Leveraging expected action value to measure on-court decision-making skills in basketball

Quantifying decision-making in professional basketball has been a challenging area of research in the past decade with potentially fruitful insights to be drawn as NBA organizations seek to better understand cognitive aspects of athlete performance. In this paper, we develop an objective framework for evaluating the decision-making capabilities of individual athletes, also enabling us to make inferences about team strategy and execution efficacy. Our method leverages high resolution tracking data to quantify the expected point value of the immediate potential actions a ballhandler can take, building on the existing concept of an Expected Possession Value (EPV) metric. Unlike the traditional EPV approach that models a single expected point value over the entire possession’s time horizon, our evaluation is constrained to immediate pass and shot actions, with a value associated with each such potential action. We introduce a novel Expected Action Value (EAV) metric to capture these instantaneous expectations, and leverage it to identify scoring opportunities throughout a game. We analyze these opportunities as instances of decision-making, quantifying how often those opportunities are missed along with the potential payoff associated with each missed opportunity. Looking at team opportunities as a whole and relying on the notion of expectation, we estimate how much of a team’s performance can be attributed to their strategy (creating opportunities) versus their execution (capitalizing on opportunities). Through this analysis, we demonstrate EAV as an effective framework for quantitatively evaluating decision-making performance via tracking data.