Abstract

This paper proposes a new framework to evaluate the performance of players in the National Basketball Association (NBA) that combines both what is happening on the court and a player’s salary.  It is an important paper in the field of sports analytics because many current approaches to assess player performance do not give any consideration to a player’s salary, despite salary being a critical component of making roster building decisions.  This is especially so in professional sports leagues that institute a salary cap, such as the NBA.  The main challenge is to convert basketball activities, such as points, rebounds, and assists into dollars, which can then be used to construct a series of cash flows.  From these cash flows, traditional financial calculations follow.  In doing so, the proposed framework takes a first principles approach to allocating the estimated dollar value of a single game to each of its players, with better performing players taking a larger share.  Notably, because of its attractive statistical properties, the proposal allows an analyst the freedom to plug-and-play with various traditional measures of player performance based on preferences.  The proposed approach also allows for breakeven calculations between high-performing players with frequent missed games and average-performing players with consistent availability.  In terms of its reach, the paper has also appeared on the program of major sports analytics conferences, such as the 2025 Sloan Sports Analytics Conference and the 2024 Cascadia Symposium on Statistics in Sport.  Finally, all methods are illustrated with data from the 2022-2023 NBA regular season, with all data and replication code made publicly available.