Identifying Defensive Matchups and Their Effect on Scoring
This project introduces new defensive metrics for the NBA, based on statistical modeling of optical player tracking data. We introduce counterpoints--a measure of "points against" for defenders--as well as defensive shot charts, which track how defenders impact shot selection and effectiveness throughout the court.Blog Posts
A New Microeconomics for the NBA
This project uses optical player tracking data to estimate, in real time, the expected number of points obtained by the end of a possession. This quantity, called expected possession value (EPV), derives from a stochastic process model for the evolution of a basketball possession, and acts as a stock ticker of a possession's value as it unfolds. By tracking EPV, we are able to quantify the effect of every action, movement, and decision made during a game, revealing new ways in which players affect their teams' possession outcomes.Blog Posts
Metrics for Metrics
While new sports metrics are changing our understanding of the sport, they can also make finding the right metric to support a particular decision like finding a needle in a haystack; this clutter threatens the untility of advances in sports analytics. Meta-Analytics analyzes sports statistics themselves, and introduces metrics for metrics: reliability, discrimination, and independence.
A Spatial Glossary of Shot Types
Modeling NBA shot attempt data as a point process, we create a low dimensional representation of offensive player types using non-negative matrix factorization (NMF), an unsupervised dimensionality reduction technique. The spatial representations discovered by the algorithm correspond to intuitive descriptions of NBA player types, and can be used to model other spatial effects, such as shooting accuracy.
The Value of Ball Movement in the NBA
Throughout the NBA, coaches consistently stress the importance of ball movement in a functional and efficient offense: “either you move it or you die,” according to San Antonio Spurs coach Gregg Popovich. To test Pop's theory, we introduce entropy and opportunity metrics, and show how teams and players balance capitalizing on immediate shot opportunities with entropic strategies yield better scoring opportunities down the line.
Identifying NBA Tactics and Set Plays from Tracking Data
Every NBA team employs a small army of scouts, video coordinators, and analysts to dissect the strategic tendencies of their competition. This project integrates optical player-tracking data with emerging computational methods to offer a brand new scouting mechanism for the so-called “big data” era. Integrating new techniques in machine learning, statistics, and visualization, we "reverse engineer" the playbook for every team in the NBA, easily finding each similar instance of a particular play's ball and player movement.
Inferring the Value of Positioning and Spacing
Continuously throughout a basketball possession, offensive and defensive players battle for control of different regions of the basketball court. From this competitive dynamic alone, we can infer the implicit value of positioning and spacing among NBA players and teams. This allows us to quantify the NBA court real estate market, enabling new insights and metrics for both offense and defense.