Magic Formula Investing Explained
Table of Contents
Magic Formula investing tries to answer two questions at once: how much operating profit does a business earn on its tangible operating capital, and how much must an investor pay for that profit? Joel Greenblatt’s approach ranks companies on both measures, then combines the ranks. The result is a systematic way to look for profitable businesses at relatively low prices, rather than simply buying the lowest P/E stocks.
The arithmetic is straightforward. The harder work is defining the inputs, using information that was actually available on the screening date, and turning a ranked list into a portfolio with realistic costs and risks. This guide uses U.S. investment and general federal-tax context; its fictional ranking example is educational, not a current stock list.
What Is Magic Formula Investing?
Earnings yield measures operating earnings relative to the market price of the enterprise. Return on capital measures operating earnings relative to the tangible capital used in the business. Rank eligible companies from highest to lowest on each measure, add their rank numbers, and prefer the smaller combined scores.
Greenblatt describes this separate-ranking and rank-summing process in his own explanation of the method. A company can therefore outrank the cheapest stock by offering a better combination of cheapness and capital productivity. The method belongs within value investing, but a high rank is neither an appraisal of intrinsic value nor proof that a company’s advantage will last. [1]
The Two Magic Formula Ratios
1. Earnings yield: EBIT divided by enterprise value
EBIT means earnings before interest and taxes. Greenblatt’s enterprise-yield explanation uses this operating measure rather than net income. Pairing enterprise earnings with enterprise value recognizes that the business has financing claims beyond its common stock. EBIT retains depreciation and amortization; replacing it with EBITDA changes the screen. [1]
For the simple company used below, with no preferred equity or noncontrolling interests, the valuation bridge is enterprise value = market capitalization + interest-bearing debt − excess cash. Market capitalization is share price times shares outstanding. Debt includes relevant current borrowings as well as long-term borrowings. Excess cash means cash not needed for operations under the stated cash policy.
Standard enterprise-value definitions commonly subtract cash and investments and include preferred equity where present. A reconstruction must state which cash is removable and which financing claims are included; a vendor’s generic EV field need not use the same adjustments as a particular strategy. Match the scope of consolidated earnings and ownership claims, and document lease treatment consistently. [3]
With positive EBIT and EV, a higher EBIT/EV is equivalent to a lower EV/EBIT multiple. It is different from the reciprocal of the P/E ratio, which uses equity earnings and equity value. An enterprise earnings yield is also neither a dividend payment nor a forecast of the shareholder’s return.
2. Return on capital: EBIT divided by tangible operating capital
The authorized public preview of Greenblatt’s 2010 Appendix A states this equation and identifies the denominator as tangible capital employed. The relevant capital is the operating base supporting earnings, rather than the market price of the shares. Goodwill and other intangible assets are not part of this defined tangible denominator. [2]
For the transparent reconstruction in this guide, operating net working capital means operating current assets minus non-interest-bearing operating current liabilities. Receivables and inventory tie up funds; supplier payables provide operating financing. Remove excess cash and nonoperating investments from current assets. Remove interest-bearing current debt from current liabilities, so financing debt is not treated as an operating payable. Cash genuinely required for daily operations remains in the operating base under this convention. Damodaran explains these operating-versus-financing distinctions in his valuation treatment of working capital. [4]
Net fixed assets are the relevant tangible fixed assets after accumulated depreciation, represented by net property, plant and equipment in the simple example. Use a capital balance, not the year’s change in working capital or the year’s capital expenditures. A low historical book balance can make the reported return very high without establishing that new investment can earn the same return.
The public book preview does not expose the full adjustment discussion, and the official website says it makes accounting adjustments without publishing a complete calculation dictionary. The cash, balance-date and exception rules used here are explicit teaching conventions. They should not be represented as an exact reproduction of the current official screener. [2] [5]
Magic Formula Ranking Example
Consider five fictional manufacturers on the same screening date. All amounts below are in USD millions. EBIT covers the latest available trailing year; enterprise values use the same market date; tangible operating capital uses the latest available balance sheet under one consistent definition. All three inputs are positive. This five-company universe makes the calculation visible; it is not a suggested portfolio size.
| Company | Annual EBIT ($m) | EV ($m) | Tangible capital ($m) |
|---|---|---|---|
| Aster Tools | 12 | 100 | 60 |
| Beacon Pumps | 20 | 200 | 50 |
| Cedar Controls | 16 | 200 | 32 |
| Delta Parts | 18 | 300 | 60 |
| Elm Equipment | 10 | 250 | 100 |
Calculate each ratio, assign rank 1 to the highest value in each column, and add the two ranks. The percentages determine the separate orderings; they are not added together.
| Company | Earnings yield (rank) | ROC (rank) | Rank sum |
|---|---|---|---|
| Aster Tools | 12 / 100 = 12% (1) | 12 / 60 = 20% (4) | 1 + 4 = 5 |
| Beacon Pumps | 20 / 200 = 10% (2) | 20 / 50 = 40% (2) | 2 + 2 = 4 |
| Cedar Controls | 16 / 200 = 8% (3) | 16 / 32 = 50% (1) | 3 + 1 = 4 |
| Delta Parts | 18 / 300 = 6% (4) | 18 / 60 = 30% (3) | 4 + 3 = 7 |
| Elm Equipment | 10 / 250 = 4% (5) | 10 / 100 = 10% (5) | 5 + 5 = 10 |
Beacon and Cedar tie at 4, ahead of Aster at 5. Aster has the highest earnings yield, but its weaker capital-return rank offsets that advantage. A score of 4 is an ordinal screening result; it does not mean a 4% return, a four-year payoff, or a measurable margin of safety.
Specify ties before selecting stocks
For this example, break a combined-score tie by the higher unrounded earnings yield, placing Beacon before Cedar. If individual ratio values tie, assign the average of the rank positions they occupy: tied positions 2 and 3 each receive (2 + 3) / 2 = 2.5. If the combined score and earnings yield still tie, use a fixed company identifier as the final ordering rule. These are declared example rules, not verified rules of the official screener.
Rank from unrounded inputs, then round for display. A nearly identical pair can swap positions after a small data revision. Adding a company to the eligible universe can also change the ranks of existing companies even when their own ratios stay fixed. Ranks discard distance: moving from 10% to 9.9% may cost the same one position as moving from 10% to 6%, depending on the rest of the universe.
Reconcile the inputs before trusting a rank
To build Aster’s inputs, suppose its market capitalization is $90 million, total interest-bearing debt is $25 million, and total cash is $20 million. Assume $5 million of that cash is required for operations, leaving $15 million excess. Its $70 million of current assets include all that cash; its $30 million of current liabilities include $5 million of interest-bearing debt. All other current items are operating, and net fixed assets are $30 million.
EV = $90m + $25m − $15m = $100m.
Operating current assets = $70m − $15m excess cash = $55m. Operating current liabilities = $30m − $5m financing debt = $25m.
Operating NWC = $55m − $25m = $30m. Tangible operating capital = $30m + $30m net fixed assets = $60m.
Using annual EBIT of $12m gives the table’s 12% earnings yield and 20% ROC. The $5m cash requirement is an assumption for this fictional business, not a universal percentage or an official screening threshold.
Subtracting every dollar of cash would use a different operating-capital assumption. Treating the current borrowing as an operating payable would mix financing with business operations. Neither choice should be hidden inside an unexplained vendor field.
Choose the Universe and Handle Unusable Inputs
The official website excludes financial-industry companies and utilities because of their financial-statement characteristics. It describes its database as U.S.-listed equities. Additional exchange, issuer-domicile, security-type, market-capitalization and liquidity filters should be documented rather than inferred from that description. [5]
JPMorgan Chase illustrates the banking problem. Its 2025 Form 10-K reports $193.341 billion of interest income and $97.898 billion of interest expense, producing $193.341b − $97.898b = $95.443 billion of reported net interest income. Interest is central to these operations. Treating it like a manufacturer’s incidental financing item would remove a major part of the business being evaluated. This is an exclusion example, not a Magic Formula assessment of JPMorgan shares. [7]
For the reconstruction here, require positive EBIT, positive EV and positive total tangible operating capital before ranking. Missing data are missing, not zero. These conservative eligibility rules keep the illustrated ranking economically interpretable; they are not a claim about the official site’s undisclosed exception handling.
- Negative EBIT: the company is not a profitable-business candidate under this example’s rule. Two negative inputs must not create a misleading positive score.
- Zero or negative EV: do not interpret the ordinary positive earnings-yield ordering as a ranking of bargains. Examine cash availability, liabilities and business losses separately.
- Zero or negative total tangible capital: zero makes ROC undefined; a negative denominator changes its economic interpretation. Flag the case rather than assigning it a favorable ordinary rank.
- Very small positive capital: investigate an extreme ROC before relying on it. Depreciation, write-downs and classification choices can shrink the denominator.
- Negative working capital alone: this does not necessarily make the total denominator negative. Keep the working-capital component distinct from the sum after fixed assets.
Apple shows why a raw working-capital field is insufficient. Its September 28, 2024 balance sheet reported current assets of $152.987 billion and current liabilities of $176.392 billion: $152.987b − $176.392b = −$23.405 billion of unadjusted working capital. That subtraction includes cash, marketable securities and current debt classifications that still need examination. It neither supplies the strategy’s operating-capital denominator nor establishes Apple’s current eligibility or rank. [6]
Magic Formula vs ROIC, P/E and EV/EBITDA
Similar labels can conceal different numerators, denominators and decisions. Keep the named strategy’s two-factor ranking separate from the broader valuation and profitability questions below.
| Measure | What it measures | Difference from this strategy |
|---|---|---|
| Return on invested capital | Commonly NOPAT divided by average invested capital. | An after-tax operating-return framework; it is not automatically Greenblatt’s pre-tax tangible ROC. |
| P/E or its E/P reciprocal | Equity price relative to earnings after interest and taxes. | Changes the earnings claim and price denominator used in EBIT/EV. |
| EV/EBITDA | Enterprise value relative to earnings before depreciation and amortization as well as interest and tax. | Its reciprocal is EBITDA/EV, not EBIT/EV. |
| Fundamental analysis | Financial statements, business economics, valuation and risks. | The broader research process can investigate what a two-ratio screen leaves unexplained. |
In particular, do not compare this pre-tax tangible ROC directly with an after-tax WACC and call the difference economic profit. The existing ROIC framework requires a consistently defined after-tax operating return and capital base. Nor does a higher ROC establish that today’s purchase price is attractive; that is why the strategy has a separate price-related ranking.
How to Implement and Rebalance a Magic Formula Screen
- Fix the investable universe and definitions. Record exclusions, minimum size and liquidity, EBIT period, cash policy, capital balance date, exceptional-input rules and ties.
- Use information available at the decision time. A fiscal year-end is not the date its accounts became public. Record the release or filing date, vendor availability and market-price timestamp; allow a realistic delay before trading.
- Reconcile unusual observations. Check the underlying financial statements, debt, one-time gains or losses, acquisitions, leases and denominator changes. Apply adjustments consistently across the universe.
- Rank both ratios and combine the ranks. Save the input version and selected list so later revisions do not silently rewrite the decision.
- Set portfolio and trading rules. Specify position sizes, industry limits if any, execution assumptions, treatment of existing holdings and a reassessment schedule. Adding constraints creates a variant whose outcomes may differ.
The official website discusses a basket of 20–30 holdings, equal-dollar positions and holding for approximately one year before reassessing with a fresh screen. It does not call for selling immediately whenever a holding drops off a later list. Equal starting dollars also do not keep weights equal as prices change. [5]
Turnover creates costs even with zero advertised commissions. As a hypothetical cost check, trade $100,000 of sales and $100,000 of purchases to replace a $100,000 portfolio, with costs paid separately. Assume execution drag of 0.20% on each traded side. Selling costs $100,000 × 0.002 = $200; buying costs another $200. Total drag is $400 / $100,000 = 0.40%, before taxes or other fees. This is 100% of the portfolio sold plus 100% bought; quoted turnover conventions may count only one side. Actual spreads, slippage and impact depend on the securities and trade size.
In a U.S. taxable account, sales can realize gains and losses. The general federal long-term rule requires holding an investment more than one year, rather than merely reaching one year. Loss sales followed by substantially identical purchases within 30 days before or after the sale can trigger wash-sale restrictions. Coordinate tax lots, repeat selections and purchases in other accounts instead of assuming an annual screen automatically delivers a tax benefit. [9] [10]
Does Magic Formula Investing Work?
Its economic idea is understandable: combine a price discipline with an operating-profitability discipline. That rationale does not establish an achievable future excess return. Greenblatt’s historical explanations discuss tested results, but an author’s reported backtest, a later provider’s reconstruction and an investor’s actual account are different evidence. The SEC’s investor bulletin stresses that backtested performance is hypothetical. This article supplies no new backtest or promised annual return. [1] [11]
Before treating any performance chart as evidence, ask how it handles the following:
- Survivorship: did the historical universe include companies later delisted, acquired or bankrupt, with their actual terminal outcomes? Starting from today’s surviving stock list changes the test.
- Look-ahead: were financial statements and subsequent restatements used only after they became available? A December balance sheet cannot simply be assumed available on December 31.
- Data mining: were the universe, cash formula, rebalance month and extra filters chosen after seeing which version performed best? Reserve later data for evaluation rather than repeatedly tuning to the same history.
- Implementation: could the portfolio trade at the assumed prices and sizes? Include spreads, trading impact, cash balances, turnover and applicable fees; explain the tax assumptions.
- Comparability: do strategy and benchmark both include dividends and comparable cost assumptions over the same periods? Examine drawdowns and long stretches of underperformance as well as a compound return.
CFA Institute identifies these data and implementation pitfalls in quantitative equity investing. These are questions to ask of a study, not an allegation that every Magic Formula test contains them. Out-of-sample evidence and a documented live record improve the assessment, while neither removes future uncertainty. [8] [11]
Common Mistakes
- Substituting a convenient ratio. Net income/market cap and generic after-tax ROIC do not reproduce EBIT/EV and pre-tax tangible ROC.
- Adding percentages. The method combines ordinal ranks; adding 12% yield to 20% ROC constructs a different scoring rule.
- Using peak earnings without inspection. A cyclical profit surge or disposal gain can improve both ratios at once. Check whether the reported EBIT describes repeatable operations.
- Confusing cash with automatically removable cash. A restricted or operationally necessary balance is not interchangeable with freely distributable excess cash.
- Forcing unusable inputs into the rankings. Zero denominators and missing data need explicit exception rules, not a zero fill or a favorable artificial score.
- Changing the rules after losses. Selective overrides, extra filters and faster trading create a different strategy. Record them and assess their consequences rather than borrowing a published record.
Limitations of Magic Formula Investing
A business can be cheap because its earnings are about to fall. A high tangible ROC can reflect a small accounting denominator. Combining the ranks does not resolve either problem or prevent a permanent loss.
The ratios omit much about future demand, competition, governance, financing stress and free cash flow. EBIT is an accounting profit measure; neither its enterprise yield nor its return on tangible capital deducts all future reinvestment needs. Excluding goodwill and intangible capital can also make comparisons favor some business models over others without proving superior economics.
A basket may concentrate in the same out-of-favor industries. Holding more names does not ensure diversification of common economic risks. Industry caps, larger-company filters and liquidity requirements may help align a portfolio with its constraints, but they also change the strategy being evaluated.
Finally, systematic rules still require human discipline and usable data. The strategy can lose money or underperform for extended periods; the official website itself disclaims guaranteed success. A long holding horizon does not convert the screen into a safe investment. Assess the method alongside your portfolio’s risk capacity and a suitable benchmark. [5]
Frequently Asked Questions
Sources
- Forbes: Joel Greenblatt Briefing Book, 2010. Greenblatt’s direct interview, including the January 22, 2010 discussion on printed pages 3–6; EBIT/enterprise yield and separate combined ranks.
- Joel Greenblatt: The Little Book That Still Beats the Market, Appendix A, Wiley 2010; licensed O’Reilly public preview. Visible return-on-capital definition only; the full appendix is not available in that preview.
- CFA Institute: Market-Based Valuation: Price and Enterprise Value Multiples, 2026 public summary. Standard EV claims, valuation multiples, data timing and screening limitations.
- Aswath Damodaran, NYU Stern: Working capital in valuation. Operating cash, noncash working capital and the exclusion of financing debt.
- MagicFormulaInvesting.com: Frequently Asked Questions and How it Works. Official portfolio process, exclusions, accounting-adjustment notice and risk disclaimer; undated living pages.
- Apple Inc.: October 31, 2024 results release, SEC Exhibit 99.1. Unaudited condensed balance sheet as of September 28, 2024.
- JPMorgan Chase & Co.: 2025 Form 10-K, Note 7. Reported interest income, interest expense and net interest income for the year ended December 31, 2025.
- CFA Institute: Active Equity Investing: Strategies, 2026 public summary. Quantitative-investing process and backtest/data/turnover pitfalls.
- IRS: Topic no. 409, Capital gains and losses. General holding-period distinction; annual tax brackets are not reproduced here.
- IRS: Publication 550 (2025), Investment Income and Expenses, Wash Sales. General loss-sale and substantially identical repurchase restrictions.
- SEC investor education staff: Investor Bulletin: Performance Claims, September 15, 2022. Hypothetical results, costs, methodology and benchmark comparisons.
Sources accessed November 15, 2026. Ranking inputs and execution costs are hypothetical and independently calculated. Apple and JPMorgan examples are dated accounting illustrations, not current screen selections or recommendations.
Disclaimer
This article is for educational and informational purposes only and does not constitute investment, tax or financial advice. Screening ranks and historical or hypothetical results do not guarantee future performance. Investments can lose value, including permanently. Consult a qualified financial or tax professional about your circumstances.
