Dieter Armbruster, School of Mathematics and Statistics, Arizona State University
Netball - The Network Structure of NBA Basketball
We analyze the network properties of NBA basketball games. Defining players as nodes and ball movements as links, we can characterize a ball game as samples of random networks. We have studied the playoff games of the 2009/2010 NBA series. Using network theoretical constructs, we determine average properties for these networks and characterize the game, specific teams and specific players by their network properties. We show that there is a typical network structure for an NBA team. Most teams present networks that are close to this typical structure with the Los Angeles Lakers (low degree centrality) and the Milwaukee Bucks (high degree centrality) marking the extremes. We identify two different playing strategies corresponding to network flows with high entropy and to flows in networks that increase the shooting efficiency of the team, respectively. A particular notion of flow centrality allows us to determine the value of a player for the team, in contrast to the usual individualized statistics. The possibility to use such network analyses to study functional small team networks is discussed.