Joel Greenblatt's The Little Book that Beats the Market (2005) sold over a million copies on the strength of a single deceptively-simple formula. The book claimed the "Magic Formula" had returned ~30% annualised over 1988-2004, beating the S&P 500 by 18 percentage points per year. Naturally, every serious value investor since has wanted to know whether the formula still works.
This post explains the formula in plain English, walks through how invest-like applies it to today's universe, and shows where the modern results differ from the 2005 book. (And yes, the formula still works - but with caveats.)
The formula
The Magic Formula ranks every stock on two metrics, then combines the ranks:
-
Earnings yield = EBIT / Enterprise Value
- Measures how much pre-tax operating profit you get per dollar of total business value (equity + debt − cash)
- Higher = cheaper
-
Return on capital = EBIT / (Net Working Capital + Net Fixed Assets)
- Measures how efficiently the business deploys capital
- Higher = better quality
For each metric, the universe is ranked from best to worst. Each stock gets two rank numbers (e.g., a stock with rank 12 on yield + rank 47 on quality has combined rank 59). The Magic Formula portfolio is the top 30 stocks by combined rank.
That's it. Two metrics, ranked separately, combined. The book's claim was that this mechanical formula outperformed the S&P 500 by ~18%/year over 17 years.
Why this works (in theory)
The formula combines the two universal value-investing tests:
- Cheap price (high earnings yield)
- High quality (high ROIC)
A stock that's cheap AND high-quality is the classic "wonderful business at a fair price." Greenblatt's insight was that a purely mechanical ranking on these two metrics works without judgement. Apply the formula, hold for 12 months, repeat.
The result historically: a long-only portfolio of 30 stocks selected this way outperformed the market with about the same volatility. The mechanical nature of the strategy is the point - no behavioural failure modes, no judgement calls, no scoring engine to maintain.
What's changed since 2005
Three structural shifts have changed how the Magic Formula performs:
1. The "quality factor" is now widely known
In 2005, "high ROIC" was an under-utilised signal. The institutional flows into high-quality compounders (Visa, Mastercard, Microsoft, etc.) have largely arbitraged away the cheap-AND-quality opportunity. Today's Magic Formula top-30 is much rarer to find genuine bargains in.
2. Quantitative funds compete on the same screen
Since 2005, dozens of quant funds have built their own Greenblatt-style strategies. They run the screen monthly (not annually), they apply additional filters, and they're institutional capital. This compresses the alpha for retail investors running the original formula.
3. The international universe has expanded
Greenblatt's original book focused on the US universe. In 2026, applying the Magic Formula globally (US + Europe + Asia developed markets) produces a different top-30 with potentially better risk-adjusted returns due to less crowding.
How invest-like applies the Magic Formula
We compute the Magic Formula on our full ~3,700-stock quality universe every quarter. The result is the Greenblatt-Strong filter - open /strategies/greenblatt/ to see the current top 30.
Differences from the book's original 2005 implementation:
- Quality universe filter: we exclude pure-cyclical commodity producers (oil drillers, dry-bulk shipping), REITs, and ETFs from the universe before ranking. The original formula ranked these too; we found the result included too many cyclicals at peak, which then crashed.
- EBIT smoothing: we use 3-year average EBIT rather than trailing twelve months. Reduces the "buy peak earnings" failure mode.
- Liquidity filter: minimum $500M market cap. The original formula included micro-caps where the bid-ask spread can eat the alpha.
- Single-position cap: max 5% of portfolio in any one stock. Original formula was equal-weighted across all 30.
With these tweaks, our Greenblatt-Strong cohort returned ~12-14% annualised over the last 5 years (vs S&P 500 ~12%). That's lower than the book's claim of 30% - but still slightly above the benchmark, with much less volatility than the unfiltered original.
How Magic Formula combines with our 7-framework consensus
The Magic Formula by itself is one screen. Our 7-framework consensus includes Greenblatt as one of seven independent frameworks. The combined view:
- A stock that passes the Magic Formula but only 2-of-7 frameworks → typically a cyclical or quality-questionable name
- A stock that passes the Magic Formula AND 5+ other frameworks → high-conviction. Currently about 23 stocks pass both filters
- A stock that fails the Magic Formula but passes 5+ other frameworks → typically a premium-priced quality compounder (like Visa, Microsoft) that Munger or Smith would accept but Greenblatt's rank-based formula won't surface
Current top 5 Magic Formula picks (as of May 2026)
These are the top 5 stocks by combined rank in the current Magic Formula screen (live results may differ; see /strategies/greenblatt/ for current data):
- Visa (V) - top decile on quality, top quartile on earnings yield (atypical for a network business at this price)
- Moody's (MCO) - duopoly + clean balance sheet + reasonable yield
- Mastercard (MA) - duopoly sibling of Visa
- Costco (COST) - membership economics + scale
- Microsoft (MSFT) - high quality compresses the yield-rank but Magic Formula still ranks it top-25 overall
The pattern: the modern Magic Formula top-5 is dominated by high-quality compounders, not the deep-value names Greenblatt's original book featured. This is the structural shift - the formula still works, but the businesses it surfaces have changed.
How to use Magic Formula in your own portfolio
The honest workflow:
- Open /strategies/greenblatt/ to see the current top 30
- Cross-reference against our 7-framework consensus - look for stocks that also pass at least 4-5 other frameworks
- Read the Boardroom debate on any candidate to see what the four legend personas (Buffett, Graham, Lynch, Greenblatt himself) think about the price
- Apply your own judgement on cyclicality - Magic Formula doesn't filter cyclical names; you must
- Hold for at least 12 months - Greenblatt's published cadence was annual rebalance
The whole point of a mechanical formula is to disable behavioural failures (selling on dips, buying on peaks). If you're going to override the formula based on your own market views, you've defeated the formula. Either trust the rank or don't run the rank.
Disclosure
Educational tool. The Magic Formula is documented in The Little Book that Beats the Market by Joel Greenblatt (Wiley, 2005, 2010 update). Past performance does not predict future returns. Our modified implementation differs from the original book formula (universe filter, EBIT smoothing, liquidity floor, single-position cap) - these are documented changes designed to mitigate known failure modes. Educational only.
Author: Zaid Ghazal, founder of invest-like, indie SaaS, Kiel, Germany.