Player transfers form the squad of the football clubs and play an essential role in the success of the teams. A carefully selected player squad is a prerequisite for successful performance. Consequently, the main topic of the football world during summers is the transfer rumors. The aim of our research is to predict future player transfers using graph theory. In this paper, first, we examine the networks formed in the football world and whether if these networks have small-world property. To do this, we set up an acquaintance graph among professional footballers based on if they have ever been teammates. We make a similar graph for the managers, in which we consider two coaches connected if they have coached the same club. Moreover, we also analyze the network that has developed among the teams in the past 14 years, in which links illustrate player transfers. Using the graphs` metrics and the information about these transfers, we make a data mining model for predicting the future transfer of players. The model can be used to predict who will transfer into a selected league. Different leagues show different features as the most important ones when it comes to buying a player, but in every case that we studied, the features extracted from the graphs are among the most essential ones. These features improved the performance of the player transfer prediction model, giving sensible possibilities about the transfers that will happen. Network science has become widespread in recent years, allowing us to explore more and more networks. By examining complex networks, we can obtain information that would not otherwise be possible and that can have a massive effect on predictions. We show that by using this information we can create meaningful features that can improve the performance of the predictive models.
© Copyright 2022 Machine Learning and Data Mining for Sports Analytics. MLSA 2021. Communications in Computer and Information Science. Veröffentlicht von Springer. Alle Rechte vorbehalten.