A Heterogeneous Ensemble of Classifiers for Sports Betting: Based on the English Premier League

This article presents sports betting as a classification challenge within the realm of machine learning, emphasizing responsible gambling practices. It outlines the challenges and prospects of accurately predicting the outcomes of sporting events, involving complex dataset utilization and sophisticated algorithm application to predict winners, losers, or draws in order to profit from sports betting. The article further discusses the hurdles of maintaining high model classification quality amid data relating to real football matches. A data analysis solution was proposed, followed by the construction of a heterogeneous ensemble of classifiers designed to make sports betting profitable, validated by application on meticulously prepared English Premier League data. Experiments conducted under near-real conditions confirm the research hypothesis, demonstrating the efficacy of heterogeneous ensemble methods in improving betting accuracy.