Detecting the hot hand in amateur bowlers

The hot hand, often invoked in describing athletic performances, refers to the notion that the probability of success is conditional upon recent success or failure. The present paper describes a method of identifying hot-hand behavior in which data from amateur 10-pin bowlers were fit with a model containing dynamically varying probabilities of success and failure (hot-hand model) and one with fixed probabilities (stable model). Full-season data of 27 out of 40 bowlers were found to be better fit by the hot-hand model. Following success, different bowlers showed increased or decreased probability of success (hot-hand behavior or underperformance, respectively). Analysis of the two halves of each bowler’s data indicated that the extent of hot-hand behavior or underperformance varied across the season. Overall, results suggest that the conditional probabilities that define performance can vary in an individual over time such that anticipation of hot-hand performance can be a reasonable expectation in certain circumstances.