Abstract: To revolutionize the concept of basketball “positions”, we use a multivariate cluster analysis to group NBA players based on several statistics. Once these clusters have been formed, we analyze how each cluster affects winning in the regular season. Along with the main effect of each cluster, we analyze the 2 and 3-way interactions of the clusters. The ultimate goal is to determine which types of players are most relevant to winning, and also what combinations of players affect winning. This will give us a way to scientifically measure team chemistry.