Estimating team possessions in high-level European basketball competition

Abstract

The focus of this work is to estimate the number of team possessions in Euroleague basketball for seasons 2017–18, 2018–19 and 2019–20. To achieve this goal, we implemented the approaches proposed by Kubatko, et al. (2007). The statistical analysis on Euroleague data suggests a model similar to the one which is currently used in NBA with one, minor, difference. In Euroleague data, the coefficient of the percentage of the free throw attempts that ends a current possession is slightly higher than the corresponding coefficient in NBA (0.454 vs. 0.44). In a second stage, we have implemented the estimated Euroleague possessions equation in order to construct indicators for the performance evaluation of each Euroleague team. The most important indexes for team performance evaluation are the Net Rating and the Four Factors of Dean Oliver (2003). However, by applying a backwards stepwise regression analysis, a slightly improved model, in terms of fit to the data, has been identified. The new proposed “Three Pillars” model includes the offensive and defensive True Shooting percentage , Offensive Rebound percentage and Turnover percentage. Hence, this new proposed “Three Pillars” model can be used for the performance evaluation of European teams instead of the standard Four Factors model.