Thesis (M.Sc.) - Cairo University - Faculty of Computers and Artificial Intelligence - Department of Operations Research and Decision Support

Prediction of players{u2019} market value for the most popular sport on the planet through a data driven approach has no more became an option, it has always been left up to the subjectivity of the decision maker, however in the last decade football transformed from a sport to an industry where a systematic framework became crucial to address simple questions such as: what really determines the value of a footballer? How much should a club pay when purchasing a certain player in a certain position on pitch in a certain season? And based on which criteria such that any subjectivity is eliminated?. With the right answers to these questions, in the right time, a club as a revenue maker is assured to gain an unparalleled edge to its rivals. The Egyptian Premier League (EPL) is one of the top 5 leagues across Africa with total net worth of 159, 28 million Euros1, yet it still lacks a scientific approach to assign a proper estimate for a player{u2019}s market value in a given season. In an attempt to bridge this gap, this research presents an end-to-end fully data driven scientific framework covering all given functions and positions of a footballer on pitch, be it a defender, midfielder, attacker or even a goalkeeper. Players are modeled to be classified to one of three categories: High, Medium or Low price players according to their history of performance depicted on pitch for the last three seasons. The results were tested and validated on real data for the Egyptian Premier League players in seasons 2015-2016, 2016-2017 and 2017-2018 and as will be shown model evaluation showed very promising results for prediction

Issued also as CD