Using Networks to Analyze Basketball
https://www.wired.com/2012/12/basketball-network-analysis/
Basketball is my favorite sport. The fluidity of the game and the constant switching of momentum makes the game unpredictable. The spectator never knows exactly what is going to happen. Recent progresses in analytics and modeling, however, make predicting the outcome of a game more and more feasible. This feat is made possible by equating the game of basketball to networks, where each player is a node and the connections between the nodes are paths defined by the ball being passed between players. The strength of any given connection is defined by how often a pass occurs and its success rate. A complex web is created and serves to explore the complexities and connections between a team’s players.
Jennifer Fewell and Dieter Armbruster parsed data from the 2010 NBA playoffs and applied their network model in an attempt to better understand the game. They found that the entropy, defined by the amount of passes and randomness in the game, was a successful metric for determining who would win the game. The higher a team’s entropy the greater the odds they win the game. The potential benefits for a system like this are endless. One could use these metrics in selecting a team and recruiting talent to try and optimize the entropy of the team. This would directly benefit the teams and also the quality of basketball as more competitive teams would become the norm. Another potential application could be gambling. If an individual developed their own model and analyzed the game real-time, they could predict the outcome of the game with some certainty. The gambling would therefore be more calculated and the likelihood of accurately predicting the outcome of the game would be increased. These are just a few applications of networks in the context of the NBA. The evolution of the models and there applications with undoubtedly continue to flourish and contribute to our understanding of the game.