Performance Persistence and Survivorship Bias

If alpha is positive this year, will it be positive next year? Persistence is the condition under which past performance is a useful predictor of future performance — and therefore the condition under which investors should deviate from a passive strategy. Without persistence, even a fund with a long history of high alpha offers no forward-looking information. This note develops the empirical evidence on persistence, the Berk–Green theory of why competition should erode it, the momentum explanation for spurious persistence found by Carhart (1997), and the survivorship bias that inflates measured alpha in any database of fund returns.

1. Testing persistence: the contingency table
2. Carhart's momentum explanation
3. Elton–Gruber–Blake evidence on persistence
4. The Berk–Green theory
5. Survivorship bias
6. Characteristics that predict future performance