Are Transfer Values Objective? The Impact of Football Players' Passing Performance on Their Transfer Prospectives.

This work is devoted to modeling the transfer мфдгуы of a football player using network metrics as a reflection of the utility of players passing performance, and statistical data on the playing activities of players of the RPL and FNL teams in the 2018-2019 season from The Wyscout website. The result of the work was the developed algorithm for intelligent comparison of players by game characteristics, participation in passing networks, and validation of their transfer values. More specifically, 4 network metrics (Closeness centrality, Betweenness centrality, Transitivity and PageRank) were built on 7 types of passes, which showed the importance of each player in the organization of passing networks. Next, we created an algorithm for grouping, searching for similar players with the ability to compare them using the K-Means and Gaussian Mixture Model (GMM) clustering algorithms. The last step was to combine these data and select the most important features that were used to create a model for predicting changes in transfer values based on gradient boosting of the Catboost library