Understanding links between biomechanical data and on-court NBA production

About the Project

Over the past decade, resources poured into analytics efforts and performance staff in the NBA (comprising sport scientists, strength coaches, physical therapists, dietitians, athletic trainers, and physicians) have consistently grown. However, despite continued investment in both of these practices, the understanding of relationships between human performance data and on-court productivity has not yielded the expected return. As a field, our understanding of the physical traits that comprise a competent NBA athlete is limited by a lack of consistent data collection processes, team staff churn, the inherent complexity of studying the human body, and the diversity of player roles in basketball.

Beginning in 2013, P3 (Peak Performance Project) began collecting biomechanical data on professional and collegiate basketball players performing ballistic movements in their laboratories. As it currently stands, P3 owns the largest biomechanics database on NBA athletes in the world. This dataset comprises 3D Motion Capture and Force Plate data – allowing P3 to create skeletal models of their athletes for in-depth movement analysis. While many of P3’s efforts to date have involved studying injury-risk and training adaptations within this cohort, the extent to which this data can help inform which physical tools are important in today’s NBA landscape remains largely unknown.

P3’s in-house dataset consists of 3D Motion Capture and Force Plate data conducted on professional and collegiate basketball players – approximately 1300 assessments in total. This data has been collected on a series of vertical plane (jumps) and lateral plane (lateral acceleration drills) actions in P3 labs in Atlanta, GA and Santa Barbara, CA. In addition to P3’s in-house data, scraping publicly available basketball data web will be integral to project success.

For this project, students will build a web-scraping tool for the purposes of comparing P3’s in-house data with on-court production at the professional and collegiate levels, build ML models to understand links between biomechanics data and on-court production, and develop an R Shiny application to visualize relationships between P3 data and on-court performance.