Slamball_Analysis_and_Predictions

Slamball is an alternative sport with a relatively small amount of statistical data available. However, game outcomes in the sport can be predicted with reasonable consistency when a simple form of gradient boosting is used to forecast the impact of individual player performances, even when they are estimated in advance. In essence, player success is voted on by an assortment of stochastic gradient descent models, and team success is in turn voted on by players weighted by their participation in a game. Simple regression testing can also be used to determine whether a subset of actions associated with a given style of play, including those associated with high risk levels for participants, lead to winning outcomes. This demonstrates that while physical contact is rewarded through defensive metrics such as hits and stops, violations from excessive contact or unsafe behavior are equally punishing, leading to a game that can easily be "balanced" to optimize safety under current parameters.