A moneyball approach to Fantasy Premier League

Abstract

Human emotions are often present in football as biases towards specific players or teams. In the context of Fantasy Premier League (FPL), this could influence the choices made by players of the game. The aim of this research was to eliminate any form of favouritism or bias by utilising an AI approach to determine the most suitable players to select. This was achieved by using historical data to train a number of machine learning models which are capable of predicting the number of points which each player is to obtain in a particular FPL gameweek. Apart from eliminating bias, the models can be used to help players perform better at the game by taking informed decisions and selecting players who are expected to perform well based on data and statistics.