Nowadays, the relevant technology based on Mathematics, and more specifically, in Probabilities and Statistics, along with the rapid advancements in computer power has deepened our understanding on a football game’s result, giving us the capability of estimating future game results based on past performances. The ability to estimate the number of goals scored by each team in a football match has revolutionized the perspective of a game’s result for both the betting market professionals and fans alike.
The Poisson distribution has been widely accepted and used in a number of studies to model the number of goals a team is likely to score in a football match, and, as such, the match result can be estimated using a double Poisson model – one for each participating team. Furthermore, the expected number of goals scored by a team, which is a prerequisite in the Poisson distribution formula, can be estimated with a Poisson regression model, using a number of independent variables as predictors, referring mostly to past performances of the corresponding teams – attacking and defensive ability – and other factors, such as home advantage for the Home team, and used as input into the Poisson distribution. Using the above, and other cutting-edge statistics, we have managed to achieve one primary goal : understand who is going to win and why.
Following the above principles, the aim of the current thesis is the development of an algorithm, which, by using Poisson distributions along with football teams’ historical performance will be able to predict future football game results. The aforementioned algorithm is developed based on the Premier League – England’s top-flight football championship – results of the 2022-2023 season. The Premier League was selected as being one of the most high-profile football championships in the world, and as the most-watched sports league in the world, broadcasting in more that 200 countries, with a potential TV audience of more than 4 billion people.
Konstantinos_Loukas - Thesis