Performance variability and team interaction in basketball: a regime switching approach
In this contribution we propose a three-step procedure in order to model performance variability and team interaction in basketball. A case study using real playby-play data, coming from the 2016/2017 European Champions League has been used to describe the methodological procedure. First, we define a shooting performance index, whose alternating dynamic is modelled by Markov Switching models, borrowed from econometrics. Second, we model the probability of being in a good performance regime (improving of shooting performance) as a function of the presence of teammates on the court and identify significant (both positive and negative) interactions between players, by means of ARIMA model with covariates and networks analysis tools. Third, we validate the results, investigating how the relationships among players effectively lead to improve the team performance (in terms of scored points).
Enlace: team interaction in basketball