Does the Coach Matter? Value-Added of Coaches in NCAA Division 1 Men’s Basketball

Abstract

This paper estimates the value-added of NCAA Division 1 men’s basketball coaches using panel data from 2012–2025. I measure coaching effectiveness along two dimensions: player development, captured by changes in Bayesian Performance Rating (BPR), and team performance relative to preseason expectations, measured by changes in KenPom Net Rating. Controlling for roster talent, injuries, playing time, and teammate quality, I isolate the contribution of coaches conditional on team composition. Fixed effects estimates reveal substantial heterogeneity in coaching ability. Top coaches generate economically meaningful improvements in player performance, on the order of one BPR point, and consistently lead teams to outperform preseason projections. In contrast, observable characteristics such as age, experience, and playing background explain little of this variation. These results parallel findings in the teacher value-added literature, which also show that individual teachers have significant, persistent effects on student performance, yet observable characteristics explain very little of this variation. Extending this framework to a high-variance sports setting, this paper provides evidence that coaching matters, but that its impact is largely driven by unobservable traits.