Joel Greenblatt
The Magic Formula: rank everything on quality and price, buy the best of both
The formula is two questions asked of every stock at once: is this a good business, and is it cheap? Neither question alone is safe — quality at any price buys expensive greatness, cheapness alone buys dying companies. Greenblatt's contribution was refusing to choose between them, and then publishing the uncomfortable truth that makes the method durable: it spends years at a time looking broken.
- Style
- Factor & anomaly
- Approach
- Mechanical
- Difficulty
- Intermediate
- Horizon
- Long term (years)
- Holding period
- One year per position
- Time needed
- A few hours once a year
- Markets
- US large and mid-cap stocks
- Source
- The Little Book That Beats the Market — Joel Greenblatt
The rule set
- Rank the entire universe on return on invested capital, from highest to lowest
- Rank the same universe again on earnings yield, from highest to lowest
- Buy the names with the best combined rank across both lists
- Hold each position for one year, then rebuild the list and replace what no longer qualifies
- Exclude financials and utilities — the capital measures do not mean the same thing for those businesses
What makes it distinctive
- Two inputs only, and both come straight off reported financial statements — nothing to forecast, nothing to interpret
- It requires quality and cheapness at the same time, because either one alone is a documented way to lose
- The one-year hold is part of the formula itself, not an implementation detail — the book rebalances annually by design
When it works
Broad markets with genuine dispersion in valuation — especially the aftermath of indiscriminate selling, when good businesses have been marked down along with everything else and the two rankings agree on more names than usual.
When it fails
It has underperformed for multi-year stretches long enough to make almost anyone abandon it — which is, on Greenblatt's own argument, why the edge has survived publication. Reported return on capital also flatters companies with large intangibles and off-balance-sheet obligations, so some of what ranks as quality is an accounting artefact.
How a decision moves through it
Input
Five years of key metrics, plus daily prices
Fundamentals refreshed monthly, not tick data. The formula reads reported statements; the price series exists so the exits have something to act on.
Measure
Return on capital and earnings yield
The two measures, computed per company. Return on invested capital answers 'is this a good business'; EV/EBITDA — the earnings yield read from the other end — answers 'is it cheap'.
Decide
Good business at a cheap price
In the product's single-symbol graph the two rankings become two thresholds: ROIC above 15% and EV/EBITDA below 10. Same spirit, different failure mode — the ranked version always buys the best thirty available, while a threshold version buys nothing in an expensive market.
Act
Buy, and hold for one year
The time cap is 252 trading days, matching the book's annual rebalance. The graph also exits early if EV/EBITDA rises above 14 — once the discount is gone, the reason for holding has gone with it.
Size & protect
A 30% catastrophe stop
Not in the book, which relies on the basket instead. The product ships one because a single-position graph has no basket to absorb the name that goes badly wrong.
The design is a refusal to choose between quality and price
Buy only high-quality businesses and you end up paying any price for them; buy only cheap stocks and you end up owning companies that are cheap because they are dying. Both failure modes are well documented, and the formula's entire design is a refusal to accept either. Every stock in the universe gets two ranks — one for return on capital, one for earnings yield — and the portfolio is whatever sits at the top of the combined list.
A ranking never returns an empty list. That is not a convenience — it is a design decision about what the method is allowed to have an opinion on. The formula never says 'the market is too expensive'; it says 'these are the best trade-offs available', whatever the market is doing.
Note what is missing: there is no judgement anywhere in the process. No reading of annual reports, no view on management, no opinion about the industry. The formula is what you get when you take the two questions a value investor asks about every company and hand them entirely to arithmetic.
Greenblatt published the flaw with the formula, and the flaw is the moat
The obvious objection to any published formula is that publication should kill it — if the edge is in a book anyone can buy, everyone buys the same stocks and the mispricing closes. Greenblatt's answer, printed in the book itself, is that the formula goes through stretches of underperformance long enough and painful enough that most people who try it give up. In his test period it trailed the market for years at a time, not weeks.
An edge that never hurt to hold would be competed away. This one persists precisely because holding it through the bad stretch is something most people — including professionals answerable to clients — demonstrably cannot do.
That turns the method's real requirement inside out. The formula asks nothing of your analytical ability and everything of your persistence: a few hours of work once a year, and the willingness to keep doing it in year three of looking stupid. The second part is the entire difficulty, and it is why the failure page of this dossier is the one to read first.
How this differs from the older value screens
Graham's defensive screen
- Pass/fail thresholds — can return nothing
- Safety first: four survival tests before price
- Cheap relative to assets and earnings
- Quality means 'will not fail'
Magic Formula
- A ranking — always returns a portfolio
- No survival tests; the basket absorbs failures
- Cheap relative to operating earnings
- Quality means 'earns a lot on its capital'
Both columns are marked positive because they are answers to different questions. Graham screens for companies that can survive neglect; Greenblatt ranks for the best available combination of business quality and price, and accepts that some holdings will fail. The formula is the more modern idea — it is essentially the academic quality and value factors run together, published for a retail reader three years before factor investing became a marketing category.
Five ways into this system
- The two measures, the combined rank, and the universe they run overEvery stock gets two ranks and the portfolio is the top of the combined list. The precision is in the definitions — which earnings, which capital, which companies are excluded and why.8 min read
- Equal weights, staggered cohorts, and no stop in the bookThe formula offers no basis for preferring one of its picks over another, so the sizing answer is equal weights across 20 to 30 names — and the risk control is the basket itself, not a stop.6 min read
- A once-a-year method, and the data that has to be rightA few hours once a year, but over the entire market: the formula is a cross-sectional ranking, and its data requirement is breadth rather than depth.5 min read
- The stretches that make people quit, and the measures that misleadFour failure modes, and the first is load-bearing: the formula's worst stretches are, by its author's own argument, the reason it has survived being published.7 min read
- The Magic Formula from scratch: two numbers and one hard habitThe formula asks two questions a child could phrase — is this a good business, and is it going cheap — and answers both with division. What it asks of you is not intelligence but a particular kind of stubbornness.6 min read
The ideas behind it
This system assumes you already know these. Each one is explained from scratch in Investing 101.
Compare with
- Piotroski F-ScoreTake the market's statistically cheapest stocks, score each one on nine yes/no accounting checks, and buy only the ones passing nearly all of them.
- Graham Defensive InvestorScreen for companies that are large, financially sound, consistently profitable and demonstrably cheap, then hold twenty of them and re-check once a year.
These are documented methods described for study. Nothing here is investment advice, a recommendation, or a claim about future returns — every system on this page has losing periods, and the pages say where.
Reading about a system is not having one.
Plutux is where you write your own rules down, test them against real data, and keep the record your memory would otherwise rewrite. Join the waitlist for early access.