Predictive Modeling of Association Football Scores Using Bivariate Poisson

Formative outcomes and terminal scores of association football serves as the basis for most placed bets in Ghana. These formative outcomes and terminal scores are largely dependent on the quality of teams’ performance as against that of the opponent(s). The paper proposes a predictive model that assesses the quality of teams’ performance via teams’ scoring intensity by encapsulating Foul for teams (FoulF), Fouls against the team (FoulA), Red cards against the team (Red CA), Corners profile for the team (CornP), Yellow cards against the team (YelCA) and Shots on targets (ShotT) of home and away teams, and further used these scores’ intensities between any two teams to calculate the probabilities of win, draw or lose between teams using bivariate Poisson distribution. The paper concludes that, the higher the scoring potency of teams, the higher their probability of winning and vice-versa. The paper also avers that, on average there appears not to be any “wild” difference in playing at home or away contrary to conclusions drawn by previous researchers in playing at home without skills dominance of the home team. Both home and away scores are influenced by corner profiles and shots on targets of teams.

Keywords: Poisson Regression, Bivariate Poisson, Scores Modeling, English Premier League (EPL), Home advantage, Probability matrix