An Analysis of NBA Roster Construction in the Modern Era

Abstract

Over the past two decades, the National Basketball Association, or the NBA, has changed significantly – both in terms of the commercial standing of the league and the tactics and play styles of its teams. In light of these changes, how should NBA franchises optimally construct their rosters with players from the five playing positions? We analyze the two main considerations of an NBA roster – performance of a team’s players and overall cost of their salaries. We estimate a Cobb-Douglas production function for NBA teams using productivity from each of the five playing positions, and we also estimate optimal ratios of salary spend across positions. Through these analyses, we find that as the modern NBA has evolved, perimeter positions now feature more prominently in an optimal roster. We then use these frameworks to carry out two further analyses: first, we construct an optimal NBA roster for each of the past 18 seasons, and second, we test “what-if” roster scenarios that almost materialized in recent years to see how much better off the teams could have been. Findings here include that a Chris Paul trade to the Los Angeles Lakers in 2011 could have turned the Lakers into the NBA’s best team, while the Detroit Pistons drafting Carmelo Anthony instead of Darko Milicic in 2003 could have turned their mid-2000’s team into a dynasty.