Data analysis in professional sports Artificial Intelligence in the game

SAS and SCI Sports are developing AI algorithms to capture and analyze football match data.
SAS and SCI Sports are developing AI algorithms to capture and analyze football match data.

By international comparison, German football is currently lagging behind the top European leagues and there is hardly any improvement in sight. Artificial intelligence could soon change this.

»AI at the Arena« - under the slogan the company SAS invited to the Allianz Arena in Munich. Artificial intelligence (AI) is now used in football, insurance and finance. But how do AI and football fit together?

The basic idea is easy: simple data analyses and algorithms were already used in the first »Bundesliga manager« game. The championship could only be won through a clever transfer policy and analysis of the players and the game system. The analyses are primarily based on game data collected during a game. In reality, it also works like this: data such as kilometres run, passes played or goals scored are recorded for each player. This makes it possible to evaluate how good the player was in the respective game.

But the data is not always informative. If, for example, Kevin de Bruyne plays a wonderful pass to the striker, who then awards the goal, de Bruyne does not receive a scorer point. The pass was therefore worthless. In return, Lionel Messi uses a simple pass to score a dream goal. The passing player receives a scorer point. This makes it clear that the data are not as revealing as expected.

Data analysis with algorithms

The companies SAS and SCI Sports developed an algorithm for the analysis of game data. By analyzing hundreds of games and evaluating their performance, each player is assigned a so-called »SCI skill level«. The level is equal to the current performance value of the player. At the same time, the player's potential is opposed as an »SCI skill value«. It indicates on which level the player can evolve. Especially the SCI skill value is interesting for the transfer policy of the clubs.

SCI Sports' AI algorithm has already been used for the 1. FC Nürnberg or the VFL Wolfsburg. The scouts from Wolfsburg, for example, found the player Wout Weghorst through the analysis. The transfer was worth it - he currently scored 12 goals in the Bundesliga. The example of Memphis Depay shows that it also works the other way round: He wanted to play for a new club and found it through the analysis of SCI Sports. The algorithm said that his way of playing fits best to the team of Olympique Lyon.

Support for transfer policy

The only problem for the algorithm is the human factor. A player can injure himself at any time or be deleted from the squad by escapades. Such uncertainties are partly taken into account, but it doesn't really work out yet.

In the future the algorithm should contribute to a better transfer policy of the teams. It remains to be seen whether and to what extent this will work. Perhaps this will even make German football more competitive again at the European qualifiers.