More than 1 million people in the Netherlands play football under the Royal Dutch Football Association (KNVB), while only 26 men and 23 women make it to the national team. Scouting the right talent in the early stages of their career is an art itself, and professionals are trained to find the few players that grow out to be one of the successful ones. These scouts use data such as speed, goals, completed passes among others, but also review the role a player has in a team outside of these metrics. Is talent quantifiable in every aspect? Or do we need the human skills of a football scout to find the needle in the haystack?
Nnamdi Emefo, CEO of Afriskaut, spoke to the Global Business Report about using Artificial Intelligence to scout for young African sports talent. He talks about the model he built over the course of five years, where he now is able to extract all statistics from a video of a football match within 24 hours, for all 22 players on the field. He created a platform where all this data is available for the whole world to see (against payment). You might think, what is new about that and why can big professional football clubs not generate this theirselves? Nnamdi’s platform is so valuable that even the big clubs like Bayern use his platform. And it is not just football, other sports like tennis or basketball can also be converted into statistics by Nnamdi’s model.
And it appears to be successful in scouting talents. Right now African basketbal players are scouted using this model. These players were successfully matched with clubs in higher divisions. His model is not only able to generate a personal file of performance statistics, but is also able to see if there would be a match with a possible transfer team, taking all data of current players in that team into account.
Another success story besides in basketball can be found in football again, young African boys are also scouted for the youth teams of professional European clubs. His data base is used to find the optimal match that is a fit for both the player, as the team as the club. This way he gives young African boys the opportunity to be taken serious by European clubs, by providing them to prove their worth with data. He believes his model had the ability to mitigate racism in the scouting process, and thrives to be the key player in the scouting proces in the sports industry in Africa in the next couple of years.
See interview here: https://www.youtube.com/watch?v=74pzE6AofG4
I think it is a very interesting topic to see if AI can receive actual scouts. It’s a topic that I get questions right away. For example, I am very curious about how the software also looks, for example, at the level of the opponent someone is playing against. This obviously has a lot of influence on your own results.
But ultimately, I think one of the biggest problems will be collecting all the data from all the clubs in the world. It seems to me that you need to put a huge number of good cameras and equipment at football pitches to get good results.
In addition, I am also curious if eventually a football club can really take the lead with this and thus improve itself compared to its competitors. Maybe there will be some kind of monopoly of a football club that will always win.
Again, very interesting post and looking forward to seeing what the future of AI and top sports will look like!
Hi Jip, it is interesting that you mention the software’s ability to take the opponent’s level into account. I could not find how they account for that aspect, but really curious now you mentioned it!
Interesting topic, never even thought about it. I think that the most valuable thing about the AI technology is that it bases their decisions purely on data over the long-term, which makes it objective. With ‘human’ scouts there are often a lot of biases which leads to missing opportunities or subjective decisions. Furthermore human scouts watch mostly two games of a player. Therefore a scout bases his decision on a one or two time performance of a player, which can be a wrong decision(if player was feeling sick one game or just did not play well). Although I do not think decisions should be based purely on data, the AI technology is definitely an excellent complementary tool for scouts.
Hello, first of all, thank you for this very interesting article, i really enjoyed it! As a football fan myself, I can see how convenient it could be to have such a classification software. I would be interested to see the limitation since one’s performance does not mean that it is a successful players. For example, Kante is a french player that plays for the national team, but he barely do any assist, nor goals. However he is seen as one of the best in the team because of effort. Would the software detect metrics such as skills and if the player is critical in the team even though there are no goals. However, such a software is really convenient since analysing all the young players is impossible to do. But a generalized software that classifies the players could be a way to detect the ones that are in remote clubs for example.