Football analysis has come a long way since the heady days of the first expected goals model.
With advancements in event data, player-tracking data and even limb-tracking technology, the value of a striker’s back-to-goal actions and off-ball runs are just as measurable as their goal tallies.
From machine learning to neural networks, the latest trend is artificial intelligence (AI). There are many sub-sections of AI, but its core principle is simply the ability of a computer to perform tasks we typically associate with humans.
Advertisement
Work within this space has been established for many years. For example, Zone7 is an AI company that works with clubs to forecast injury-risk using machine-learning methods.
On the pitch, semi-automated offsides are managed by AI analysis to monitor the movement of the ball and player to generate its decision.
At club level, Liverpool recently worked with Google DeepMind to create a “TacticAI” platform that improves strategy for defensive corners.
But the biggest impact of analytics in football is from a recruitment perspective. Investment has created research teams that operate more like a Silicon Valley tech start-up than a club’s data department.
Unfortunately, not everyone down the football food chain can build that infrastructure on a limited financial budget, but is that starting to change as AI grows in prominence?
The rise of generative AI — think ChatGPT — has transformed how many people engage with technology. At the risk of getting too technical, this method is underpinned by large language models (LLMs) that are trained on a huge amount of data to recognise patterns, allowing them to produce content including text, images or audio.
Soccerment’s xvalue and SentientSport’s ScoutGPT are two examples of generative AI models being used within player scouting. While the statistical analysis might be complex in the background, these platforms allow users to engage and ask questions about a specific player in simple football language.
For David Sumpter, co-founder of analytics company Twelve Football, the democratisation of such platforms means that data science is no longer reserved for those operating at the elite level. In the modern day, clubs have the opportunity to chat with the equivalent of a data scientist by simply clicking on an app on their phone.
“Some of the biggest clubs in the world engage with this, but it can be used by clubs in the National League or League Two,” Sumpter told The Athletic. “Essentially, they can have a data science department that is on the level of a Premier League club. That’s the dream.”
Advertisement
With clients across the top five European leagues, Twelve’s latest release has been its AI-powered analytics tool, Earpiece, delivered via WhatsApp. It enables clubs to explore the strengths and weaknesses of a player, initially without a number in sight.
While data and visualisations can be interrogated at the user’s discretion, Sumpter spoke of the platform as the “wordilisation” of complex information into a simple message — providing a bitesize analysis of a player as though you were chatting with a coach or sporting director.
The output has already helped clubs, with Sumpter highlighting the role it played in the January transfers of a League One club, whose performances have helped the team shoot up the table in recent weeks. The same is true at the player level, with reports sent after a game regularly read by those in the squad to assess their match performances.
As simple as it may sound, removing barriers to access is not to be underestimated when considering the limitations that some clubs face in terms of basic infrastructure.
Speaking with multiple people working within the game, there was a unanimous agreement that WhatsApp was the tool most used to operate and communicate in the football industry.
The app’s end-to-end encryption might play a part, given the confidentiality and high stakes that exist within football. Encryption keeps personal messages and calls between the parties communicating, meaning nobody else — not even WhatsApp — can read, listen to or share them.
A platform that operates via WhatsApp can feel like messaging a scout or data scientist on your journey home. The image below shows an example message exchange when looking for a player analysis of Bournemouth left-back Milos Kerkez.
“It’s incredible how WhatsApp works in football — most people don’t use anything other than that,” Sumpter said.
“From an early stage, we found that we needed to send match reports on a PDF (document) via WhatsApp. That is the way that the coach wants things communicated, and this is pretty much across all clubs — from Manchester United to the smallest clubs in the world. A WhatsApp with a report is what people use.
“We recognised quickly that this is not just the coaches, but the chairmen too. Most people in football don’t sit with a laptop because they are out on the pitch, at the stadium, talking to people. So we have had to have a solution that really communicated directly with these people.”
As The Athletic recently reported, the transformational effects of AI could threaten the future of traditional scouting, with many fearful of losing their jobs.
The skills required to evaluate a player’s subjective, qualitative attributes cannot be replaced by a single algorithm, so scouts should not be too affected — but could data scientists be replaced?
“There are nuances to it, but we have seen that many EFL clubs have recently hired a data scientist or two and said, ‘Let’s do data science and see if we can solve everything’,” Sumpter said.
“Those people will be happy working in football, but if another company comes along that can build a tool that will do data science within football, that can be a big risk for them. A lot of their job might be time-consuming tasks like data engineering or creating data pipelines (to help with their workflow) and it is not easy.
Advertisement
“Now, a lot of that work can be replaced by these AI systems, where you can bring in all the data and send it directly to the decision-maker. So the chairman or sporting director can have equal access to the information at hand.”
Though these platforms can analyse and present detailed information, an element of caution is still required when considering whether such technology can supersede a human analyst.
The soft skills that come with data science roles are where analysts truly earn their money, evaluating the strengths and weaknesses of certain analyses to form their conclusions.
“For the scouting department, there are plenty of restraints as to why certain players are not being looked at, but now suddenly the chairman has 60,000 options from the data using platforms like this, so you have to get that balance right,” Sumpter said.
“It is still important to have a data scientist in the club who can offer that experience or balance. It’s not just about being able to pull a machine-learning model off the shelf, but being able to interpret those models and look at their limitations — those skills are going to become much more important again.
“There is a buzz around AI, but those people who can show critical thinking will be valued at a premium.”
The democratisation of data insight is a huge positive in the wider football landscape, but data departments should not fear for their jobs just yet.
(Top photos Getty Images)
