Visualizing a team's goal chances in soccer from attacking events: A Bayesian inference approach

Abstract

We consider the task of determining the number of chances a soccer team creates, along with the composite nature of each chance-the players involved and the locations on the pitch of the assist and the chance. We infer this information using data consisting solely of attacking events, which the authors believe to be the first approach of its kind. We propose an interpretable Bayesian inference approach and implement a Poisson model to capture chance occurrences, from which we infer team abilities. We then use a Gaussian mixture model to capture the areas on the pitch a player makes an assist/takes a chance. This approach allows the visualization of differences between players in the way they approach attacking play (making assists/taking chances). We apply the resulting scheme to the 2016/2017 English Premier League, capturing team abilities to create chances, before highlighting key areas where players have most impact.