Benjamin Graham's Intrinsic Value Formula: How to Calculate It and Find Undervalued Stocks

Benjamin Graham is known as the father of value investing. He spent decades building a framework for deciding what a stock is actually worth and how big a discount you need before you buy it. Two formulas sit at the heart of that framework: the Graham Number and his intrinsic value formula.

Both are still used today. Neither is complicated. This article explains how each one works, walks through the calculations, and covers when to use which.

Who was Benjamin Graham?

Graham was a professor at Columbia Business School, born in 1894 and died in 1976. He published Security Analysis in 1934 with David Dodd, a book that became the foundation of fundamental investment analysis. His follow-up, The Intelligent Investor, came in 1949 and is still one of the most widely read investing books ever written. Warren Buffett studied under Graham at Columbia and has described him as the second most influential person in his life, after his own father.

Graham's starting point was straightforward: stocks are partial ownership of real businesses, and businesses have measurable worth. His goal was to find stocks trading well below that worth. He called the gap the margin of safety.

Formula 1: The Graham Number

The Graham Number is a purely mechanical estimate of the maximum fair price for a stock. Two inputs, no forecasts.

Graham Number = √(22.5 × EPS × Book Value Per Share)

The 22.5 constant is not arbitrary. Graham set two rules for defensive investors in The Intelligent Investor: never pay more than 15 times earnings, and never pay more than 1.5 times book value. Multiply those two ceilings together and you get 22.5.

The inputs:

• EPS: trailing twelve months net income divided by diluted shares outstanding
• Book Value Per Share: shareholders' equity divided by shares outstanding

Worked example

Suppose a company reports EPS of $4.00 and book value per share of $30.00.

• 22.5 × $4.00 × $30.00 = $2,700
• √$2,700 ≈ $51.96

If the stock trades at $38, you're buying it at roughly a 27% discount to the Graham Number. That's your cushion. If it trades at $60, you're above it. By Graham's rules, you're paying more than a defensive investor should.

Formula 2: Graham's Intrinsic Value Formula

Graham's second formula appears in the 1962 revised edition of Security Analysis. It was designed to account for growth, which the Graham Number does not.

Intrinsic Value = EPS × (8.5 + 2g)

Where:

• EPS: trailing twelve months earnings per share
• 8.5: Graham's assumed P/E for a company with zero expected growth
• g: expected annual earnings growth rate over the next 7 to 10 years (as a whole number, so 6 for 6%)

Worked example

Same company: EPS of $4.00, expected annual growth of 6%.

• 8.5 + (2 × 6) = 8.5 + 12 = 20.5
• $4.00 × 20.5 = $82.00

At $60, that stock looks undervalued. At $90, it's above intrinsic value.

A note on the original formula: Graham's 1962 version also multiplied the result by 4.4/Y, where Y is the current yield on AAA corporate bonds (4.4% being the rate at the time). Most people drop this today because it produces odd results when interest rates are very high or very low.

The limitation: you have to estimate g

The intrinsic value formula gives you more precision than the Graham Number, but it comes at a cost. You have to estimate future earnings growth, and that estimate is entirely subjective. Two people looking at the same company can get very different intrinsic values just by changing their growth assumption.

That sensitivity is real. For EPS of $4.00: at g = 5, intrinsic value is $74. At g = 8, it's $98. The $24 difference has nothing to do with the business. It's entirely down to which number you plugged in.

The margin of safety

The margin of safety is the most important idea in Graham's entire framework. Both formulas are only useful if you combine them with it.

The rule is straightforward: only buy a stock when it trades at a real discount to your estimate of intrinsic value. Graham typically suggested no more than two thirds of intrinsic value. If you calculate a stock is worth $75, pay $50 or less.

The logic is not complicated. Your estimate could be wrong. The business could hit problems you didn't see coming. The market can stay mispriced for longer than you expect. A discount gives you room for any of those things to go wrong without wiping you out. The bigger the discount, the more room you have.

A stock right at its Graham Number has no margin of safety in this framework. At 60% of the Graham Number, you have a 40% cushion. Graham didn't set a single exact threshold, that's a judgement call, but the principle is consistent: the bigger the gap between price and value, the less has to go right for the investment to work.

Which formula should you use?

The two formulas answer slightly different questions.

The Graham Number is purely mechanical. No forecast needed, no assumptions about the future. It checks whether the current price is reasonable given current earnings and book value, using the same yardstick across every company. That makes it good for screening: you can run it across thousands of stocks without making any discretionary calls.

Graham's intrinsic value formula accounts for growth, which matters if a company is genuinely expanding earnings. A business growing at 10% a year is worth more than one growing at 2%, even if their current EPS and book value look the same. The Graham Number misses that. The intrinsic value formula captures it, but only if your growth estimate is sensible.

In practice, most systematic value investors use the Graham Number to build a shortlist and then apply the intrinsic value formula (or a discounted cash flow model) when looking closely at individual candidates. Use the number to find the names worth researching. Use the formula to think through what they're actually worth.

Walking through a real example: Pfizer (PFE)

Pfizer peaked at around $61 in late 2021, driven by COVID vaccine revenues. By early 2026 the stock sat at $27, a fall of roughly 55% from that peak. On the surface it looks cheap: P/E of 11, dividend yield above 3%, revenues still above $60 billion a year. The question is whether the Graham Number confirms that cheapness.

The inputs from Pfizer's latest annual filing:

• EPS (trailing twelve months): $2.44
• Book Value Per Share: $16.35 (shareholders' equity divided by diluted shares)

Graham Number = √(22.5 × $2.44 × $16.35) = √$898 ≈ $29.94

At $27.14, Pfizer is trading about 9% below its Graham Number. That is technically a margin of safety, but a thin one. Graham himself wanted to buy at no more than two thirds of intrinsic value, specifically a discount of 33% or more. At 9%, the cushion is small and leaves little room for things to go wrong.

There is also a nuance worth flagging. Graham set two separate rules for defensive investors: a maximum P/E of 15 and a maximum P/B of 1.5. Pfizer's P/E of 11 passes comfortably. Its P/B of 1.66 does not. The Graham Number combines both into a single formula, so the low P/E is enough to pull the result below the current price. But strictly speaking, Pfizer fails one of Graham's individual criteria.

Checking the Piotroski F-Score adds useful context: Pfizer currently scores 5 out of 9. That is a neutral reading. The business is not clearly deteriorating, but it is not showing the broad improvement across profitability, leverage, and efficiency that Graham would have wanted to see alongside a cheap valuation. Post-COVID revenues have stabilised, but the financials are not yet showing a clear recovery trajectory.

The Graham Number is saying Pfizer is worth putting on your watchlist. The thin margin of safety and neutral F-Score suggest that waiting for a larger discount before committing would be more in the spirit of what Graham actually advocated.

Limitations of both formulas

Neither formula works for every type of company, and applying them without understanding their limits will produce bad results.

Negative EPS or negative book value makes both formulas undefined. You can't take the square root of a negative number, and a loss-making company has no meaningful Graham Number. This rules out a fair chunk of the market: early-stage companies, cyclical businesses in a down year, companies in distress.

Financial companies (banks, insurers, REITs) have balance sheets that work differently from industrial or consumer businesses. Their assets, things like loans and securities, are already marked close to market value, so book value has a different meaning. The Graham Number tends to produce misleading results for these companies. Graham himself did not apply these formulas to banks in the same way.

Asset-light businesses (software, media, professional services) often have very low book values relative to their earning power. The most valuable things in those businesses, intellectual property, brand, customer contracts, do not appear on the balance sheet. A software company with $2 of book value per share and $5 of EPS gives a Graham Number of around $4.74, which looks absurd next to a $20 stock price. But the low book value is an accounting artefact, not a sign of overvaluation. For these companies, the intrinsic value formula makes more sense than the Graham Number. Bristol-Myers Squibb is a real example: strong earnings, a 73% gross margin, and a P/B of 6.74 that produces a Graham Number of just $26.86 against a stock price above $60. That does not mean BMY is dangerously overvalued. It means the formula is not the right tool for a company where most of the value sits in drug patents rather than tangible assets.

And finally, neither formula says anything about competitive position, management quality, or industry dynamics. A company can pass both tests and still be a poor investment if the business is in structural decline or being outcompeted. These are starting filters for analysis, not a replacement for it.

How to screen for stocks below the Graham Number

Doing the Graham Number calculation by hand for a handful of stocks takes a few minutes. Across thousands, you need a screener.

When I run the Graham Number filter on StockPik right now, around 2,965 stocks out of 6,000+ screened are trading below their Graham Number. That is a broad starting pool. The Graham Number casts a wide net by design. It is a conservative ceiling, not a high bar. The job after that is to narrow the list down to companies where the business is actually in decent shape. You can browse the current list of stocks below their Graham Number, sorted by margin of safety, updated weekly from SEC EDGAR data.

StockPik calculates the Graham Number using SEC EDGAR financial data, updated weekly. You can filter directly for stocks where the price sits below the calculated value, and combine it with filters like P/E ratio, Piotroski F-Score, or debt-to-equity.

A combination I find useful: stocks below their Graham Number with a Piotroski F-Score of 7 or higher. The Graham Number finds cheap stocks. The Piotroski score filters out the ones where the underlying business is quietly getting worse. That pairing directly addresses the value trap problem.

Add a P/E below 15 (Graham's own ceiling) and the list tightens further. You won't have many names left, but the ones that remain will have passed three independent tests using the same underlying logic.

You can see how each metric is calculated on the methodology page.

Bottom line

Graham's formulas are not complicated. The Graham Number is a square root of two multiplied numbers. The intrinsic value formula is a single multiplication. The maths is not the point. The point is the discipline. They force you to connect the price you're paying to something real, rather than buying based on recent momentum or what everyone else is doing.

Combine them with the margin of safety concept and you have a systematic way to look for stocks where the odds are in your favour. That's been the basis of value investing since Graham published it, and it still works.

---

Sources

• Graham, B. & Dodd, D. (1934). Security Analysis. McGraw-Hill.
• Graham, B. (1949). The Intelligent Investor. Harper & Brothers.
• Graham, B. (1962). Security Analysis, 4th edition. McGraw-Hill. (Source of the V = EPS × (8.5 + 2g) formula.)
• Pfizer Inc. FY2024 Annual Report (Form 10-K), filed with the SEC.
• SEC EDGAR: source of all financial statement data used by StockPik.

---