“Hot Hand” on Strike: Bowling Data Indicates Correlation to Recent Past Results, Not Causality
ROC curves are calculated for different values of and following the stochastic simulations described in the text. The left column refers to while the right column refers to . The different rows correspond to different aggregation levels of the test (games, tournaments, seasons and career: G/T/S/C). One sees how the presented method starts to show impressive power for very small values of . However, aggregation levels of T/S/C detect streakiness for while in aggregation level G, it is not detected as designed.