Concept

Magic Formula

investingvalue-investingquantitativescreening

Magic Formula

A quantitative stock-screening approach developed by Joel Greenblatt and described in The Little Book That Beats the Market (2005). The formula ranks the market by two crude metrics — return on tangible capital and earnings yield — combines the rankings, and selects the top-ranked stocks. Backtesting from 1987 shows the top decile beating the market, with returns ordered by decile (top beats second, second beats third, etc.), not just top-vs-bottom noise.

The two inputs

Return on tangible capital — the quality filter. Operating earnings divided by tangible capital (plant, equipment, and net working capital — intangibles and goodwill stripped out). This is Buffett’s actual quality criterion: how much does the business earn on the physical capital actually deployed? A high ratio indicates pricing power, operational efficiency, or structural competitive advantage. It is emphatically not return on equity, which can be inflated by leverage or accounting intangibles.

Earnings yield — the cheapness filter. Operating earnings divided by enterprise value (market cap + debt − cash). The inverse of the EV/EBIT multiple. A high yield means you are paying little for each dollar of earnings. Enterprise value is preferred over market cap because it captures the actual cost to own the business, including its debt obligations.

Why two inputs

Either metric alone produces noisy results. A business with very high returns on capital may already be priced for perfection — expensive. A business with a very high earnings yield may be cheap for a reason — a structurally declining business with low and falling returns on capital. The formula filters by both simultaneously: the goal is businesses that are both good and cheap.

Joel Greenblatt notes that the “cheap-only” approach — ranking by earnings yield alone (associated with Tobias Carlisle’s Acquirer’s Multiple) — also beats the market and may produce slightly higher raw returns, but with greater volatility and drawdown. The quality overlay stabilises the portfolio emotionally, making it easier for investors to hold through periods of underperformance.

Intellectual honesty in design

Greenblatt ran the formula on backtested data starting from 1987 and chose not to iterate the design based on results. The metrics were chosen first on fundamental grounds (these are what sensible investors care about), tested once, and published as-is. This distinguishes the Magic Formula from data-mined quantitative strategies that overfit to historical noise. “It’s not the optimal result of thousands of back-tested combinations; it’s one test of something that makes sense.”

The ordered-decile result

The telling validation: when Greenblatt sorted the market into deciles by combined rank, the top decile beat the second, the second beat the third, and so on — a monotonically ordered relationship. Random variation would produce noisy decile ordering. An ordered result across all ten deciles is evidence of a real signal, not a statistical artefact of selecting the best extreme.

Limitations

The formula works as a systematic approach but is vulnerable to underperformance periods of two to three years — long enough to make most investors abandon it. Greenblatt regards this as a feature rather than a bug: persistent behavioural friction is what keeps the edge alive. If the formula always worked in the short term, arbitrage would eliminate the excess return quickly.

The formula is also agnostic about business quality beyond capital returns. A cheap, high-capital-return business facing secular disruption (a newspaper in 2005) will screen well even as intrinsic value erodes. The screen does not read the future.

Where mainstream views differ

Factor investing sceptics argue that published factors decay after discovery — the “value factor” has underperformed since ~2007. Greenblatt’s response is that the underperformance of the cheap-only value factor is different from the Magic Formula: the quality overlay has held up better because it filters out structurally declining businesses that trade cheaply for good reason.

Deep-value tradition holds that quality is already priced in and that buying at a sufficient discount provides more margin of safety than filtering by quality. Tobias Carlisle’s Acquirer’s Multiple research supports this view empirically: cheap-only may outperform cheap-plus-quality in raw terms even if the ride is rougher.

Quant critics note that backtesting from 1987 is a short window. A 35-year backtest includes the entire secular bull market of the 1990s and 2000s; out-of-sample performance (after 2005 publication) has been more mixed.

Sources