Recently, the "hot hand" phenomenon regained interest due to the availability and accessibility of large scale data sets from the world of sports. In support of common wisdom and in contrast to the original conclusions of the seminal paper about this phenomenon by Gilovich, Vallone and Tversky in 1985, solid evidences were supplied in favor of the existence of this phenomenon in different kinds of data. This came after almost three decades of ongoing debates whether the "hot hand" phenomenon in sport is real or just a mis-perception of human subjects of completely random patterns present in reality. However, although this phenomenon was shown to exist in different sports data including basketball free throws and bowling strike rates, a somehow deeper question remained unanswered: are these non random patterns results of causal, short term, feedback mechanisms or simply time fluctuations of athletes performance. In this paper, we analyze large amounts of data from the Professional Bowling Association(PBA). We studied the results of the top 100 players in terms of the number of available records (summed into more than 450,000 frames). By using permutation approach and dividing the analysis into different aggregation levels we were able to supply evidence for the existence of the "hot hand" phenomenon in the data, in agreement with previous studies. Moreover, by using this approach, we were able to demonstrate that there are, indeed, significant fluctuations from game to game for the same player but there is no clustering of successes (strikes) and failures (non strikes) within each game. Thus we were lead to the conclusion that bowling results show correlation to recent past results but they are not influenced by them in a causal manner.
All Science Journal Classification (ASJC) codes
- Biochemistry, Genetics and Molecular Biology(all)
- Agricultural and Biological Sciences(all)