A simple score-driven time series model for forecasting basketball match results

Paolo Vidoni, University of Udine

Abstract: This paper proposes a simple dynamic multivariate model for describing the outcome of sport matches, with particular regard to basketball. The model belongs to the class of score-driven models and it is characterized by the presence of a set of team-specific, time-varying parameters, i.e. the offensive and defensive ratings of the teams, which describe the evolution over time of the strength of the teams involved in the competition. This proposal can be viewed as a statistical dynamic extension of the popular Massey’s method for rating players and teams. Finally, an application to the Euroleague Basketball 2018-2019 is presented.