Regression to the mean (in sports)

Regression to the mean describes the statistical tendency for an extreme result, an unusually hot shooting streak, an unusually high or low ERA, a big early-season point differential, to move back toward a player's or team's true underlying level as more data accumulates. It is a general statistical phenomenon, first described by Francis Galton, that applies whenever a measured result mixes real skill with random variation.

The mechanism is straightforward: any small sample of results reflects both a player's true talent and luck, and luck by definition does not persist, so as sample size grows, the noise averages out and the observed rate converges toward the true rate. This is why metrics known to carry heavy luck components, such as batting average on balls in play and hockey's PDO, are flagged for likely regression when they sit far from a player's or league's typical range.

Analysts use regression to the mean to avoid overreacting to hot or cold starts and to build more accurate rest-of-season projections, often by blending early-season data with prior expectations. The common misreading is the "gambler's fallacy" version, expecting a specific next game or at-bat to correct a streak, when the correct reading is only that a large future sample is expected to look more typical, with no guarantee about any single next outcome.