Magic Formula Stocks

What Is the Magic Formula?

Joel Greenblatt introduced the Magic Formula in his 2005 book The Little Book That Beats the Market. The premise is deliberately simple: buy good businesses at cheap prices, hold them for a year, and repeat. "Good" is measured by return on invested capital (ROIC). "Cheap" is measured by earnings yield — earnings before interest and taxes divided by enterprise value (EBIT/EV), inverted from the more familiar EV/EBIT ratio.

The screen works by ranking every stock in the universe separately on each dimension, then adding the two ranks together. The companies with the best combined rank — not necessarily the best on either metric individually — make the list. A mediocre business at a very cheap price and an expensive business with extraordinary returns can score similarly; Greenblatt's insight is that the combination is what drives returns.

Historically, Greenblatt's own research showed the strategy generating substantial excess returns over the S&P 500 from 1988 to 2004. Academic follow-up studies have been more mixed. The strategy works best during periods when value investing broadly outperforms — it struggled significantly during the 2010s growth run — and tends to produce a portfolio of unloved, unglamorous companies that are psychologically difficult to hold through drawdowns. Greenblatt himself emphasizes that the strategy underperforms the market roughly one out of every three to four years, which is precisely why it cannot be arbitraged away.

Common exclusions in a faithful implementation: financial companies (banks, insurers) and utilities, where capital structure makes ROIC and earnings yield comparisons unreliable. Very small-cap stocks are often excluded as well because liquidity makes execution impractical.

The criteria in plain English:

• High ROIC: the business earns strong returns on the capital it deploys
• Low EV/EBIT: you are paying a low multiple of operating earnings for the business
• Combined rank: you want companies near the top of both lists simultaneously

The strategy has no fundamental quality filter — it will surface turnaround candidates, cyclical companies at trough earnings, and businesses in secular decline. That is a feature, not a bug, but it means position sizing and sector diversification matter more here than in a screen with explicit quality guardrails.

---

How to Run This Screen in GeminIQ

GeminIQ's screener lets you filter on both ROIC and EV/EBIT directly from as-filed SEC data. Here is how to replicate the Magic Formula logic:

Step 1. Open the GeminIQ Screener

Step 2. Add the following filters:

Filter	Operator	Value
ROIC TTM	≥	25
EV/EBIT TTM	≤	15
EV/EBIT TTM	≥	1

The EV/EBIT TTM floor of 1 removes companies with negative or zero operating earnings, which produce misleading low multiples.

Step 3. Set Sort By to EV/EBIT TTM, direction Ascending — cheapest first.

Step 4. Review results. For a true Magic Formula implementation, you would want to rank by the combined ROIC + earnings yield rank rather than sorting on either alone. Use the results as a starting pool and review each company's source 10-K or 10-Q in GeminIQ to verify the EBIT figure is not distorted by one-time items.

GeminIQ Tip: Use the Calculated Metrics view to see ROIC trended over 8–12 quarters. A company with a high trailing ROIC that has been declining for three straight quarters is a different investment than one with a stable or rising ROIC at the same headline number.

---

What Aggregator Data Misses for This Screen

The Magic Formula depends on two calculated figures: ROIC and EV/EBIT. Both are sensitive to how capital and earnings are defined — and aggregator normalization choices materially affect which companies appear on the list.

Invested capital distortion. Aggregators frequently capitalize operating leases into debt and right-of-use assets when calculating invested capital. Under ASC 842 (effective for most companies after 2019), lease obligations are already on the balance sheet. An aggregator that double-counts lease capitalization will inflate invested capital, suppressing ROIC — and could push a genuinely capital-light retailer or restaurant company off the screen entirely.

EBIT normalization. Many aggregators adjust EBIT for restructuring charges, stock-based compensation, or amortization of acquired intangibles. A company carrying $300M in acquisition-related intangible amortization will show a much higher "adjusted EBIT" on an aggregator than its as-filed EBIT. GeminIQ surfaces the as-filed EBIT figure, which is what Greenblatt's original formula uses. If you are comparing results to another screener and seeing different companies, this is the most likely explanation.

Enterprise value. Minority interest treatment, pension obligations included or excluded from net debt, and how deferred revenue is handled all affect EV. Aggregators make different choices. GeminIQ's enterprise value is constructed from as-filed balance sheet components with full XBRL traceability, so you can audit exactly what went into the number.

---

GeminIQ builds its financial statement database from raw SEC filings, not from third-party financial data APIs.

This screen is educational and does not constitute investment advice. Past performance of any strategy does not guarantee future results.