news agency
Artificial intelligence rules the club.  Team “for free” beats richer teams many times

Artificial intelligence rules the club. Team “for free” beats richer teams many times

In Toulouse, the computer should get a lot of applause for recent successes. Club coach Philippe Montanier was nominated for the position by a special programme. Another algorithm was also behind the employment of the player who became the top scorer in Ligue 2. As the club’s authorities assure, the guidelines provided by the system made Toulouse gain an advantage in more than one match. And while most teams use data and analytics in modern football, a few do so on a grand scale. “Moneyball” exists and it’s right next to us.

“Moneyball” in the French edition

The association with “Moneyball” – a book and a film based on facts – is most appropriate. In that story, Billy Beane, the new manager of the Oakland Athletics baseball team, abandoned traditional methods of running the team and relied on computer analysis. He got a lot out of it. His club recorded historical successes, beat stronger ones, and Beane’s American dream aroused media interest and recognition so great that others wanted to employ him. One of the creators of Toulouse’s success, director Julien Demeaux, was delighted with the analysis and approach to data in America. At that time, he was working in high school as a football and… baseball coach.

“I started delving into how baseball clubs use data to make critical decisions and I was figuring out how that could be translated into football,” he told The Athletic. After a few years, Demeaux returned to France and worked at one of the companies providing data to football, basketball and rugby clubs. He has always liked sports, and he had a talent for exact sciences. He used to work in aeronautics, worked for Airbus and developed a flight simulator for a pilot training center in the UK.

During the pandemic, the American club Seattle Sounders organized a competition for data analysts. The topic of the presentation was free. Bored at home in Toulouse, Demeaux did a pre-match analysis of Barcelona. After a few days, it turned out that he won. It attracted the attention of RedBird Capital Partners. in various initiatives, including sports. Through investing in Fenway Sports Group, he has indirect stakes in Liverpool, the Pittsburgh Penguins (NHL) and the Boston Red Sox (MLB). Recently, it also bought shares in Milan, and during the aforementioned competition for analysts, it was just taking over FC Toulouse. Its representatives decided to take advantage of the brilliance of the man who is on site. After talks, they hired the Frenchman as football data manager.

“Clubs use data to support what scouts believe. It’s a mistake”

Since RedBird, among several hundred companies in which it has invested, also has one dealing with sports data analysis, the path to cooperation between Zelus Analytics and FC Toulouse was natural. The club now has seven people working with Demeaux to create the algorithms needed for various tasks, prepare specific reports or just transfer data. “As a club, we are very data oriented,” said Demeaux. They decided to be the team’s coach When scouting players, it all starts with them.

– Scouts are here to see what the details say. They report already armed with numerical knowledge. Many other clubs use data to back up what scouts believe, and that’s a quick way to get it wrong, says Brendan Macfarlane, head of Toulouse’s player acquisition team.

Macfarlane is a Briton studying French in Scotland who was once sent to Niort for an apprenticeship. In his free time, he watched the matches of the local team there. He took notes on players. Not only about those from the local team, but also about other Ligue 2 players. After some time, he handed over a large database to his beloved Celtic. No reaction. However, after the publication of some observations on the internet blog, clubs from the Netherlands and England became interested in him. He was brought to Brentford. Then he appeared on the radars of Toulouse. He has been working there since mid-2021. It is he and his colleagues who are responsible for giving the data a human dimension or adding what the numbers will not show in the event of serious interest in a footballer. For “receipts” describe what they look like, for “fouls” add a comment about their nature to the report. Check what’s going on with the player off the pitch. Talk to his parents.

“We recently found a rider who looked amazing in numbers, but when I did an environmental interview, I discovered DUI and drugs. We let it go, says MacFarlane.

Simulation of results after league change. “Colleagues called and could not understand how they missed such a player”

How does Toulouse work? Its programs monitor 70 football leagues. Of course, they contain data on players from the strongest leagues, but for algorithms catching talents, information that comes from, for example, the fourth English or third German league is more important. Proprietary algorithms reward players up to the age of 25 (so far there have been only two exceptions), and even having data on style, speed of play and other parameters, they forecast their performance after changing the league. It sounds strange, but in practice it works.

Toulouse’s 2020/21 (14 goals) and 2021/22 (20) top scorer Rhys Healy joined the club after playing in the 3rd and 4th tier in England. It was imported for 500,000. euro. With the central midfielder Branco van den Boomen (EUR 350,000), the algorithm said that after moving from the Dutch second division to the French Ligue 2, his statistics should remain at a similar level. In the Netherlands, his best tally is 11 goals and 15 assists. At Toulouse, he had 12 goals and 21 assists in his best season. It turned out better than the processors calculated. 21 assists was a new record in the second tier of French competition. “Colleagues from the Netherlands called later and couldn’t understand how they had missed such a player,” MacFarlane later said.

The algorithm monitors the players after the match and also presents their long-term analysis. After it turned out that the current main goalkeeper Maxime Dupe would perform much worse in Ligue 2 than in Ligue 2, a rival from the Norwegian Valerenga was brought in for him. Dupe has 16 clean sheets in Ligue 2 now he can dream of, but he is still number one. Thanks to the system, he knows what he needs to work on.

The algorithm also analyzes Toulouse’s game and highlights elements that can still be improved. In addition, before each match, he points out the weaknesses of rivals. In the corridors of Toulouse, a story still circulates how, before the match against Paris FC, Demeaux told the players that the rivals’ goalkeeper had the greatest problems with reacting to shots on the ground from his right side. The game was evenly matched until Stijn Spierings beat the Parisian goalkeeper just like that. Apparently, this was not the only time when the advice from the computer came in handy.

“I see so much more than anyone else.” Six hundred players per window

The proprietary algorithm is also used to shine in conversations with potential new players. During these conversations, the players get a lot of information about each other. In a short presentation on charts and on field maps, they see advanced statistics with a description of in which situations they are most effective.

– Football is unpredictable. With our method, we want to protect ourselves from irrationality. Of course, mishaps will happen, but a rational approach and systematization of knowledge creates a comfortable safety cushion – describes the club’s president Damien Comolli (former chairman of Liverpool). This is also known to football managers, who rarely call Toulouse with their offers. Their stories about the great left foot and compilations of the best moves of the players are collided here with the novels of analysis programs. Demaux admitted that during the summer transfer window in 2021 he analyzed 600 players quite efficiently. “Thanks to these programs, I see a lot more than anyone else and I don’t need a lot of time to do it,” he commented. Demaux in his office is a modern version of the world-wide network of Scouts.

Demaux already provides a limited list of players to the recruiting department, and in practice only those names are dealt with in the field. Looking at the last three years, it can be seen that Toulouse most often reaches for players from the Netherlands, Belgium and Scandinavia, but the list of acquired players also includes players from Japan, Australia and Slovakia. There is also a price limit in the algorithm. Comolli in one of the interviews mentioned that a player from Scandinavia costs on average five times less than a player from France of similar quality and skills. The policy of methodical management seems to be beneficial for the club. Last season, Toulouse were promoted to Ligue 1, scoring the most goals in the history of the 20-team league (82). She earned about EUR 28 million on transfers over three years. Currently, in the top flight, it occupies a place in the middle of the table.

Team from the computer also in Belgium. She beat much richer rivals

Toulouse is not the only team functioning thanks to algorithms. The term “computer team” can also be reserved for the Belgian Royale Union Saint-Gilloise. Since 2019, the owner of the club is Tony Bloom, at the same time the president of Brighton & Hove Albion, also a professional poker player. It was his company that once developed a computer program that analyzed data on any sporting event and indicated points that could give an advantage over the bookmaker. After investing in football, he created algorithms that, fed with data, predicted the best and almost free team composition.

The computer vision was implemented. and what is important, she quickly and in great style won promotion to the top league. Then, quite unexpectedly, she fought for the Belgian championship, dominating her rivals in the first part of the season. After promotion, the Union roster was refilled at almost no cost. In total, according to data published on Transfermarkt, Union spent just over two million euros on transfers over four seasons. The media praised the ability to find the right tactics, selection and quick fitting of new players. Rivals could have sour faces. Genk, whose team cost €134 million, and Anderlecht, worth €90 million, found themselves behind Union, a €20 million club.

What other teams in Europe rely the most on algorithms and processor calculations and partly trust them in football? In an interview with Europe-cities.com, Demeaux admitted that only a small number of clubs rely so much on data for now. He listed England’s Liverpool and Brentford FC, as well as Danish Football Club Midtjylland, which has the same ownership as Brentford. None of these teams in recent years could complain about the results. Liverpool have triumphed in England and the Champions League, most recently being Champions League finalists. Brentford was promoted to the Premier League two years ago, and the Danes have been at least runners-up in the country since the time of the new owner, also enjoying the championship titles.

A year and a half ago, it was also loud in Poland that Ireneusz Mamrot was helped by an external company that commissioned a computer to analyze the data. mumbling in . But will soon what will happen in football be decided mainly by the computer?

Source: Sport

You may also like

Hot News

TRENDING NEWS

Subscribe

follow us

Immediate Access Pro