Predicting sports results using latent features: A case study

Predicting sports results is normally a challenging task, even more in case of a sport that shows a highly stochastic nature. In football, for example, numerous features are tracked and combined with expert knowledge, yielding various predicting algorithms. Our work however, is based on a case where there is no expert knowledge available and the only data comes from previous match results. We built a goal score prediction model that uses latent features obtained from matrix factorization process. We also added a Naive Bayes Classifier to be able to predict outcome of the match. The algorithm has been tested on results of the FIFA World Cup 2014. We also built a match result predictor based on the betting quotas. As these are derived from a complex algorithms that encompass also the expert knowledge, our algorithm can be used to estimate accuracy of an expert knowledge-based system. This case study shows that there is no significant difference between the two algorithms that we tested and that the latent features may provide a valid substitute for real features, when the later ones are not available.