With the data genie now firmly out of the bottle, it is a case of keep up or fall behind among elite football clubs.
Modern-day recruitment has evolved dramatically in recent years, with the use of video scouting and data-driven insights serving to optimise time and minimise risk when searching for that 20-goal-a-season striker or that box-to-box midfielder.
It is one thing to invest in data, but it is another to successfully embed it within a club’s workflow. You only need to look as far as Brentford and Brighton to see how data has been used to maximise the market and allow both clubs to punch above their weight in recent years.
Some fans might still be resistant to the idea of data in football, but few can dispute its utility as a tool to inform decision-making when scouting a player.
Follow the summer transfer window with ** ** The Athletic …
As humans, we are filled with biases — often with good reason. They provide us with rules, shortcuts, and heuristics to help us make sense of the world, but these biases can be fallible.
Of course, traditional scouting should never leave the game. It allows you to assess the body language, the temperament, and the character of a player. Nevertheless, it is widely accepted that this method is neither time nor cost-effective.
Besides, does traditional scouting work any more than flipping a coin?
Data is transforming the way that most clubs operate, and its trajectory is becoming increasingly sharp, with investments in data scientists becoming almost as important as the players themselves.
In a world where machine learning, computer vision, and neural networks are becoming increasingly recognised terms within football analysis, the final step of such work is arguably the most important one — how can we use information from the past to predict the future?
“There’s been some really good examples of Premier League clubs producing new metrics and new ways of looking at data in the last few years. But this often looks at how a player is performing now, with their current team-mates in that current team,” Dr Ryan Beal, chief executive and co-founder of SentientSports, explains to The Athletic.
“There is huge value to backwards-looking analytics, but what we’ve always done is look at changing that to forward-looking situations to challenge scouts on how well they think a player will adapt — using the numbers.”
GO DEEPER
Casemiro, Sterling and Scamacca: Analysing transfer suitability and who else was out there
The umbrella term to encapsulate this next stage of data analytics in football is artificial intelligence (AI). At its core, AI is simply the ability of a computer to perform tasks that we typically associate with humans.
Voice recognition? That’s artificial intelligence.
Face identification on your phone? You guessed it.
Identify a shortlist of players who best fit your team’s style of play? Now we’re talking.
The application — or rather translation — of these findings is where you maximise your impact. Without the message being delivered clearly, the insights that are generated can reach a cul-de-sac pretty quickly.
While spreadsheets, dashboards and data visualisations are not without their merit, SentientSports’ latest release channels the new generation of AI to help users understand information in an accessible manner.
This is using a form of AI called “Generative AI” that uses existing information to produce different types of content, including text, images or audio.
OpenAI’s ChatGPT might be the most well-known example known to many, but anyone who has used an online chatbot to help them pay their phone bill has already engaged with this powerful tool.
For Beal and his colleagues, they have designed a bespoke, football-specific version of ChatGPT — called ScoutGPT — that uses live data to allow any user to ask a question about a player or team and receive an answer, underpinned by complex analytics, in simple football language.
This technology has brought together some leading minds from across the industry including ex-Manchester United executive vice-chairman Ed Woodward and Professor Gopal Ramchurn who is the CEO of Responsible AI UK — the UKRI’s flagship AI programme.
Let’s run through an example.
Who are the top five strikers under 25 years old who press the opposition and are valued under £30million ($38m)?
A comprehensive explanation provides a shortlist including new RB Leipzig striker Lois Openda, Eintracht Frankfurt’s Randal Kolo Muani, and Athletic Bilbao’s Oihan Sancet — underpinned by SentientSports’ analytics models.
From the initial reply, the user can probe further by asking questions about the output provided. Crucially, it is designed to work for those who will use this platform the most.
“We’ve had the CEOs of Premier League clubs using it, as well as fans, club analysts and agents. We’ve really tried to broaden the net to get feedback from different user groups to work out how they would use it and what they would use it for.” Beal says.
Specifically, there is a huge potential for AI technology to transform the future of digital experiences for fans, by equipping them with the very tools and analytics that clubs use within their own decision-making.
For example, AI can help clubs to engage with fans in their native languages, fostering a sense of connection with a team that is potentially based thousands of miles away.
Building this direct engagement between club and fans can pose real challenges for clubs, but the SentientSports team are looking to bridge that gap, drawing on the experience of Woodward’s time at Manchester United.
The platform can also be used at the team performance level. Premier League analysts will no doubt be getting the breakdown on newcomers Luton Town, but a quick search into the style of play highlights Rob Edwards’ side as a direct, high-pressing team who generate chances well from crosses.
A neat starting point for analysts to dig further into the video.
Why not go straight for the hottest topic of the summer transfer window?
With Harry Kane securing his move to Bayern Munich, fans can simply search the 30-year-old’s suitability for Thomas Tuchel’s system with a comprehensive scouting report — outlining where his strengths and weaknesses lie.
For Beal, that translation from complex analysis to football language is crucial. The technology must adapt to the user, whether that is a fan, a data analyst, a sporting director, or a CEO.
“It’s about that connection between human and data science for someone who hasn’t got a PhD or doesn’t understand the mathematics that underpin it, but they can talk in that football language,” Beal says.
“I’m sure any data scientist you speak to in a club would say that, no matter how good the data science is, the hardest part is explaining this to someone who is a ‘football person’ and wants to talk in football terms — and I think the people that can do that are probably the best, the most respected data scientists in those clubs.”
The application and productisation of AI is growing in the football world, with companies such as aiScout allowing grassroots players to log their performances on an app in a ‘virtual trial’ that can be assessed by clubs.
Crucially, the use of such data should not be the end point but the starting point — a tool that can support the decision-making process, not replace it. Like any technology, these platforms come with limitations as well as strengths, but remain a powerful device if used optimally.
“We’re not saying to use AI to take over transfers, take over people’s jobs. We want to make people more efficient, and help them to make better decisions,” Beal says. “Trust is a huge part of that. People might say, ‘We trust our scouts with 20-30 years of experience predicting future performance, why should I trust the data?’. It’s our job to back-test our models, and prove the value to them — and it takes time to build that trust.”
Such AI predictions cannot be 100 per cent correct but they provide crucial risk assurance to human decision-makers, allowing users to combine their own “gut feeling” and analysis with that of the platform.
Having an interface where people can ask questions about the outputs is where Generative AI comes into its own. If a sporting director is unsure of the information provided, they can respond, ask for further examples and create a dialogue with the AI technology that is underpinned by ScoutGPT.
While football will always be for the entertainment of the fans, the spotlight on the business of football has become brighter than ever in recent years. To have a successful business, you need to optimise your processes and using cutting-edge technology is often a key way to achieve that.
Whether it is companies such as SciSports, SkillCorner, Opta or StatsBomb, AI is growing within the football industry. As clubs become increasingly attuned to this innovation within the game, this new chapter of AI could change the landscape entirely.
