Using poisson model for goal prediction in European football

Abstract

Predicting the features of behaviour of big data and multivariable systems has been a research subject in various fields of science. When it comes to football, as it is a field of sports followed by the whole world, the number of studies carried out aiming at predicting the results of football games has been increasing in the field of football science. Although the result of a football match depends on various variables, it is mainly determined over the offensive and defensive strengths of the teams. Different variables have so far been determined in the literature to figure out these strengths of the teams. In this study, it was aimed to predict correctly how many goals a team could score or concede in the last 5 weeks based on the average number of goals they scored and conceded since the beginning of the season in 6 European leagues. For this reason, a Poisson distribution model was established based on these offensive and defensive strengths. A total of 4264 matches and 5938 goals were analysed in the study and the established model yielded affirmative results at the level of 50% in the leagues analysed.