Player evaluation and team organization analysis based on network structure entropy

Basketball player evaluation and team performance focus more on technical data analysis while inadequate attention is given to the reflection of passing cooperation of the ball. The analysis of the network can make up for this defect by establishing the game as a network model. Firstly, the basketball games are constructed as ball-passing networks containing outcome nodes. The local area network that is based on the nodes defines the node entropy centrality, thus the players are evaluated from a new perspective. Taking the efficiency value of players as reference, the evaluation result of this index is compared with that of several network centrality indicators. The results show that this index has better accuracy to some degree, and also reflects player contribution in the passing process of ball. In addition, a 5-step, 40-pass network sequence is constructed for each match, and the differential passing structure entropy to measure the way of team organization is defined; and the relationship between team entropy and team performance was innovatively attached greater emphasis to on the basis of not just directing attention to the temporality of the match. The results show that the centralization or decentralization of team organization has certain impact on team performance. Too much decentralization will increase the possibility of team mistakes.