#MARKETS & INVESTING
The tenth of US stocks that rose most over twelve months has returned 17.04% a year since 1927. The tenth that fell most has returned −0.97%. Intuition says buy what has dropped; a century of data says the opposite.
Published on September 4, 2026
Kenneth R. French's data library sorts US stocks each month into ten groups on their return from month −12 to month −2, and publishes each group's return. Across the 1,194 months from January 1927 to June 2026, the top decile returns 17.04% a year and the bottom decile −0.97%. Cumulatively, a dollar on the winners becomes 6.3 million; the same dollar on the losers is worth 0.38, a 62% loss over a century before inflation. The annual gap between the two extremes reaches 18.0 points.
Should you buy what has fallen or what has risen?
Over a century, the top decile returns 17.04% a year and the bottom decile −0.97%. But the gap has halved since 2001, and it reverses violently coming out of a crash.
In other words: a low price is not a promise: it is often the one piece of information the market has already priced in.
Six caveats that belong with the number
- The gap is narrowing. It is worth 20.8 points a year from 1927 to 2000, but only 9.95 points over the 306 months from 2001 to June 2026.
- It is not a month-by-month regularity: losers beat winners in 438 months out of 1,194, or 36.7% of the time.
- Reversals are brutal. In April 2009 the losing decile gained 45.06% in the month while the winning decile lost 0.12%. From March 2009 to February 2010, losers +254.8% against winners +33.7%.
- The record reversal runs from July 1932 to June 1933: +357.4% for the losers against +57.5% for the winners.
- These are gross returns. A decile re-sorted every month turns over heavily, and the losing decile is the most expensive to trade: real costs are not deducted.
- The losing decile is far more volatile: 33.8% a year against 22.5% for the winning decile. Part of the gap pays for that risk and is not an anomaly.
Why it still matters: a gap measured over a century is not a gap you can collect: in between sit turnover, costs, and months when everything reverses.
For you
- When you buy a stock that has fallen, are you buying a discount or a decline that continues?
- A strategy that loses 45 points in a single month — would you hold it into the next one?
Source: reproducible Ze Finance calculation on Kenneth R. French's data library (Tuck School of Business, Dartmouth), "10 Portfolios Prior 12 2", value-weighted monthly returns, CRSP database vintage 202606 — 1,194 months from January 1927 to June 2026.
View the source →