Possession stats are one of modern football’s great obsessions — they appear on almost every post-match graphic, at the end of countless match reports and we spend endless hours studying them, even during the course of a game.
But there is growing evidence they are almost obsolete in terms of what they tell us about the performance of a team.
Last Saturday, out of eight games played in the Premier League six were won by the side with least possession and yet commentators and amateur statisticians still talked animatedly about what the figures meant.
Looking back Arsenal, by all accounts, were the poorer side at Stamford Bridge as they were beaten 3-1. Some reports even talked about Arsene Wenger’s side being bullied and dominated as their title dream took a traditional dip; and yet they enjoyed 58% possession. On the same afternoon Everton thumped Bournemouth 6-3 in a thriller at Goodison Park with only 42% of the ball — while Liverpool enjoyed 72% possession at Hull but still lost 2-0.
The confused story is also played out in the Premier League table where runaway leaders Chelsea are nine points clear and widely recognised as the strongest team in the country; but if the table was drawn up purely on the amount of ball possession then Antonio Conte’s men would sit joint sixth alongside Southampton, a team that in reality has picked up less than half the number of points.
Leicester City are perhaps responsible for bringing the debate to the fore. After all, they were ranked only 18th in the Premier League based on possession last season but were still crowned champions. Their percentage of 42.3% was the lowest of any Premier League winner in the previous 10 years. So, is it time to find a different measuring stick?
Rasmus Ankersen is joint director of football at Championship side Brentford, a club famous for embracing the use of in-depth performance statistics as part of their recruitment and playing philosophy, and his views are clear cut.
“We don’t look at possession stats at all,” he revealed. “There are a lot of better metrics available which tell you far more about the effectiveness of a team’s performance. I don’t think bright coaches in the game look at them either — it’s more of a media thing — and I suspect we’ll be far less obsessed with them in future as more useful measurements become available in the public domain.
“I always say there is a split in football between the IQ group and the EQ group. In the IQ group you have the statisticians and in the EQ group the coaches and managers — but they don’t communicate. This means the IQ group often produces statistics and measurements, like possession stats, which are not actually useful to the football side. We should be asking a question and using data to answer it.
“The more important metric for us to measure is the creation of meaningful chances which could lead to a goal — but it doesn’t matter how you get there or how much possession you have. You could get there the Stoke way or the Barcelona way.”
The history of possession stats, and their perceived connection with success, dates back to the ‘tiki taka’ philosophy of Johan Cruyff which was later embraced by Pep Guardiola at the Camp Nou. Yet it appears to have taken a strong hold.
In England, the backlash to the infamous coaching manual of Charles Hughes, the director of coaching at the FA in the 1990s, has been almost puritan in its intensity on the back of a modern possession obsession. Hughes’ belief that most goals are scored from three passes or fewer was derided by many as long ball football. However, primarily he stressed the importance of areas of the field from where goals were most often scored — Positions of Maximum Opportunity (POM) — and the importance of getting there quickly.
Leicester did that often last season, although they did so with pace on the break and swift incisive passing in the final third rather than with a long ball, inspired by the hard work of Kante, the speed of Vardy and the more flamboyant skills of Mahrez.
In a way there are echoes of Hughes’ research in modern performance analysis but with a far more open-minded approach (in fact an entirely neutral approach) to how teams reach those POMs — and a much greater understanding of the randomness of football.
Ankersen believes his metrics could burst the Leicester fairytale, saying: “The problem with football is that because it is a low-scoring sport there is a lot of randomness and variation. So, we look at two league tables at Brentford. The real one and our own one (we call it the table of justice) which shows where we deserve to be based on performance — and that’s what we run the club on.
“Last year, for instance, Leicester won the Premier League but most models showed them as only the fifth best team in the league. They had a 5% chance of winning the title but converted chances at an unsustainable rate in the first half of the season when Vardy and Mahrez did so well. That figure slowed down to the norm eventually — but at the same time in the second half of the season the defensive performance of their rivals dropped. So they got lucky twice.
“That doesn’t work for the media — you can’t tell the story like that because the brain likes a narrative. It prefers to think it was down to genius coaching or team spirit. Data cuts through all that. We think the biggest lie in football is that people say the table never lies – it does.”
So, too, do possession stats — or certainly they are being read in the wrong way and given too much credence.
Ankersen added: “There are better metrics to use. We look at shot probability and the expected goals method which measures the quality and quantity of chances created. How often are we reaching positions where the probability of scoring is high? Then you look at how well you are preventing the opposition getting into those positions — and finally you can measure the probability of winning.
‘Of course, those measurements can also help influence the style of football played but communication with the football side is key. The statistics have to be useful if they are to help give the team an extra edge. Possession stats certainly don’t do that.”
© Irish Examiner Ltd. All rights reserved