Welcome to the source code repository for the blog post, “Animating Expected Possession Value (EPV) in the NBA”.
In a MIT Sloan Sports Analytics Conference Research Paper, Cervone, D’Amour, Bornn, and Kirk Goldsberry create a metric called Expected Possession Value, EPV. They use optical player tracking data to calculate the number of points the offense is expected to score by the end of the possession in real time, given everything we know now. In other words, EPV of a possession is the weighted average of the outcomes of all future paths that the possession could take (Cervone et al.)
The source code for creating the plots and figures in the blogpost is contained in final-result.Rmd.