Recognizing and Analyzing Ball Screen Defense in the NBA

As the NBA’s go-to offensive play, determining how to defend the ball screen is among the most critical decisions faced by NBA coaching staffs. In this paper, we present the construction and application of a tool for automatically recognizing common defensive counters to ball screens. Using SportVU player tracking data and supervised machine learning techniques, we learn a classi ier that labels ball screens according to how they were defended. Applied to data from four NBA seasons, our classi ier identi ied 270,823 screens in total. These labeled data enable novel analyses of defensive strategies. We present observations and trends at both the team and player levels. Our work is a step towards the construction of a coaching assistance tool for analyzing one of the game’s most important actions.