sport news Football's A.I. future? How former Premier League player is showing top clubs ...

sport news Football's A.I. future? How former Premier League player is showing top clubs ...
sport news Football's A.I. future? How former Premier League player is showing top clubs ...

During the first half of last season, Sportsmail sat down with former Premier League midfielder Matt Oakley and the professor behind the recruitment tool that, they believe, will revolutionise the way in which clubs obtain and retain players. 

In short, their technology tells you what will happen, not what has happened. We needed convincing. 

Now, with the benefit of hindsight, almost every forecast made came to fruition. 

They spoke then of how, last summer, they were asked to run a 'player impact' report on Pierre-Emerick Aubameyang, prior to Arsenal's decision to award him a lucrative new contract, worth £55million over three years. 

Oakley and Professor Ian McHale, of the University of Liverpool, had warned how keeping the striker at the club was perhaps not a prudent move. 

A new artificial intelligence tool has predicted Harry Kane's potential value to Man United

A new artificial intelligence tool has predicted Harry Kane's potential value to Man United

It successfully predicted that the bumper new contract signed by Pierre-Emerick Aubameyang would not prove to be worth the money for Arsenal

Gunners boss Mikel Arteta could perhaps have benefitted from the tool

It successfully predicted that the bumper new contract signed by Pierre-Emerick Aubameyang (left) would not prove to be worth the money for Arsenal and Mikel Arteta (right)

Their system, Real Analytics, predicted that Arsenal would finish in an average of eighth position in the Premier League with Aubameyang in the squad, and ninth without him. It also projected that they would score 55 goals with, and 50 without. Arsenal finished eighth, with a goals tally of 55. 

'Being an Arsenal fan, like many others, we were getting excited about Mikel Arteta and the team last summer,' says Oakley. 'Our system said otherwise, with or without Aubameyang. Real Analytics opinion was that giving him a new contract was not worthwhile, in terms of the difference it would make in the team's performance for the cost of his wages.' 

Oakley and the team had presented us with several other reports on the back of them simulating the 2020-21 Premier League season one million times over. 

That included the data predicting Liverpool would not retain the title, while another stressed the difference Allan Saint-Maximin makes to Newcastle. According to their software, the Magpies would finish close to 17th without their talisman, as opposed to 13th with him. They ended up 12th. 

So how does it work? 

'There are a lot of companies doing 'data', but what they basically do is simple counting. That isn't analytics,' says Oakley. 

'What Professor McHale has created, that is real AI, it has 10 years' of data that is used to predict forward, rather than looking back.' 

Former Premier League midfielder Matt Oakley, pictured in action for Southampton in 1999 is a key backer of the A.I. tool

Oakley pictured more recently as MK Dons assistant manager

Former Premier League midfielder Matt Oakley, pictured in action for Southampton in 1999, (left) and more recently as MK Dons assistant manager (right) is a key backer of the A.I. tool

Real Analytics claims to be 'the future of decision making in football'

Real Analytics claims to be 'the future of decision making in football'

Oakley and Professor Ian McHale (pictured), of the University of Liverpool, created the technology that tells you what will happen, not what has happened. Now, with the benefit of hindsight, almost every forecast they made to Sportsmail last season came to fruition

Oakley and Professor Ian McHale (pictured), of the University of Liverpool, created the technology that tells you what will happen, not what has happened. Now, with the benefit of hindsight, almost every forecast they made to Sportsmail last season came to fruition

McHale explains: 'The numbers that go into it are called 'event data'. 

'Every match has around 2,000 events. A pass is not just recorded as a pass. We have the x-y co-ordinates of where the pass came from and where it went and who it went to. 

'Our AI engine learns the value of every action in terms of what it contributes to the likelihood of that possession ending in a goal. Every pass, every tackle, every interception, everything is given a positive or negative score. 

'It can be used for coaching to look at the moments where you have added value, or it can deem that you have taken value away. You can see how some players, who might be completing lots of passes, actually impact negatively on the team. 

'But when it comes to recruitment, it is no use knowing how good a player was last week. We need our tools to be predictive. We want to know how good he will be next year, or in five years' time. 

'It turns out, things like pass completion percentage reveal little about a player's future performances. 

'We have been working on these types of models for 20 years. They were originally built for the gambling industry and designed for

read more from dailymail.....

PREV sport news Beth Mead warns Arsenal are 'hungry' for revenge on Chelsea in Continental ... trends now
NEXT Jake Paul, Amanda Serrano in action: Fight card, odds, date, opponents, start time, complete guide