From unlocking football's 'magical' unknown fifth element using brain tests to the AI-wielding 'unicorns' identifying talent around the world, as part of the Future of Football series Sky Sports takes a look at the innovative people and technology changing recruitment
Thursday 10 August 2023 12:40, UK
"These unicorns they do exist."
There are a special breed of of people that will change the way transfers and recruitment in football are done for years to come, according to Real Analytics' Ian McHale, also a professor of sports analytics at the University of Liverpool.
The appointment of Laurie Shaw as lead AI scientist at the City Football Group, the parent company of Manchester City, is a signal that this shift has already begun.
Shaw, formerly an astrophysicist at Yale University, left his role as a research scientist and lecturer at Harvard University to join last season's treble winners in 2021.
Astrophysicists are better known for studying black holes than football but could they, and other highly qualified scientists and mathematicians, really become key figures at Premier League clubs?
"I think this will happen eventually, and it will certainly be the case that the quants [short for quantitative analysts] are listened to," McHale tells Sky Sports as part of the Future of Football series.
Being a brainiac or a genius won't be enough, though.
Like Shaw, who specialised in data analysis of team sports at Harvard, McHale explains that clubs will need people that can combine a technical understanding with a love of the sport.
"These unicorns, they do exist, I'd like to think we have them at Real Analytics," says McHale. "Those will be the people who you will eventually need on the board, or helping the board, and the people who sign the cheques, to make better decisions."
Jens Melvang, a former footballer now working as senior product manager at Stats Perform, the parent company of Opta, shares a similar perspective. Part of his role is translating the more technical aspects of the data they produce for coaches at football clubs. However, integration, it would seem, is not always easy.
"I often see that data scientists can be isolated at clubs," he tells Sky Sports. "That's a shame because they are adding a lot of value. So the biggest obstacle is making sure these more technical people are part of the process."
Football, apparently, hasn't quite woken up to the full power of the resources at its disposal. Here, Sky Sports takes a look at those putting science, data and innovation at the heart of their decision-making in transfers and recruitment.
What sets an elite footballer apart? That was the question Eric Castien, the founder of BrainsFirst, posed to scouts at Barcelona and Real Madrid as part of his research for a book he was writing about talent identification in 2011.
There was an agreement across the board that talent can be broken down into four elements: technical, tactical, physical and mental. But there was another, rather unfathomable, fifth aspect.
"Some of them call it magic, others the black box," Castien, a former journalist, tells Sky Sports. "You either have it or you don't.
"When I asked them to point to it on their bodies, they said, 'Definitely it is here in between the ears'. I said, 'Ok, maybe it is brain function'. They said, 'Maybe you are right, but we don't have the understanding of this'."
Castien took this theory to neuroscientists whom he had interviewed for his book. Somewhat bemused by the suggestion of magic, they agreed with his hypothesis about brain function.
They needed four years to show they could pinpoint the brain functions needed to be a top player and a further six to demonstrate they could use it to predict the future.
Now youth team players at clubs in Europe, like AZ Alkmaar and PSV Eindhoven, are having their cognitive performance measured with brain tests to establish if they have what it takes to be elite footballers. BrainsFirst are measuring magic.
"Today we understand that cognitive performance is as crucial as physical performance for football," Castien, whose role was to bridge the gap between football and neuroscience, explains.
"We now have the research to understand what football at the elite level requires from the human brain. If you have an assessment of someone's brain you can compare that benchmark with their threshold levels. If you understand threshold levels and the brain scores then you have a match or not."
Both clubs' youth academies have experienced great success off the back of this partnership. AZ Alkmaar lifted the UEFA Youth League in April and Jong PSV were crowned Premier League International Cup winners in May.
Jurrit Sanders, the lead sport scientist at PSV's academy, tells Sky Sports the cognitive tests are helping the club understand "a new piece of the puzzle in football".
"The programming and the training we deliver are much more detailed now than before," he adds, "especially when we are focusing more and more on the individual athletes and individual programming based on outstanding skills."
Once a young footballer crosses the threshold at PSV's facility one of the first priorities is getting their brains tested.
"We are measuring about 53 cognitive factors," says Castien, whose company now conducts these tests in the form of games. "To play football at the highest level you at least need to score very highly on about 10 or 12 of them.
"If you want to perform at a high level in school, there are maybe five cognitive factors in common for being a top-level footballer. But there are seven or eight others as well."
This technology has created an objective measure of ability, eliminating bias from the scouting process, while recalibrating the way clubs identify talent.
But which clubs are most interested in working with BrainsFirst?
"Those very hungry to outsmart the competition with innovation," says Castien.
"I think it is not by coincidence that now a club in England like Southampton chose BrainsFirst. This is about innovation and if you have to challenge the status quo, if you want to become one of the best elite teams in Europe, then you have to do something different.
"If you copy them you will lose because the budget is smaller and now with this technology you can try to jump higher. That is the reason why the challengers in Europe choose this innovation."
Brain scans aren't the only way to learn about a player's cognitive skills. Companies are beginning to harness the power of AI to produce data that can provide insights about the decision-making and performance of footballers around the globe.
This allows clubs to spread their net far and wide to identify the talent that may otherwise go unearthed until established in a major league. Think Moises Caicedo to Brighton from Ecuadorian club Independiente del Valle for £5m in 2021.
For so long, event data - what happens on the ball - has only given us a fraction of the story. There is, of course, so much more going on around that, which is where tracking data - everything that happens off the ball - comes in, enabling companies like Real Analytics and Stats Perform to provide a more holistic understanding of a footballer by combining the two.
"The event data gives you the what, then the tracking data enables you to start looking at the why," Brad Griffiths, the senior vice president of innovation at Stats Perform, tells Sky Sports.
"The key thing that tracking data gives us is that it gives you the positional information and the movement of all players on the pitch all the time and that gives you context. Where you had your event data, you knew that a player had the ball and he passed to this player from this point in the pitch. What you don't know is what drove that decision."
The advancement of AI is making it possible to train a computer using deep learning to identify how a player reacts in certain conditions to predict how they would fare in a different league or fit into the system of a certain team.
"A Premier League club could look at South American leagues and see how the player behaves on and off the ball, how they run, how they react when under pressure from opposition players," says McHale, whose company were given a £100,000 grant by Nvidia to cover computing costs such is the power required to run these models.
"You can see how much their performance drops when they have less time on the ball, when they are directly against a faster and quicker player. That club could get a real insight into how the player might adapt to football in the fastest league in the world."
Melvang, senior product manager at Stats Perform adds: "We provide clubs with the tool that allows them to make these searches for players with the inputs of we would play this style, look for this type of player, this role of player and the system would come back with suggestions about this."
As data analysis becomes more sophisticated, it is going beyond just the performance of a player and beginning to shed light on the inner workings of their mind.
"You can start to see how players respond to different scenarios and it comes down to their cognitive and decision-making process," says Griffiths.
"Is a player perceptive at going, 'Well I know I can put the ball down there and my player is going to get there first, even though it is not going directly to him?'. How quickly are they making these decisions? Are they willing to run the ball past players or do they try and offload the ball as soon as they are under threat?
"You can see these things within the data that we are collecting that does help you gain insight as to how a player behaves on the pitch."
The emergence of chatbots such as Open AI's ChatGPT and Google Bard could have future application in the field of data analysis in football transfers.
Soccerment, a company that tracks and analyses data of performances in football, is already taking steps to integrate these language models into their product.
"We have rebuilt all the database so it is able to plug the main AI tools, including for instance, natural language models so we can interrogate our database through the APIs of ChatGPT and others," Soccerment's CEO Aldo Comi tells Sky Sports.
"This is allowing us to come up with what I think is a very powerful tool that will be published soon."
Soccerment's data analytics platform, xvalue.ai, is already delivering a product that makes it accessible enough for someone without expert knowledge, like a journalist, to search their database.
The platform enables the user to search for the most like-for-like replacement for another footballer. For instance, if Arsenal were to lose Bukayo Saka, this could help them identify which targets are most similar.
"Bukayo Saka has a function, which according to our clustering model, is to be a one-to-one explorer," Comi says.
"A player that does a lot of one-to-ones, a lot of expected threat from carries and so forth. So we compare Saka to those types of players only. We also have an AI algorithm which shows us the players that are most similar to Bukayo Saka."
Barcelona's Raphinha - a former Arsenal target - came up as the most similar player to Saka, which is perhaps why Mikel Arteta was keen on bringing the Brazil winger to the Emirates a few summers ago.
The integration of language models such as ChatGPT would only serve to enhance the experience of the user, making the platform more accessible for the less technically qualified.
"Our additional effort is to give you a virtual data analyst," adds Comi. "It will go into our little tool and ask who are the most similar Bukayo Saka and why. Basically the data analyst will answer in a very precise way, giving you the players but also why those names appeared in the answer. It will give the answer in a way that will be more comprehensible to you or anyone."
Language models are in their infancy and rapidly improving but, at this stage, still throwing up random inaccuracies, making it hard to rely on them completely.
"The natural language aspect of asking it questions and having data presented back to you is really exciting," says Griffiths, Stats Perform's senior vice president of innovation.
"The one thing we have to be very mindful of - certainly in the space where we are talking about sports data and recruitment - is the fact you can't make mistakes. You have to be getting data back that you know is reliable and is correct.
"It absolutely will have a place, it is something that we are already working on but it has to be leveraged in the right framework with the right guardrail so that you can be 100 per cent confident that what you get back is the truth and reliable."
