The football transfer market involves many big money moves. Every club want to get the next big player and fill the gaps they have in their squad.
The football transfer market involves many big money moves. Every club want to get the next big player and fill the gaps they have in their squad, all whilst trying to get players as cheaply as possible and ensuring when they sell, they get the most amount of money for their existing players. There are millions of active players worldwide, all of them recruitable and with vast amounts of data available for each player, there is a challenge for how a club develops it’s recruitment strategy.
Using SPSS and Watson Machine Learning this application utilises data from WhoScored to offer insights and suggestions to clubs for which players they should be buying/selling and which clubs they should be targeting to do so, in order to get the cheapest purchase and the most expensive sale.
The machine learning algorithms predict both the “likelihood” that a team would be open to selling a particular player, in addition to the predicted “movement” of that player, i.e. assuming the player is going to leave, where are they most likely to move to? These models were trained on historical transfer window data and involved the key step of using a team “metric” when considering a transfer between two teams. This enriches the training data considerably and enhances the accuracy of the models.