An artificial intelligence (AI) predicts the best strategies of footballers |

Scientists at the Alan Turing Institute in the UK have developed an AI algorithm to predict which team has the best chance of winning the World Cup in Qatar. It mainly takes into account the results of previous championships.

However, important factors such as the performance of individual players are left out. It would therefore be interesting if it were complemented by another algorithm, developed in 2020 by data scientist Carter Bouley. Analyze the different types of passes players can make, in order to calculate the best strategies.

It should be noted that this algorithm was not made for the World Cup in Qatar, nor for any specific championship. It’s simply a way of using statistics and computer science to develop the best game strategies for footballers.

It relies heavily on passing, so it could be very useful for teams like the Spain national team, whose passing game is often one of the keys to their success. The ideal is to optimize them, so that they involve the minimum expenditure of energy and that they can also be linked with precision until they end up on goal. There is no magic formula, but at least this algorithm can help design the best possible strategies.

The perfect pass algorithm, even beyond Qatar

Of course, not all players are the same. They are more or less qualified and more or less trained. However, optimizing their passing strategies can help them all.

That’s the goal of this algorithm, which was trained using data from 358,753 passes made in 380 games involving 20 teams. Several factors were taken into consideration. First of all, it must be established whether the players are in their own half of the pitch or in that of the opposing team. For another, minute-by-minute results and full match results. Furthermore, the passes were drawn graphically, with the ends of the pitch as X and Y axes. Finally, the type of pass was taken into account: normal, header, cross, corner kick, lineout, kick goal kick or free kick.

With all this data, AI was put in place to look for patterns that linked a specific type of player pass with better results. They discovered data that a large percentage of passes are missed at a very short distance, less than 5m. Also, between 15 and 30 m, “ there are much more successful passes than missed passes, and after 30 m the proportion of successful passes drops sharply, while missed passes begin to level off“.

Another key factor turned out to be where the pass is made on the pitch. For example, the closer they get to the opponent’s goal, the more passes they miss. Logically, this is a very important area, so knowing which strategies work best in this position is important.

Particular attention for football players

In this algorithm, the individual role of the players is taken into account. As Bouley himself explained at the time, ” if the model predicts that a pass with a probability of 0.8 will occur and it is made, 0.2 is added to the player’s pass score“. On the other hand, ” if the pass was not made, minus 0.8 for the passing rating of the players”. The average is then calculated on the number of passes made by the player, in order to define an average pass risk score. “This score allows players to be compared by risk taken and passed in the pass“.

Because, logically, it’s not just about knowing which steps are best. You also need good players who can run them. It also means they need to be able to take risks, but without being too bold. There is virtue in the middle ground. This also applies to winning a football match. It doesn’t matter if you are at the World Cup in Qatar or in a local league.

Post up Hide the summary

Leave a Reply

Your email address will not be published. Required fields are marked *