From core to peripheral: A network analysis of lineup types in NBA playoff teams

Abstract

This study aims to identify the types of lineups based on their topological structure within a lineup network and to explore the relationship between lineup types and team standings during 10 NBA playoff seasons from 2012-2013 season to 2021-2022 season. A total of 15,699 lineups from 1,655 playoff games were collected to construct lineup networks. Three roles of the lineup, called core, connector and peripheral lineups, were found through community detection and unsupervised clustering of within-community degree, participation coefficient, and playing time. The percentage presence of connector lineups showed a positive correlation with the number of playoff wins (r = 0.45, p < 0.001), while peripheral lineups demonstrated a negative correlation (r = −0.33, p < 0.001). Additionally, the study found that connector lineups were more frequently reused than peripheral lineups (H = −14.90, p < 0.001) and that stronger teams exhibited lower conserved rates of all kinds of lineups. The collective performance was found to be more dependent on connector lineups (H = 926.42, p < 0.001) than peripheral lineups (H = 3342.63, p < 0.001). This study is the first to provide insights into the global lineup roles within season-scale lineup structures, offering generalizable suggestions for optimizing rotations. These suggestions advocate for the inclusion of more connector lineups and versatile players, and a reduction in the reuse rate of lineups, especially those classified as peripheral.