Development of value metrics for specific basketball contexts: a positional approach for the defensive rebound value

Abstract

Over the past decade, NBA teams are making an increasing investment in the development and growth of data analytics to help each team's ultimate goal of winning the championship. This growth has been accompanied by improvements in data collection from the league's games, moving from superficial, manually annotated statistics - box scores - to much richer data that can provide much more information - the optical tracking data. This study aims to analyze a very specific facet of the game: defensive rebounding. This section of the game is one of the least information that can be collected manually. Using optical game tracking data from the 2020/2021 season, we will try to capture the contribution that players have in achieving the rebounding for the team. In this work, multiple metrics have been developed to explain different dimensions of the defensive rebounding process, in addition to different analyses of these metrics. This has allowed us to discover different player profiles in this phase of the game and team behaviors that were invisible to the eyes of traditional statistics.