AQR Working Paper
Researchers have long argued over whether strategies based on past stock returns have the power to explain future stock returns. Some have discovered past-return-based strategies that appear powerful for predicting future stock returns. The most important are short-term contrarian strategies, near-term momentum strategies, and long-term contrarian strategies. This paper makes several important contributions to this ongoing work.
Other research has commonly focused on these strategies one at a time. We test each past return based strategy in a common framework. We define a new three-variable specification for past returns so as to clearly distinguish the strategies discussed above.
While this paper finds no convincing evidence that either short-run or long-run contrarian strategies represent important factors for explaining the cross-section of stock returns, we do find that the properly specified one-year momentum strategy has explanatory power for stock returns when used alone, when tested against size and book-to-market, and when subjected to exhaustive robustness checks.
We extend our common framework to test past return based strategies against firm size and book-to-market, variables found by other authors to predict stock returns and to subsume the power of many other measures. We test whether other effects may be driving our results (e.g., risk changes, liquidity, bid-ask effects), and we uncover a new and powerful conditional relation between liquidity and future returns (i.e., that this relation changes based on short-term performance). Finally, we test whether our strong results are period specific.