Studying Advanced Basketball Metrics with Bayesian Quantile Regression A 3-point Shooting Perspective
Taylor K. Larkin, Denise J. McManus

Abstract
In light of the recent progression in data collecting methods and the increased emphasis on the 3-point shot in today�s NBA, it is desirable to investigate the qualities of good 3-point shooters, especially in regards to more advanced player metrics. Motivated by this opportunity, we implement a regularized Bayesian quantile regression model to identify the most important non-shooting player metrics associated with the best 3-point shooters. The data used for modeling is a combination of SportVU player tracking data and other advanced metrics from the 2015-2016 NBA season. The results are positive and show support that the quantile regression model provides a more comprehensive and accurate assessment of 3-point shooters compared to using Bayesian linear regression. The application of quantile regression models on player tracking data incites opportunities for the development of more advanced analytical models that have the propensity to change the game of basketball.

Full Text: PDF