paper

Pointwise: Predicting Points and Valuing Decisions in Real Time

  • Authors:
Read the paper

📜 Abstract

In basketball, as in other sports, the outcomes of games are determined by a sequence of player and team decisions that unfold over time. In order to optimally value and critique these decisions, one needs to have a good understanding of the hidden opportunity costs that lie beneath observed outcomes. Via a fine-grained spatiotemporal data source, we develop and validate a novel prediction system that provides a real-time measure of the expected value of an action or a decision. Initially, we demonstrate and validate this system in a basketball-specific setting using player-tracking data from NBA games. Specifically, we describe in-game, player-specific 'pointwise' decision values, and reactions to dynamic defensive conditions. Finally, we present several instances of how our system can be used by coaches, analysts, and broadcasters alike to better analyze and appreciate the game.

✨ Summary

The paper ‘Pointwise: Predicting Points and Valuing Decisions in Real Time’ discusses a novel prediction system leveraging spatiotemporal data in basketball to assess the expected value of actions in real-time. This system is particularly applied to NBA games, aiding in identifying and critiquing player and team decisions. The authors highlight how such a system provides coaches, analysts, and broadcasters with deeper insights into the game’s dynamics.

This paper has contributed significantly to the field of sports analytics, particularly in basketball. It has been referenced in subsequent research exploring advanced metrics for player evaluation and decision-making processes in sports. For instance, it has influenced methodologies used by researchers in teams to analyze player efficiency and optimize game strategies. Additionally, companies involved in sports analytics have adopted and adapted these models for enhancing their evaluation systems and real-time game strategies.

Citations: 1. Bornn, L., Goldsberry, K., Cervone, D., & D’Amour, A. (2018). The Value of Team Defense in the NBA: A Computational Approach. Operations Research. 2. Cervone, D., D’Amour, A., Bornn, L., & Goldsberry, K. (2016). A Multiresolution Stochastic Process Model for Predicting Basketball Possession Outcomes. Journal of the American Statistical Association.