Viagra pour Rousseau

[Uilleam 5,409]

"AI is now expected to drive football analysis, scouting and coaching into a new age."

 

Just as success in cycling for a number of years -decades- depended upon the quality of your pharmacist, success in football may -will- come to depend upon the quality of your AI hardware (Viagra, indeed).

 

FOOTBALL | TOM ALLNUTT

How football is racing to take game to next level

They may be late to the party, but nearly every Premier League club is now experimenting with AI

Tom Allnutt

Sunday June 25 2023, 12.01am, The Sunday Times

 

https://www.thetimes.co.uk/article/how-football-is-racing-to-take-game-to-next-level-rq53mv9f2

 

Atalanta really should have won. They were the Champions League hopefuls up against Empoli, who were fighting relegation. They had 47 shots to Empoli’s three. They had 17 shots on target and seven from inside the six-yard box. According to the expected-goals tally, Atalanta should have won 6-0. According to the data, they played brilliantly. By every numerical metric, this was a hammering. The result was 0-0.

That match in April 2019 set a record in Europe’s top five leagues. Never before had a professional side been so dominant and failed to win. To many, Atalanta’s head coach Gian Piero Gasperini included, the match was an aberration, an anomaly to be forgotten. To others, there were explanations beyond bad luck: Gasperini’s lack of a plan B, the increasingly frantic attempts in the second half or the fact that Empoli had already beaten Atalanta earlier in the season.

These were the intangibles the numbers did not explain. This was the gap between the data and the reality, the gap artificial intelligence is now starting to fill.

 

Football has been slow on the uptake with AI. In the past 20 years, clubs and coaches, inspired by the Moneyball story of the Oakland A’s in baseball, have embraced a data revolution to deepen their knowledge of players and teams beyond what ever seemed imaginable.

But just as early enthusiasts for data faced resistance from the more old-school attitudes in the game, AI has found football a particularly awkward subject to master. Unlike other sports, football has a large number of players, all of them with fluid roles. They play in different styles and in long, uninterrupted sequences. What happens off the ball can be more important than what happens on it. There are few isolated events and even fewer goals.

 

Yet just as data broke through the scepticism, the rapid progress of AI is now expected to drive football analysis, scouting and coaching into a new age. Already, nearly every Premier League club is experimenting with AI, either through external agencies, at a cost of tens of thousands a year, or by hiring teams of experts in-house. It is a race of stealth, with every breakthrough a closely guarded secret.

 

Brentford, who use an algorithm to identify undervalued players, have shown the benefits of getting ahead. Liverpool have partnered with Google DeepMind to enhance their analysis of opponents. Chelsea are signed up to AiSCOUT, a programme that uses machine learning to assess the physique and technique of transfer targets.

Manchester City partnered with Google Research to develop an AI tournament, in which virtual players think intelligently and, according to the club, “test tactical concepts so that they are strong enough for a coach to stake their career on.”

 

Just as ChatGPT is now capturing information on the internet to mimic human responses, AI can pore over hundreds of years of football data and come up with intelligent conclusions. “There is more data now than ever before,” says Jelle Leenders from SciSports, a Dutch analysis company that works with Leeds United among others.

“It’s about dealing with all that data and how to bring that data to life. AI will take these applications to the next level.”

 

Some of the possibilities could transform how the game is analysed and understood. At Premier League matches, clubs already have a team of video analysts sitting in the stands, watching match footage on a screen, deciphering patterns to be passed on to the manager at half-time and after the game. But AI can “layer” all that data in real time, comparing a range of different scenarios and instantly coming up with predictions.

Researchers at DeepMind believe this means that managers could soon be working with “AVACs” (Automated Video Assistant Coaches) that could provide analysis on the bench while the match is happening. These video coaches could point out that the opposition left back is running 20 per cent slower than at the start of the game, suggesting fatigue, and that attacks down that side might have a better chance of success. They could advise a change of formation, based on how the opposition defence drops deep late in games, or substituting a player whose passing has deteriorated in the last few minutes. How could players complain when they are hauled off by a computer?

 

This could also revolutionise analysis before and after matches. When carried out by humans, reports on opponents are dependent on an individual’s skill and experience but AI can be more extensive in its research and less affected by bias. Specific training sessions could be devised to target vulnerabilities in the opponent’s line-up while pre-match pep talks could involve players listening to an AI “coach” detailing the strengths and weaknesses of their opposite number. Routine tactics could also be reconsidered.

A study by the School of Psychology at Queen’s University Belfast has used a virtual simulator to challenge the idea that goalkeepers should use walls to defend free kicks.

 

Scouting has already been transformed by data, with clubs using statistics to narrow down a list of targets that human scouts then assess. But there are limits to letting the numbers filter out so many possible signings. Clubs that spend hundreds of millions in the transfer market will pay to get better guarantees.

“You have to build metrics that say something useful about the data,” Leenders adds. “If a central defender plays five passes and they are all backwards, it doesn’t matter that they have 100 per cent accuracy. That doesn’t tell you about their quality on the ball. An AI algorithm can rate every one of those passes and its impact on the team. It can then look at thousands of passes, across several years, and hundreds of leagues, and tell you what kind of pass you want. You can’t do that in person.”

 

The limitations of AI are still considerable, which means technology will be there to support, not replace. AI could recommend tactics that offer the highest probability of winning but it might not be popular with the fans if the advice is to let the opposition have 90 per cent of the ball.

AI could recommend dropping an expensive new signing but it might not consider how that would disappoint the player’s team-mates or irritate the chairman.

Humans will stay in charge but how they understand and analyse the game is already changing millions of data points at a time.

 

[Rousseau 9,361]

Oh la la

 

[compo 6,122]

3 minutes ago, Rousseau said:

Oh la la

 

Memories 

[Gonzo79 13,589]

Another nail in the coffin of individuality and personality in football.  

[Scott7 5,150]

Does this AI stuff know if some continental bloke can hack a November day in Glasgow?

 

I can’t take AI seriously. My rural background tells me it’s something else.

[ranger_syntax 3,765]

43 minutes ago, Scott7 said:

Does this AI stuff know if some continental bloke can hack a November day in Glasgow?

 

I can’t take AI seriously. My rural background tells me it’s something else.

We will know soon enough.

 

First thing is first though.

 

Please send very many wheelbarrows of cash to the tech sector.

[Uilleam 5,409]

3 hours ago, ranger_syntax said:

Please send very many wheelbarrows of cash to the tech sector.

Bitcoin, surely?

 

 

 

 

Edited June 26, 2023 by Uilleam