Profil

Profil
CV - Click here

Mittwoch, 19. April 2017

The numbers don't lie: why football clubs place such importance on analytics

Leicester's direct style makes them a distinct Premier League outlier in the data


In a quiet corner of West Ham United’s Chadwell Heath training ground, long after the players have left for the day, Rory Campbell is staring intently at a computer screen. The residential surroundings have changed only superficially in the 50 years since this was a second home to Bobby Moore and Sir Geoff Hurst but the complexity of the off-field preparation is transformed.
Campbell is West Ham’s technical scout and analyst. An Oxford graduate, one of Alastair Campbell’s sons and a successful poker player with a moderate playing and coaching background, his focus is to make sense of the infinite statistical data about football and then communicate what really matters to the club’s key decision-makers. It is a specialism that has been profitably applied to sports betting but is increasingly now being employed across the professional game.
N'Golo Kante has been one of the signings of the season and a triumph for football analytics
N'Golo Kante has been one of the signings of the season and a triumph for football analytics CREDIT: EPA
What’s more, the potential correlation between some of football’s most effective analytic operations and a Premier League table that has rarely been less reflective simply of spending power is obvious. Leicester City and West Ham meet today but how did they, for example, identify N’Golo Kanté, Dimitri Payet and Riyad Mahrez at a combined cost of £16 million while Manchester United were splashing out more than £70 million on Marouane Fellaini, Ander Herrera and Bastian Schweinsteiger? And what led Tottenham Hotspur to Dele Alli and Eric Dier or Southampton to Sadio Mané and Virgil van Dijk? Why are teams crossing less than ever? What is the unique characteristic of Leicester? Why do managers like Pep Guardiola discourage shots from distance? And, in this most surprising of all seasons, is it really true that the table never lies?
 
Paid content
Campbell is West Ham’s technical scout and analyst
Campbell is West Ham’s technical scout and analyst CREDIT: DAVID ROSE
Analytics provide at least part of the answers, even if Campbell is adamant that their value lies in supporting rather than somehow replacing the experiences, intuition, innate knowledge and contacts that still form the basis of decision-making. “Any market with inefficiency is an opportunity,” he says. “The fact that football doesn’t have a set or agreed way of valuing talent and is so arbitrary is an opportunity. There is a difference between statistics and analytics. Statistics tell you about events that have happened. They don’t mean anything without context.
Leicester's direct style makes them a distinct Premier League outlier in the data
Leicester's direct style makes them a distinct Premier League outlier in the data CREDIT:DAVID SUMPTER
“Analytics is interpreting those stats to predict future performance. You can measure everything. The hard bit is working out what’s important. One good thing is that football is quite simple. Everything must relate somehow to goals, whether that is enhancing our chances of scoring or preventing them. It must also work within a framework of how the manager wants the team to play.”
Further insights can be found at Southampton’s training complex, where the most striking room is also the base for Ross Wilson, the club’s 34-year-old director of scouting and recruitment. Straight in front of him is a row of 15 screens at which a team of young staff are processing information. Some are interns direct from degree courses in the specific field of football analysis. To Wilson’s right is a greyer haired contingent, including Rod Ruddick, the scout who discovered an eight-year-old Gareth Bale on the playing fields of Newport. To Wilson’s left is a door with the words “Black Box” hanging mysteriously on it.
Paul Mitchell was lured from Southampton to Tottenham to become their head of recruitment and analysis
Paul Mitchell was lured from Southampton to Tottenham to become their head of recruitment and analysis CREDIT: SOLENT NEWS
Southampton are constantly modifying the bespoke software used in this mini-cinema and, with just a few clicks, can be watching just about any player in the world. Other clubs are developing similar technology and, among all these mathematical talent spotters, a transfer market in itself has emerged. Arsenal tempted Ben Wrigglesworth from Leicester this year and have spent £2 million on buying statDNA, a US-based analytics company. As well as Mauricio Pochettino, Tottenham recruited Paul Mitchell from Southampton as their head of recruitment and analysis. “I work off the simple theory that I had one good game once but I think the 80 other times I played I wasn’t so good,” says Mitchell, whose career was ended by injury at 27.
Just as in poker, Campbell calls player recruitment “managing the economic risk of your bet with the available information” but stresses a further point which, upon meeting other members of his industry, is clearly crucial. Where attempted innovation in football has previously fallen, it has often stemmed from lapses in communication or clashes of personality.
Sir Clive Woodward had a better relationship with Harry Redknapp than commonly assumed but add in Rupert Lowe, Dave Bassett, Dennis Wise and Simon Clifford and you hardly need the detective powers of Hercule Poirot to deduce where it might have gone wrong. Where analytics is making a difference, the culture is usually aligned. “There will be some clubs steeped in the tradition that you need to have played but, fortunately the clubs I’ve been at, the mindset has been very open,” says Wilson.
Campbell adds: “Where I think the analytics world has struggled is building a bridge to the traditional football world to infiltrate the information better. It is actually quite presumptuous to give out loads of information, that makes perfect logical sense from a mathematical standpoint, and expect a sport that has developed for decades to accept it overnight. I would say this remains the biggest challenge. You have to understand the dynamics and the personalities of the people you are working with to be able to communicate the information. I think that is why analytics has not penetrated football like maybe other businesses would have expected.”
Arsenal manager Arsene Wenger has also embraced analytics
Arsenal manager Arsene Wenger has also embraced analytics CREDIT: GETTY IMAGES
This is changing, however. Campbell says he is very lucky to work under Slaven Bilic and Tony Henry, the director of player recruitment, and there is clarity to everyone’s role. What they want is simply a trusted evaluation on which to help inform their decision. Older managers are also reaching out. Claudio Ranieri is one example. Arsène Wenger caused a stir this season when he publicly referred to Arsenal’s “expected goals” [xG], which is the key measure within sports betting and analytics of how often a team were statistically likely to score.
Borussia Dortmund coach Thomas Tuchel has sought out Matthew Benham to learn more about xG. Benham made millions in sports betting and has since bought Brentford and FC Midtjylland. Like Campbell, who also travels extensively to watch players in person, Benham has stressed the complementary importance of “scouting with the eyes” as well as the significant variance of any maths model in a sport as low scoring as football. Just as in poker, random and uncontrollable events play a part that is frequently overlooked amid the rush to form judgments.
“This is something that makes football exciting but unpredictability always presents significant inefficiencies,” says Campbell. “Professional poker players who moan about luck are narrow-minded. Luck is what enables them to make a living. If I play poker with a really bad player, he might win 40 out of 100. If I played Garry Kasparov at chess, he would win 100. Chess players don’t make money because no one bets them.”
Leicester's fairytale season has confounded the statistical models
Leicester's fairytale season has confounded the statistical models CREDIT: REUTERS
Analytics are also competing to be heard within an increasingly opinionated landscape. “Arsène Wenger said we have moved from a vertical to a horizontal society,” says Campbell. “The vertical is where you have a leader at the top and everyone follows. The horizontal is where you have a leader bombarded by information and opinion. That’s where the leader has to be so clinical about what’s important and what’s noise.”
And so back to those earlier questions. The likes of Kanté, Mané and Payet were ultimately astute football decisions but long triumphed in the analytics community for their underlying performance indicators. The clear statistical evidence is that crosses are a low percentage tactic compared to through-balls and that shooting from range yields fewer goals than trying to pass into better positions.
As for Leicester, according to research by maths professor and Soccermatics author David Sumpter, one striking difference with the rest of the league is how they get the ball forward quickly with relatively long and straight passes.
And that cliché about the table never lying? Well, perhaps it just does not tell the whole truth. Virtually every xG model says that Arsenal have indeed missed a huge opportunity and should be top. Most models would have Leicester between fourth and eighth if this season had been played an infinite number of times. Variance and luck, then, remain sizeable over even a 38-game programme. Yet the gaps are narrowing and, if the past year has been refreshing proof of anything, it is that working hard and smart really can count for more than the size of a club’s bank account.

How analytics inspired five success stories  

Billy Beane and the Oakland A’s
The story that inspired Brad Pitt's Moneyball film and statistics across numerous sports. Working with Harvard statistician Paul DePodesta and by using sabermetric principles, Oakland achieved unprecedented success. Beane has led Oakland since 1997 but admits that he would one day be tempted by a job in football.
Team Sky in cycling
Sir Dave Brailsford’s philosophy was to work back from the question ‘How do we win the Tour de France’ and one key analytical innovation was to establish the power output required throughout the key moments of an entire three-week race.
English rugby under Sir Clive Woodward
The first coach to adapt Prozone to rugby, installing it at Twickenham four years before the 2003 World Cup triumph. "It made a big difference – it removed a lot of the preconceived notions we had about how other teams played,” he says.
Matthew Benham in football
Made a fortune in the sports betting industry and has since bought Brentford and FC Midtjylland where he has applied analytics to decision-making. Midtjylland won the Danish league last year for the first time while Brentford recorded their highest league finish since 1947.
Formula One and Paddy Lowe
Arguably the most data influenced of all sports. Virtually every innovation can and is clearly measured. Lowe, a Cambridge engineering graduate who is now the executive director of Mercedes, is credited with taking analytics to a new level.

Keine Kommentare:

Kommentar veröffentlichen