Exploration of various predictive models for the National Basketball Association

This research paper explores the application of various predictive models for the National Basketball Association (NBA), with the aim of predicting game outcomes and margins of victory for the 2015- 16 season. The research focuses on three, distinct models, all created using the statistical analysis software Rthe mean score difference model, the weighted mean wins model, and the team strength model. Each model employs different methodologies and data analysis techniques to predict game outcomes and team strengths. Our study found that the team strength model and mean score difference model performed best due to their use of point margins as their primary variable, while the weighted mean wins model was less accurate due to its use of wins and losses. None of the models, however, were able to match the predictions from sports gambling organizations, called the bookies, showing that much more work is needed, not only to surpass the bookie models but to inch closer from prediction to actuality.