 1.5× book value. A price below the resulting number means the stock clears both tests at once.

Worked: EPS of $4 and book value per share of $20 give √(22.5 × 4 × 20) = √1800 ≈ $42.43. A stock trading at $35 sits below its Graham Number — the classic "undervalued" flag on a Pythia company page.

NCAV, and what a net-net actually is

The second test ignores earnings entirely and asks what the business would be worth if it stopped:

NCAV per share = (current assets − total liabilities) ÷ shares outstanding

Note what is missing: property, plant, goodwill, and every other long-lived asset are set to zero. Graham then demanded a further discount, buying only below two-thirds of NCAV — so a company with $25 of NCAV per share is a candidate below about $16.67.

At that price the market is saying the business is worth less than its own liquid assets net of every debt. Graham's point was not that liquidation was likely; it was that the gap gave him room to be wrong.

What a discounted value is actually made of

Graham's formulas are deliberately crude because he distrusted precision built on guesses. This calculator shows why. Move the terminal growth rate half a percent and watch the answer move — and watch the share of the value that comes from the years you never actually forecast.

Discounted cash flow (simplified)

$m
Accepts 0 to 1000000. Values outside that range are adjusted when you leave the field.
%/yr
Accepts -10 to 40. Values outside that range are adjusted when you leave the field.
%
Accepts 1 to 30. Values outside that range are adjusted when you leave the field.
Accepts 1 to 25. Values outside that range are adjusted when you leave the field.
%
Accepts -2 to 6. Values outside that range are adjusted when you leave the field.

Present value

$20,534m

58% of that value is the terminal assumption — the part beyond the years you actually forecast.

A teaching model, not the Pythia DCF that computes PVS — that one carries sector base rates, a real cost of capital and a refusal set. The point here is how much of the answer the terminal assumption owns.

Worked example

Graham's own defensive test pairs the two ceilings rather than applying either alone: a price no more than 15x average earnings of the past three years, and no more than 1.5x book value.

The product of those two, 22.5, is where the Graham Number comes from - it is not a separate rule, it is the same pair written as one expression. That is why a company can clear the combined number while failing one leg badly, and why the legs are worth reading separately.

Figures: Graham, The Intelligent Investor (1949), ch. 14 - the defensive-investor criteria

Where these tests stop working

  • Asset-light businesses break the book-value leg. A software company with almost no tangible book value will fail the Graham Number no matter how profitable it is. That is not a verdict on the business.
  • Negative earnings make the formula undefined, not "very cheap" — the square root of a negative number is not a low price.
  • Net-nets are nearly extinct in large caps. Oppenheimer's returns came from a universe of small, ignored companies; the same screen run on the S&P 500 today usually returns nothing, and when it does return something the reason is normally distress rather than neglect.

Graham himself was blunt about the limits: his formulas were, in his words, precise arithmetic applied to highly imprecise assumptions. Pythia shows the Graham Number where its inputs exist and says n/a where they do not, rather than printing a number the formula cannot support.

Check yourself

3 questions. Nothing is recorded unless you are signed in, and nothing here affects anything else.

1. Where does the 22.5 in the Graham Number come from?
2. What does net current asset value deliberately ignore?
3. In a discounted cash flow, why does the terminal growth assumption matter so much?

3 questions left. An unanswered question counts as a miss, so the check waits for all of them.

Sources

Further reading