Thesis - An Effort To Reevaluate Individual Defense In Nba Games
Purpose - Alleviate team-success bias present in some NBA defense metrics.Objective - Find regression lines between team defensive stop rate, foul rate, and forced turnover rate and wins for playtypes tracked on NBA.com. Resulting slopes were used as weight coefficients multiplied by centered player rates. The results were summed, and then multiplied by proportion of team minutes played to give an estimate for wins added by playing more and playing better.Results - Gaps in data used and methods used leave room for improvement. Some data was available for teams but not players on NBA.com, and some slopes had incorrect signs due to sample size issues. The model was systematically overrewarding a lack of fouls committed, leading to the top of the list including many poor defensive guards who do not foul. A positional adjustment could fix this. Four versions of the model were fit, all identifying Draymond Green as the NBA's best defender in 2023.
