Wong, Zhen Ping (2021) Predicting soccer result using dixon coles model and its applications. Master dissertation/thesis, UTAR.
Abstract
The Dixon Coles model uses attack and defend parameters from the team’s perspective to predict a soccer result. This approach can be improved by adding the player’s rating parameter. This study aims to improve the Dixon Coles model by implementing attack and defend parameters at the player’s level. In this context, the player’s attack and defending parameter are calculated from the player’s rating data, coming from whoscored.com (n.d.). To test the hypothesis that the player’s parameter will improve the model, we collect the player’s rating data and calculate the player’s attack and defending parameter. The parameters were then added to the Dixon Coles model. We predicted season-long matches using the Dixon Coles model (before and after adding player’s parameter), and the result from both models was compared. Overall, the results showed an improvement: the player’s parameter did improve the original Dixon Coles model prediction. The underlying statistical distribution for the Dixon Coles model is Poisson distribution, its application is extensive, as long as the expecting event is statistically independent and the rate of happening is constant, such as packet loss per hour in networking field, number of customer arrival.
Actions (login required)
![[img]](http://eprints.utar.edu.my/style/images/fileicons/application_pdf.png)
