Game-to-game variability of technical and physical performance in NBA players

This study aimed to compare the differences in game-to-game variability of technical and physical parameters of basketball players, according to game location and game outcome. Game data (n=712) were collated from the official box-score and player-tracking of the 2013-14 NBA regular season. The players were separated according to their court specific position. A two-step cluster analysis was performed to group players according to time played into 3 groups: short, medium, and long-time played. The coefficient of variation (CV) was calculated from game-to-game parameters. The results showed that short-time players demonstrated great CV for technical and physical statistics, long-time players displayed larger CV in losing and away games, technical indicators such free-throws revealed substantial variability in losing games and guards and centers presented high CV for some performance statistics. CV values were inversely proportional to time played, probably because playing less time decreases the probability of maintaining stable performance across games. Long-time players displayed larger CV in away and losing games, possibly due the constraints imposed by opponent teams. Free-throws seems to be the variables that best discriminate between winning and losing teams. Forward players are a very homogeneous group and mainly composed by all-round players with multiple roles.