Total Football: A Machine Learning Approach to Utilize Holistic Football Data for Match Outcome Prediction

As the world’s biggest game, football enjoys the interest of billions across the globe. The prediction of football matches fascinates everyone, from the fans to the managers themselves. Football is an inherently chaotic game, and so, the prediction of matches is not a simple task. In this paper, we have tested out various machine learning classification models, trained on historical match data of both sides. To improve performance, we have considered the sides’ midfield and defensive parameters, which are not used often, in addition to the more commonly used attacking parameters, along with data on the player attributes of both sides to predict the outcomes of matches. We have seen that this new dataset improves the performance of the popular basic models considered. Additionally, using these basic models, we have built a modified heterogeneous ensemble classifier, with the aim of maximally utilizing the strengths of its constituent models.