Neural Networks Predictive Modeling for Football Betting

Abstract

The aim of this paper is to explore neural network-based modelling strategies for football betting. A neural network model and a challenger model based on a traditional econometric approach introduced by Dixon were estimated on data from five national leagues results including France, Spain, Italy, Germany, England) Our results show that the Neural network-based model has better predictive accuracy compared with the traditional econometric models. Betting strategies were implemented using prediction outputs generated with both econometric and neural networks models. The latter provides with a better return over investment. Nevertheless, both approaches lead to losses in the long run.

Keywords: Sport Betting Football, Machine Learning, Neural Networks, Statistical Models, Forecast

JEL Classification: G12