The Graham Technique: Buying Businesses Below Their Own Assets
Most stock analysis tries to forecast a company's future. A different, older discipline asks a narrower question instead: can I buy this business today for less than a conservative estimate of what it is worth right now, with enough cushion that being partly wrong still leaves me whole? This article walks through that quantitative screening method, named for the economist and investor who formalized it in the mid twentieth century, Benjamin Graham, with the arithmetic worked in full.
The core principle: margin of safety
The foundation of this approach is margin of safety: the gap between the price paid for a stock and a conservative, quantitative estimate of what the underlying business is worth, a cushion designed to protect an investor even if their analysis turns out to be partly wrong or the future turns out worse than expected. This is a fundamentally different posture than growth investing, which requires forecasting a company's future and paying for that forecast today; the margin-of-safety approach instead tries to pay so little relative to what already, verifiably exists on the balance sheet that the investor barely needs to forecast anything at all.
This does not mean the approach ignores the income statement entirely; earnings still matter, both as a filter for business quality and as an input into the second screen described below. What it deliberately avoids is paying a premium today for earnings growth that has not yet happened, the central bet underlying most growth investing, in favor of paying at or below a conservative estimate of what already exists, with any future growth treated as a bonus rather than a requirement for the investment to work out. That reframing, from forecasting the future to pricing the present, is the single habit of mind that separates this discipline from most other forms of stock analysis.
Two specific screens operationalize this idea. The first is net current asset value (NCAV), sometimes called a net-net, calculated as NCAV = current assets − total liabilities. This figure represents a highly conservative liquidation estimate: what would theoretically remain for shareholders if the company sold off only its current assets (cash, receivables, inventory) at close to book value and paid off every liability, ignoring the value of any factories, equipment, real estate, patents, or brand entirely. The classical version of this screen looks for stocks trading at or below two-thirds of NCAV per share, a deliberately conservative threshold that builds an extra cushion into an already conservative number. The second screen is the Graham number, a rough fair-value ceiling calculated as Graham number = square root of (22.5 × earnings per share × book value per share), where the constant 22.5 comes from combining a maximum price to earnings ratio of 15 with a maximum price to book ratio of 1.5 (15 × 1.5 = 22.5). A stock trading meaningfully below its Graham number is priced cheaply on both an earnings basis and an asset basis simultaneously, rather than merely one or the other.
Beyond these two headline screens, the broader discipline traditionally layers in several additional quantitative filters meant to weed out statistically cheap stocks that are cheap for a genuinely bad reason. These typically include a minimum current ratio, often set at 2.0 or higher, to confirm adequate short-term liquidity; a requirement for positive earnings in most, if not all, of the trailing several years, to screen out companies with an unstable or deteriorating earnings history; a cap on total debt relative to net current assets, to avoid businesses that are cheap because they are over-leveraged rather than merely overlooked; and, where available, a history of uninterrupted dividend payments, treated as a rough proxy for durable, cash-generative operations rather than accounting profit alone. No single one of these additional filters is decisive, but stacking several of them together meaningfully reduces the odds that a stock passing the NCAV or Graham number screen is cheap because the business itself is failing.
Two worked examples
Take a company with earnings per share of $3.20 and book value per share of $28.00. Its Graham number is square root of (22.5 × $3.20 × $28.00). Working the multiplication first: 22.5 × $3.20 = $72.00, then $72.00 × $28.00 = $2,016, and square root of $2,016 ≈ $44.90. That $44.90 figure is a rough fair-value ceiling combining both an earnings-based and an asset-based cap. If the stock currently trades at $30.00 per share, the margin of safety is ($44.90 − $30.00) ÷ $44.90 ≈ 33.2%, meaning an investor is paying roughly a third less than this conservative ceiling, a substantial cushion, before any credit is given for future earnings growth at all.
The second example works a net current asset value screen. A small manufacturer reports current assets of $50,000,000 and total liabilities of $30,000,000. NCAV = $50,000,000 − $30,000,000 = $20,000,000. With 4,000,000 shares outstanding, NCAV per share = $20,000,000 ÷ 4,000,000 = $5.00. The classical net-net screen looks for a purchase price at or below two-thirds of that figure: $5.00 × (2 ÷ 3) ≈ $3.33. If the stock trades at $3.00 per share, it qualifies: an investor buying at $3.00 is paying less than the liquidation value of the company's current assets alone, net of every liability, receiving the company's factories, equipment, brand, and any future earnings power entirely for free on top of that already conservative valuation.
Applying the additional filters described above to this same manufacturer sharpens the picture further. Suppose the company's current ratio is 2.4 (comfortably above the 2.0 threshold), it has posted positive earnings in each of the last seven years, and its total debt of $8,000,000 sits well below its $20,000,000 of net current assets. Each of these facts independently supports the case that the stock is cheap because the market has neglected it rather than because the underlying business is deteriorating, and together they build a considerably stronger case for the position than the NCAV screen alone would provide, since a company can technically trade below NCAV while also carrying warning signs, such as declining earnings or a thin liquidity cushion, that a single-metric screen would miss entirely.
What the evidence shows
Academic tests of quantitative net current asset value screening, run across multiple decades and multiple markets, have generally found that baskets of stocks meeting the net-net criterion delivered above-average returns relative to broad market benchmarks over the periods studied, with the important caveat that qualifying stocks in modern developed markets, particularly large, well-covered U.S. exchanges, have become considerably scarcer than they were in the mid twentieth century, as more market participants, faster information flow, and a larger universe of quantitative screening tools have competed away much of the easiest mispricing. The strategy has, historically, worked better and more consistently in smaller, less-followed stocks and in markets with less analyst coverage, precisely the corners of the market where genuine neglect rather than efficient pricing is more likely to explain a stock trading below its own liquid assets.
A broader body of research on value investing generally, buying stocks that are statistically cheap relative to earnings, book value, or cash flow, has found a persistent, if inconsistent and cyclical, historical premium for cheap stocks over expensive ones across long time horizons and multiple countries, though the premium has come with extended stretches, sometimes lasting the better part of a decade, where expensive stocks outperformed cheap ones, testing the patience of anyone following the discipline mechanically through a full market cycle.
Studies specifically layering additional quality filters, positive earnings history, adequate liquidity, and manageable debt, on top of a pure statistical cheapness screen have generally found the combined approach delivers a smoother, less volatile return pattern than cheapness alone, with fewer catastrophic single-stock losses from companies that were cheap because they were failing rather than merely neglected. This is broadly consistent with the intuition behind stacking filters: a pure NCAV screen with no quality overlay will capture both genuinely mispriced bargains and genuinely troubled businesses in the same basket, and the additional filters exist specifically to tilt that mix toward the former.
Applying it in a real portfolio
An individual investor drawn to this approach should treat it as a screening and diversification exercise rather than a single-stock conviction bet: run the NCAV and Graham number screens across a broad universe of stocks, build a basket of a dozen or more names that qualify rather than concentrating in two or three, and expect that some fraction of the basket will be cheap for a legitimate reason (a genuinely deteriorating business) rather than because of market neglect. Because qualifying candidates in large, liquid U.S. markets have become scarce, an investor serious about running this screen mechanically should expect to look further down the market capitalization spectrum, into smaller and less-followed stocks, where informational neglect remains more plausible than in heavily analyzed large caps.
A modern, pragmatic adaptation for investors who find genuine net-nets too scarce to build a full basket is to relax the NCAV threshold to a broader statistical cheapness screen, low price to book combined with positive earnings, low debt, and the same current ratio and earnings-history filters described above, while retaining the same margin-of-safety spirit and the same insistence on diversification across a meaningful number of names. This trades some of the extreme conservatism of the pure net-net screen for a substantially larger, more workable universe of candidates, particularly in developed markets where true net-nets have become rare outside of periods of broad market stress.
Actionable breakdown
- Running the screens
- Calculate NCAV as current assets minus total liabilities.
- Screen for prices at or below two-thirds of NCAV per share.
- Calculate the Graham number from EPS and book value.
- Filtering candidates
- Require positive earnings, not just a low asset price.
- Check debt levels to rule out distressed, failing businesses.
- Confirm the balance sheet assets are genuinely liquid.
- Building the basket
- Diversify across a dozen or more qualifying names.
- Expect some names to be cheap for good reason.
- Hold with patience through years, not months.
Common pitfalls
Falling for a value trap: a stock can be statistically cheap because the underlying business is genuinely shrinking or becoming obsolete, not because the market is neglecting it.
Concentrating in too few names: the discipline depends on diversification to offset the qualifying stocks that never recover, so a handful of positions defeats the purpose.
Underestimating how long mispricing can persist: a genuinely cheap stock can stay cheap for years before the market re-rates it, which can look indistinguishable from being wrong.
Ignoring balance sheet quality: current assets are not automatically liquid; aging receivables or obsolete inventory can overstate the true liquidation value behind an NCAV figure.
Skipping the quality filters: screening on price alone, without checking liquidity, debt, and earnings history, mixes genuine bargains with genuinely troubled businesses in the same basket.
The bottom line
This approach succeeds by systematically buying diversified baskets of financially sound companies priced below a conservative estimate of their own assets and earnings, then waiting patiently for the market to close the gap.
Profitability measures · Comparability problems · Intrinsic value versus market price · Price to earnings ratio · Value investing