Magic Formula Investing: Greenblatt's System for Beating the Market

Magic Formula Investing: Greenblatt's System for Beating the Market

Most stock-picking systems are either too simple to work or too complex to execute consistently. Joel Greenblatt's magic formula sits in a rare middle ground: it has a clear theoretical basis, a transparent methodology, and a multi-decade track record that academic researchers have scrutinized in detail. This guide explains exactly how the system works and what you need to know before applying it.

Screen for magic formula stocks with the ValueMarkers magic formula calculator.

Key Takeaways

• Magic formula investing ranks stocks by two metrics: earnings yield (EBIT/EV) and return on capital (EBIT/tangible capital)
• Greenblatt introduced the system in "The Little Book That Beats the Market" (2005)
• Backtested returns from 1988–2004 showed approximately 30% annual returns in the US, though live returns have been more modest
• The system requires holding 20–30 stocks, replacing them annually, and tolerating periods of significant underperformance
• Financial companies, utilities, and foreign private issuers are excluded from the original screen

The Origin: "The Little Book That Beats the Market"

Joel Greenblatt published "The Little Book That Beats the Market" in 2005. Greenblatt runs Gotham Asset Management and has produced exceptional long-term returns. The book distills his investment philosophy into a framework accessible enough for retail investors while grounded in the same principles driving institutional value investing.

The core thesis is simple: buy above-average businesses at below-average prices and hold them long enough for the market to recognize their value. The innovation is making that statement operational through two quantitative rankings that a computer can sort in seconds.

Greenblatt intentionally simplified the method for "The Little Book." His institutional practice at Gotham is considerably more sophisticated. But the simplified version proved sufficient to generate strong risk-adjusted returns in backtests, which is what made the book influential.

The Two Magic Formula Metrics

Metric 1: Earnings Yield (EBIT / Enterprise Value)

Earnings Yield = EBIT / Enterprise Value

This is the inverse of the EV/EBIT multiple. Greenblatt chose EBIT over earnings per share for two reasons:

1. Capital structure neutrality. EBIT is a pre-interest figure, so it allows apples-to-apples comparison between companies with different leverage levels.
2. Tax-rate independence. Effective tax rates vary across jurisdictions and are affected by deferred tax positions. EBIT avoids this noise.

Enterprise Value (EV) = Market cap + Total debt − Cash and equivalents

Using EV instead of market cap ensures the price comparison is also capital-structure neutral. A company with $500M market cap and $300M debt has the same EV as a company with $800M market cap and no debt. Comparing EBIT to EV treats them consistently.

A high EBIT/EV means the company is cheap relative to its operating earnings — the first leg of the "good company at a cheap price" formula.

Metric 2: Return on Capital (EBIT / Tangible Capital Employed)

Return on Capital = EBIT / (Net Working Capital + Net Fixed Assets)

This measures how efficiently the business generates operating earnings from the tangible capital it employs. Greenblatt uses tangible capital — net working capital plus net fixed assets — deliberately excluding goodwill and other intangibles.

The rationale: goodwill reflects what a buyer paid for an acquisition in the past. It does not represent capital the business requires to operate. Excluding it reveals whether the underlying business earns high returns on the capital it actually needs to run.

A high return on capital means the business is genuinely good — it generates substantial operating earnings without tying up excessive capital. This is the "good company" leg of the formula.

How the Ranking System Works

The magic formula does not use absolute thresholds. Instead, it ranks every eligible stock on both metrics separately, then combines the rankings.

Step 1: Rank all eligible stocks from highest to lowest by earnings yield. The stock with the highest earnings yield gets rank 1.

Step 2: Rank all eligible stocks from highest to lowest by return on capital. The stock with the highest return on capital gets rank 1.

Step 3: Add the two ranks for each stock. A stock ranked 10th on earnings yield and 5th on return on capital gets a combined score of 15.

Step 4: Sort by combined score from lowest (best) to highest (worst). The lowest combined scores represent the best combination of cheapness and quality.

This design prevents a single extreme value on one metric from dominating the result. A company with an astronomical earnings yield but terrible returns on capital will score worse than a company with a moderately good score on both dimensions.

What Stocks Are Excluded

Greenblatt's original screen excludes several categories:

• Financial companies (banks, insurance, investment firms): Their balance sheets and earnings definitions are structurally different; the metrics do not translate cleanly
• Utilities: Heavily regulated; capital structure requirements differ from ordinary industrial companies
• Foreign private issuers trading as ADRs: Accounting standards differences introduce noise
• Stocks with market cap below $50M (original threshold; many practitioners use $100M or higher for liquidity)
• Stocks with recent fiscal year-end data more than 3 months old: Stale numbers undermine the ranking accuracy

The Backtested Returns

Greenblatt's original backtest covered US stocks from 1988 to 2004. Using a universe of the largest 3,500 US stocks:

These results are striking. Academic researchers who have replicated the backtest generally confirm outperformance, though the magnitude varies with the specific time period, universe, and methodology choices.

Several important caveats apply to the historical numbers:

• Survivorship bias. Backtests using historical databases can inadvertently exclude companies that went bankrupt or were delisted, flattering the returns.
• Look-ahead bias. Ensuring the backtest uses only data available at the time of the decision is technically difficult.
• Transaction costs. The original backtest assumed low or zero transaction costs. Annual turnover in a magic formula portfolio is high; real-world friction matters.
• Capacity constraints. A strategy managing billions cannot replicate the returns shown for a portfolio of 30 small and mid-cap stocks without moving prices.

Post-publication returns have been more modest. Several studies covering the 2005–2020 period show the magic formula beating the market by 2–4% annually rather than 18%. This is still meaningful compounding over time, but the gap has narrowed.

Applying the System in Practice

Greenblatt recommends a specific implementation:

1. Size the universe. Start with the largest 3,500 US stocks or set a minimum market cap threshold (e.g., $100M).
2. Rank and select. Use the magic formula calculator to screen the universe and identify the top 20–30 stocks by combined rank.
3. Buy in tranches. Purchase 2–3 stocks per month over 12 months rather than all at once. This reduces timing risk.
4. Hold for one year. Sell each position after approximately 12 months. Sell losers in the 11th month (to realize short-term capital losses) and winners in the 13th month (to qualify for long-term capital gains rates).
5. Repeat. Re-run the screen, select the new top 20–30, and repeat the process.

The 12-month holding period is partly driven by the tax optimization logic and partly by giving sufficient time for mean reversion to play out. Greenblatt explicitly acknowledges that the formula underperforms the market in roughly 1 out of every 4 years, and that investors who bail during those periods lock in losses.

Why the System Works (When It Works)

The magic formula captures two well-documented return factors simultaneously:

Value factor. High earnings yield stocks tend to outperform low earnings yield stocks over long horizons. This is the core of the value premium documented by Fama and French.

Quality factor. High-return-on-capital businesses tend to sustain their advantages longer than the market expects. The market systematically underestimates the durability of moats.

Most purely mechanical value strategies buy cheap stocks regardless of quality. They pick up deep-discount companies with poor return on capital, which sometimes turn out to be value traps. By requiring both cheapness and quality simultaneously, the magic formula filters out many of these traps.

Limitations and Risks

No quantitative strategy eliminates the need for judgment. The magic formula has genuine limitations:

Mean reversion in return on capital. Very high ROIC tends to attract competition and capital, which erodes the advantage over time. A stock that looks high-quality today may not sustain its returns for the year or two it takes the market to rerate it.

Sector concentration. Depending on market conditions, the top-ranked stocks can cluster in specific sectors. In certain environments, a magic formula portfolio can be significantly overweight in one or two industries, adding sector-specific risk.

Small-cap bias. The best opportunities tend to appear in smaller stocks where institutional coverage is thin. These stocks are less liquid and can gap down significantly on adverse news.

Accounting quality. EBIT can be manipulated through aggressive revenue recognition, understated depreciation, or off-balance-sheet arrangements. The formula treats all EBIT equally; a diligent investor would not.

Behavioral difficulty. Greenblatt has noted that the strategy is easy to understand but hard to execute. During periods of underperformance — which can last multiple years — almost every stock in the portfolio looks wrong and the temptation to abandon the method is intense. Investors who systematically exit during drawdowns turn a potentially strong long-run strategy into a poor one.

Magic Formula vs. Other Quantitative Frameworks

The magic formula is distinguished by its explicit two-factor structure and its specific avoidance of goodwill in the capital denominator. If you want a single-number screen that balances cheapness and quality, it remains one of the most transparent and well-documented options available.

For a deeper screen, run magic formula candidates through the full ValueMarkers fundamental analysis suite — check earnings quality scores, balance sheet strength via the Altman Z-Score, and Piotroski F-Score to stress-test the quality side of any candidate the formula surfaces.

Further reading: The Little Book That Beats the Market — Greenblatt (2005) · Magic Formula Investing · Investopedia – Magic Formula Investing