AI in football: Why machines are correct with unpopular predictions

Mister Wendthow will the championship end this year according to KI?
With a probability of 97.3 percent in favor of Bayern. St. Pauli, Holstein Kiel and Bochum will be relegated. We made these predictions at the beginning of the season based on squad strengths.

Where do you get this information from?
We founded an AI data science company called “Plaier” and created a database of 300,000 players, each with 140 attributes. We put the data into perspective and can therefore make forecasts.

And are the predictions accurate?
You have to remember that we are talking about mathematical probability. There is no such thing as 100 percent security. Therefore, it is the quantity of correct predictions that matters, not individual ones.

However, their AI correctly predicted that the first and second place teams would be tied on points at the end of the 2022/23 Bundesliga season.
That’s correct. Yet such championship predictions are a byproduct of what we do. Other things in our daily work are much more important.

Namely?
Our core business is that we predict whether a player will make a club better or not. And we are 90 percent right.

Do you have an example?
Our most popular prediction is that Harry Kane will not have a positive impact on goal difference in his first year at Bayern. At the same time, we predicted that Tottenham would be better in the three areas of points, goals scored and goal difference after Kane’s departure. Both have arrived.

An unpopular forecast.
We have some of those. We also say that Stefan Ortega is the best German goalkeeper. And that Kevin Stöger is among the 100 best players in the world.

How does the algorithm work?
Every player has an untold amount of data collected about them. This flows into a value that makes it measurable. We can then say exactly which player works in which league and in which position in relation to other players.

And the clubs benefit from this.
Exactly. We tell the clubs how good their squad has to be. Or who they should sign to achieve their sporting goals.

Why is it in the modern one? football no longer without data analysis and AI?
It works. But when one person starts, everyone else has to join in. Data has been around in football for many years. The only difficulty is interpreting them. Without AI, this is impossible today.

What are the limits of technology?
The players are physically available to us almost 24 hours a day, but not psychologically. We don’t know what’s going on at home or what they experienced in their childhood. Therefore, an AI will not be able to psychologically prepare a team for a game in the near future. This will be a mystery until the data improves in five to ten years.

And then AI could replace trainers?
This is almost a philosophical question. I think people want to deal with people. But a coach’s job will change a lot.

Have you ever considered using your product for sports betting?
If you were to bet with us, you could be guaranteed to win a small percentage per bet. We have proven this in simulations. However, we decided against having anything to do with the topic. We cannot advise on the club side and bet against the club at the same time. This is impossible to reconcile with our core business.

Are you a football fan yourself?
Yes, I am a big football fan.

Isn’t it unromantic for you to calculate the game like that?
Yes and no. Romance often arises from tension, which AI simply shifts. It is then exciting to see whether the forecast is correct or not. Football still remains an unpredictable game.

We are just at the beginning when it comes to AI. What else can we expect from technology in football in the future?
AI will trigger massive changes, especially in the tactical area. There will also be a stronger relationship between a player’s quality and his market value. It doesn’t matter whether he comes from Chelsea or Ingolstadt. Currently, according to data, the best players are not always the most expensive. If that changes, more money will be invested in young players with good prognosis. Their transfer fees could then shoot up. And older players become more valuable and important to teams.

How should the implementation work? At the moment we can’t even manage it with the VAR.
That’s already happening. You can already see that in the tables. We don’t publish customers, but in the second Bundesliga you can tell that Schalke, Hertha and HSV don’t work with us. You can see who is in the top five. That’s quite noticeable.

What would you say to people who are skeptical about the use of AI in football?
I can understand that. But German football has to decide. If you want to compete in world football, you have to use everything that helps you get better. And it won’t work without AI. I originally come from motorsport. We really tried everything that could potentially benefit us. I sometimes miss this unconditional will to improve in football. Now diamonds can still be found everywhere in the sport if you calmly introduce our system. If you have to make a decision later under pressure, it will be much more difficult – and you may lose touch.

The football romantics won’t like that.
However, the romance of football in its current form always carries with it a bit of injustice. We show clubs and players undreamed of opportunities through AI. I think we are making football fairer, more democratic and more transparent.