We wish to build a model, which could predict the outcome of basketball games. The goal was to achieve an sufficient enough accuracy to make a profit in sports betting. One learning example is a game in the NBA regular season. Every example has multiple features, which describe the opposing teams. We tried many methods, which return the probability of the home team winning and the probability of the away team winning. These probabilities are used for risk analysis. We used the best model in hypothetical sports betting and measured potential net profit. The results are not perfect, so we mentioned possible improvements. I think that a lot of the ideas could also be of use in other fields, which deal with risk and predicting the future.
|
