The aim of this exploratory study is (1) to determine the relationship between substitution network (Sub-N) parameters and teams` standings and (2) to find out the key individual performance indicators that differentiated substitution groups of players, and explore the association between players` percentages and team`s standing within the obtained substitution groups. A total of 574,214 substitution events during the last 10 NBA seasons were analysed to construct Sub-N for each team observation. Three different player groups were obtained after clustering their playing time, clustering coefficient and vulnerability. Team`s clustering coefficient, standard deviation of vulnerability and out-degree centrality of starters exhibited moderate to strong correlations with team`s standing during playoffs (r = 0.54-0.76). The regression models showed that defensive win share (beta = 0.54-0.67), turnovers (-0.15 to -0.25) and assists (0.12-0.26) were predictive for all players` net ratings, and the role players who scored more points presented higher net ratings (0.34). Finally, players from top-playoff teams exhibited lower absolute value of vulnerabilities (r = 0.80). The findings demonstrate the feasibility of Sub-N for exploring the association between rotation and competitive performance, and provide quantitative reference for coaching staff to optimize substitution structures and rosters.
© Copyright 2022 Journal of Sports Sciences. Taylor & Francis. All rights reserved.