Fig 1.
(A) The permutation test [6] removes the streak selection bias in small sequences. (B) In an actual game environment we need to consider the shot quality. Our shot make probability model outputs accurate out-of-sample probabilities.
Table 1.
Kemba Walker shows significant levels of streakiness, even after adjusting for errors in the shot make probability model.
Table 2.
While overall the league exhibits shooting regression (last column), there are players that have the hot hand (HH) to different degrees (second column).
Out of the 153 qualified players for our statistical tests at least 24 players provided statistical evidence for streakiness at the 5% significance level for all values of k examined. The probability of obtaining at least 24 false positives at the 5% level is less than 1-in-1,000,000.