Virtual coaches and scouting apps: How the AI revolution is transforming football's future

QPR’s relegation at the end of the 2012-13 season acted as an unlikely precursor of sorts to the introduction of artificial intelligence in football.

At the start of the season, midfielder Esteban Granero arrived amid fanfare from Real Madrid but was unable to halt the club’s drop from the top flight.

He returned to Spain the following season, first on loan with Real Sociedad before joining permanently. While there, he befriended a team analyst, who found himself unsure what to make of the reams of data he had at his disposal.

Granero’s solution was perhaps atypical to that of his peers. “I took the data and started asking clever people about it,” he recalls. “I went to universities and asked professors about artificial intelligence, as I felt AI could improve data analysis, as in other industries.”

By 2016, he had founded Olocip, an early entry into football AI used to aid decision making when it comes to signing players and also around in-game decisions.

“We’ve been developing for years,” he explains. “Now the AI wave is coming — we see AI applications worldwide in every industry — we are taking advantage of what we’ve been building. For years, we were struggling to make people understand how AI can improve analysis of data. Now people come to us.”

Olocip is working with 40 clubs, including one in the Premier League and another in the Championship, neither of which he can contractually name.

AI in football is already big business and will only become bigger, and Granero is only one person and Olocip one company working in an ever-expanding sphere. As Granero puts it: “AI will bring efficiency in decision making at a club, not only in sports science but economics, the development of a club, fan engagement, ticketing, marketing services. It will bring efficiencies in everything.”

AI’s role is multi-faceted covering a broad spectrum of areas: scouting, coaching, athlete health, officiating, ticketing and even the matchday experience for a fan. Earlier this week, AI even created a FIFA presidential candidate to rival Gianni Infantino, Hope Sogni.

Created in partnership between creative agency Dark Horses, Twise.ai and the CEO of Lewes FC, Sogni launched a hypothetical election campaign.

AI’s impact is already playing a role in the transfer successes of Brentford and Brighton, and their seeming ability to punch above their weight in the top flight. But clubs are, for the most part, closely guarded about its use. Both Brentford and Brighton politely declined to be interviewed about their AI use.

Chelsea have been open about their scouting approach with ai.io, a business launched by Darren Peries when his 16-year-old son, Reef, was released by Tottenham.

Approached by other clubs regarding his son, he found there was no technology for amateur players to produce reliable data which could help clubs’ scouts unearth players.

It brought him into contact with sports scientist Richard Felton-Thomas, now the chief operating officer of the business.

Six years on, the company launched an app last summer, aiScout, where players can upload certain drills which are then in the system and available to the 30-plus clubs in their network, including Chelsea and Burnley.

“Before the point where you hope and dream a scout comes to watch you, we can put you in front of them with just your mobile phone,” said Felton-Thomas. A former goalkeeper who had trials at Arsenal and Watford, he says he would have benefited from such an app in his youth.

Richard Felton-Thomas of ai.io

What ai.io is potentially able to do is unearth hidden gems who have fallen beneath the cracks and/or use data from academy players, at say Chelsea, to see a benchmark to which other players compare. It also has a state-of-the-art mobile lab which provides real-time data and analysis, and travel to any location.

In an initial test back in 2018, they picked 50 players from a college local to Chelsea’s training ground who had given up on playing professionally. It was agreed the best player would team up with Chelsea’s academy for a day.

In the end, Ben Greenwood stayed 10 weeks and scored on his Under-18 debut. While he was not signed by the club, he is now at Bournemouth.

Now their digital scouting mobile phone software is the only one of its kind in the FIFA innovation programme, and Felton-Thomas praises Chelsea for their openness.

“The natural tendency from someone is that you feel you have the inside track and, if you keep it secret, you’ll get an advantage,” he said. Chelsea’s thinking was that by being an early adaptor, they would have more data and also get more player approaches, which has been the outcome.

Professor Gopal Ramchurn has been involved in AI for 20 years, and set up Sentient Sports in 2020 with the aim of taking cutting-edge research from the laboratory into sport. The company recently added former Manchester United executive vice-chairman Ed Woodward to its board.

Ramchurn’s first foray into sport came courtesy of PhD students, who came up with an AI fantasy football agent, which beat the vast majority of people in the game.

Of AI in football, Ramchurn, who was this year appointed CEO of Responsible AI UK, said: “It’s still in its infancy. It’s going to be huge, creating products that you never imagined before.”

Gopal Ramchurn of Responsible AI UK

Among myriad changes, he says it will get to a point where coaches will be able to have virtual coaches aiding their decision making in games.

He said: “The latest algorithms will enable us to tell the manager which players to swap in, the tactical set-up of the team, even the state of the players, how tired they are and how likely they are to get injured.”

Any fears AI might take over entirely are dispelled by the professor, the majority of whose work for the last 10 years has been looking at human-AI partnerships. “How do you combine humans to get the best output from AI and AI get the best output from humans, for humans to do the right thing?” he added.

“That requires a good understanding between the human and machine. When AI rubs up against the physical world, that’s when lots of sparks come out. You’re forcing something artificial onto the real physical stuff. You always need a human to mediate that stuff.”

Data is central to AI in football and it is increasingly being harvested in everything from GPS trackers on players to sensors in their boots. Ramchurn argues it will come to a point where a player’s data — much like image rights — will be a part of transfers.

“People are not fully aware of the value of data just yet,” he said. “Access to a player’s data should be part of the transfer process. It will matter.” Aldo Comi argues that football is the “sport most lagging behind when it comes to data analytics”.

His Milan-based company, Soccerment, was set up in 2017 and works in data acquisition and predictive analytics. Last week, it launched an app called AIDA, mobile AI data analyst for players, coaches, fans and media. It covers 12 countries and includes all of Europe’s top leagues.

The use of the model can range from guiding you over fantasy football selections to acting as an AI member of backroom staff.

Comi believes there are a number of areas where AI will impact football — and he argues that officiating will become one of the first things majorly overhauled by AI.

“Linesmen will be the first refereeing figure to leave the game,” he said. “Already, they are not as influential as they were a few years ago. I wouldn’t be surprised to see AI assisting referees. I don’t think you will see linesmen in 10 years.”

As for clubs in London, the wider UK and beyond, each are increasingly embracing the use of AI in its different formats. Those that do not, argues Felton-Thomas, will become the also-rans.

“Clubs will have to embrace it or will get left behind,” he said. “It’s such a big thing and is still very young. There’s a lot to come.”