The Little Book That Still Beats the Market Ch. 5: Where the Formula Breaks Down
阅读中文版Value traps, accounting distortions, and what happened to the magic formula out of sample after 2005.
🔊 Listen to Article (Chinese Audio)
The Little Book That Still Beats the Market Ch. 5: Where the Formula Breaks Down
"No widely known quantitative strategy earns its backtested return going forward." — a standing caution in quantitative investing
Investment Context
The first four chapters explain why the formula works. This one addresses a more important and far less discussed question: when it does not.
The inherent risk of a bestseller is that readers retain the 30% backtest figure without the conditions that produced it. Greenblatt does warn in the book that the strategy will fail at times, but that warning travelled far less widely than the performance number.
The Wall Street Translation
1. The Out-of-Sample Record
After publication in 2005 the formula entered a widely-known, out-of-sample phase. Independent studies and Greenblatt's own managed accounts indicate that it continued to work after 2005, but with markedly lower excess returns than the backtest and longer stretches of underperformance.
Two non-exclusive explanations: capital flowed in once the strategy was public, compressing the opportunity; and the original test window covered a period unusually favourable to small-cap value.
The pattern itself is worth remembering: nearly every published quantitative strategy underperforms its backtest in live trading. On encountering any backtested figure, the reasonable response is to discount it before doing anything else.
2. Three Systematic Failure Modes
| Failure mode | How it appears | Why the formula misses it |
|---|---|---|
| Cyclical peak trap | Cyclicals show inflated earnings at the top, ranking well on yield | The formula reads current EBIT, not its sustainability |
| One-off gains | Asset sales or litigation settlements inflate EBIT | These land in operating profit on the statements |
| Structural decline | The business is genuinely dying, so cheapness is correct | The formula cannot separate "temporarily disliked" from "terminal" |
The third is the essence of a value trap: cheapness is sometimes the market's error and sometimes its accurate judgment, and the formula cannot tell which.
3. Diversification Is the Only Remedy
Greenblatt's answer to all of this is not additional screens but holding enough names. With 30 positions, three or five value traps can be absorbed by the rest.
This stands in direct opposition to Common Stocks and Uncommon Profits: Fisher argued for deep research into few companies held concentrated; Greenblatt argues for shallow understanding of many companies held broadly. Both are coherent, but mixing them fails on both counts — running Greenblatt's shallow research at Fisher's concentration is the most dangerous combination available.
Actionable Trading Rules
- Discount every backtest: Halve the backtested excess return of any public strategy to get a realistic forward expectation. Execute only if it still looks worthwhile after the haircut.
- Never reduce the number of holdings: Diversification is the formula's only defence against value traps. Holding fewer than 20 means absorbing single-stock risk the method was never designed to carry.
- Manually exclude obvious cyclical peaks: If a name sits in an industry at a recognised boom peak — shipping rates at historic highs, energy prices after a spike — drop it. This is one of the very few justified human overrides.
Relevance to a Retirement Portfolio
The lesson generalises: any product sold on historical returns deserves the question "how unusual was that history?"
This matters most in the withdrawal phase, where you cannot console yourself that things revert in the long run — your horizon is finite and non-repeatable. That is the fundamental reason the core of the portfolio belongs in broad index funds.