B0757
Title: Can we model the hot hand phenomenon? A Bayesian hidden Markov approach for assessing basketball team performance
Authors: Gabriel Calvo - University of Valencia (Spain) [presenting]
Carmen Armero - Universitat de Valencia (Spain)
Luigi Spezia - Biomathematics and Statistics Scotland (United Kingdom)
Abstract: Belief in the hot hand phenomenon in sports is commonly assumed by both media and fans. To investigate the question in the title, the shooting performance of a professional basketball team is considered and evidence of this phenomenon is sought. In particular, a Bayesian longitudinal hidden Markov model with two connected subprocesses is developed. On the one hand, the hidden process consists of a Markov chain with two possible states for each match: cold and hot. On the other hand, the observed process follows a Bernoulli distribution with two different success parameters depending on the current state of the team. The model is applied to a real data set from the Miami Heat team during the 2005-2006 season of the USA National Basketball Association. It is shown that this model can be a powerful tool for assessing the overall performance of a team during a match, particularly for quantifying the magnitude of team streaks in probabilistic terms.
