A comparative study of football prediction models

Rapando, Reginald Tabuche
Journal Title
Journal ISSN
Volume Title
Strathmore University
This study investigates the effect of tiers on football prediction models. The tiers in this study are grouped according to the value of the teams. Many football prediction models do not take tiers as a factor when making predictions. The study used a Poisson Regression Model which assumes that the number of goals scored by a team follow a Poisson distribution. It also employed a Bayesian approach . The parameters used in the Poisson Regression Model are strength of attack, defensive strength, home advantage and club value (the tiers). From the study it was found that teams in a higher tier were given a higher probability of scoring more goals and winning games as compared to teams in a lower tier. This shows.that the value of a team has a significant effect on the predicted outcome in a football prediction model. From this, it was concluded that the value of a team is an important aspect that needs to be considered in football prediction models since it has a significant effect on the outcome of prediction.
Submitted in partial fulfillment of the requirements for the Degree of Bachelors of Business Science Actuarial Science at Strathmore University