Can AI help spot the best footballer Eredivisie?
The world of football is full of experts, not to mention those sitting at home on the bench. But the sports world is also experiencing a technological revolution, with data playing an increasingly important role. Not only for the viewers, but also in the analysis of performance and to focus tactics. What else can we do with all this data? Student Daniel Krumov found out what AI models can get out of match data.
Data and goals
Daniel started with a model called Expected Threat, which gives a chance of a goal based on a player's position. After all, you win by scoring goals. A pass back, a defensive move, or even a goalkeeper's save is more difficult for an AI to link to goals. Yet these 'progressive actions' are essential for success. You don't win matches with just Ronaldo or Messi on the pitch. So how do you extract actions from data that contribute to that win, beyond goals?
Progressive actions
And that's where AI really comes in, as each goal comes from a series of progressive actions. So by applying complex models, AI can tell a lot about each player's influence on the game. This is not only great for the match reviews and the chat in the studio, but can also help clubs in the process of scouting and drafting the right players for a match. That does not make scouts superfluous, of course, nor the opinion of all the national coaches on the bench and in the studio. It is an extra tool in top sport, which can help professionals.
Will AI find the new Ronaldo or Messi for us? Daniel tells more about this in this edition of Eyes on AI.
The project is a collaboration between Fontys ICT and partner company HERE, part of the SPARC partner collective. Together with these partners, students in our Fontys ICT InnovationLab are working on applied research in and with the working world.