Three patterns in stock returns have survived decades of scrutiny. Each has also lost half or more of its value in a stretch most people could not sit through. The evidence, the crashes, the cost of trading them, and what publication does to an edge.
A factor is a characteristic of a stock that has been tied, on average and over long periods, to a different return — cheapness, recent performance, the quality of the business. Three of them have the kind of evidence a serious person has to take seriously: decades of data, many countries, and survival after publication. Each of them has also, at some point, lost half of its value or more in a stretch that lasted years. This page is both halves. The second half is the one the sales material leaves out.
Value
In 1992 Eugene Fama and Kenneth French showed that two easily measured variables, a company's size and its ratio of book value to market value, together captured most of the difference in average returns across U.S. stocks — and that the textbook risk measure, beta, did not. Their 1993 paper turned that into factors: a size factor (small minus big, SMB) and a value factor (high book-to-market minus low, HML). Cheap stocks, measured that way, had earned more than expensive ones over the decades studied. The data behind both papers is published free on French's website and updated monthly, which is part of why the finding has been checked so thoroughly.
The drawdown: value underperformed growth from 2007 onward, and by mid-2020 the value factor's drawdown stood at 55% — thirteen years of falling behind, including, in AQR's words, "a particularly painful period" from 2017 to 2020. Whether value was dead became an academic debate with papers on both sides. Then 2022 was, again in AQR's words, "extremely strong" for value. A holder who sold in 2020 had sat through the entire loss and none of the recovery.
Momentum
In 1993 Narasimhan Jegadeesh and Sheridan Titman documented that stocks which had risen over the previous six to twelve months went on to beat stocks which had fallen, for the next several months — about 1% per month for the strategy of buying recent winners and shorting recent losers, and a compounded 12.01% a year for the six-month version over 1965–1989. It has since been examined in many other markets, and it is the factor that most offends the idea that prices already reflect everything.
The drawdown: Kent Daniel and Tobias Moskowitz called it a momentum crash, and it arrives at the worst possible moment — when a market rebounds after a collapse. From March to May 2009 the U.S. market rose 26%. The stocks the momentum strategy was short, the previous year's losers, rose 163%; the winners it was long rose 8%. The winners-minus-losers portfolio lost 42.28% in March 2009 and 45.52% in April. The precedent was 1932, when the same portfolio lost 60.98% in July and 74.36% in August as the losers rallied 232%. Momentum's long record is real; so is the fact that its worst months are concentrated exactly where a frightened holder is least able to stay.
Quality
Asness, Frazzini and Pedersen's "Quality minus junk" defines quality as what a sensible investor would pay more for — profitability, growth and safety — and finds that a portfolio long high-quality stocks and short junk "earns significant risk-adjusted returns in the United States and across 24 countries." Of the three it has had the gentlest ride, and the shortest published history, which is a reason for less confidence, not more.
The three, side by side
| Factor | Where the evidence starts | The stretch that hurt | What a holder sat through |
|---|---|---|---|
| Value (cheap minus expensive) | Fama and French, 1992–93 | 2007 to mid-2020 | A 55% drawdown over thirteen years, then a sharp recovery |
| Momentum (winners minus losers) | Jegadeesh and Titman, 1993 | March–May 2009; July–August 1932 | Monthly losses of 42% and 46% in 2009; 61% and 74% in 1932 |
| Quality (good businesses minus junk) | Asness, Frazzini and Pedersen, 2019 | Milder so far | A shorter record, which is its own warning |
Two things that happen after a factor is found
It gets traded, and trading costs money. Frazzini, Israel and Moskowitz used roughly a trillion dollars of a large manager's live trades across nineteen markets to measure real costs. Their finding was that size, value and momentum "can be deployed at very high asset size and still survive trading costs" — but that short-term reversal, one of the more exciting anomalies in the literature, does not. A factor's paper return and its net return are different numbers, and the gap depends on who is trading and how much.
It gets published, and publication eats part of it. McLean and Pontiff tracked 97 published return predictors. Their returns were 26% lower out of sample than in the papers that discovered them, and 58% lower after publication — a third of the decline from the data mining that found them, the rest from investors acting on them. Harvey, Liu and Zhu, surveying hundreds of published factors, concluded that a new one should clear a t-statistic of 3.0 rather than 2.0 and that "most claimed research findings in financial economics are likely false." The three on this page clear that bar; most of the factor zoo does not.
What this has to do with a futures desk
Nothing directly — the desk's systems are not factor portfolios, and nothing here is a suggestion to build one. The lesson transfers anyway, and it is the most important one on this shelf: every real edge has a stretch that looks exactly like death. Value's thirteen years. Momentum's two months. If you do not decide in advance what evidence would make you stop, you will decide it at the bottom, where the evidence is loudest and most wrong. The desk's own answer is to post each system's expected numbers — profit factor, win rate, worst drawdown — before it is dealt, and then to re-measure every hand against that posted expectation on a schedule, so that drift from what members were told is the signal and a person, not a job, acts on it. The drawdown page is mostly arithmetic about how long the bad stretches are entitled to be.
- The cross-section of expected stock returns — Fama and French, Journal of Finance 47(2), 1992
- Common risk factors in the returns on stocks and bonds — Fama and French, Journal of Financial Economics 33, 1993
- Kenneth R. French data library — the factor series, free, updated monthly
- Returns to buying winners and selling losers — Jegadeesh and Titman, Journal of Finance 48(1), 1993
- Momentum crashes — Daniel and Moskowitz, NBER working paper 20439, 2014; Journal of Financial Economics 122(2), 2016
- Reports of value's death may be greatly exaggerated — Arnott, Harvey, Kalesnik and Linnainmaa, Financial Analysts Journal 77(1), 2021
- Value: why now? Capturing the comeback in its early innings — AQR, 2023
- Quality minus junk — Asness, Frazzini and Pedersen, Review of Accounting Studies 24(1), 2019
- Trading costs of asset pricing anomalies — Frazzini, Israel and Moskowitz, AQR working paper, 2012
- Does academic research destroy stock return predictability? — McLean and Pontiff, Journal of Finance 71(1), 2016
- … and the cross-section of expected returns — Harvey, Liu and Zhu, Review of Financial Studies 29(1), 2016
Education, not advice. Nothing on this page recommends buying or selling anything, and nothing is a promise of results — every figure is cited above or hypothetical. TOTH does not provide investment advice. New to the vocabulary? The glossary has plain definitions.