Predictive and Prescriptive Models for playtime allocation of football players and its impact on their development

This thesis explores the development of predictive and prescriptive models aimed at optimising the allocation of playtime for professional football players, with a focus on their performance and development. Leveraging machine learning models and optimisation technique, the study builds a framework that forecasts both the number of minutes players will play and their corresponding SciSkill, a key performance indicator (KPI), development over the following year. By incorporating contextual factors such as team attributes and individual player features, the research aims to assist football technical directors and coaches in making data-driven decisions regarding player acquisitions, sales, playtime distribution with its corresponding KPI values development. The project integrates historical player performance data to create a comprehensive tool for maximising squad development. The results highlight the importance of optimal playtime allocation to maximise team development, offering insights into the impact with the addition of new players and/or the remotion of some others.