Statistics without context = Xmas without dinner

Date published: Saturday 24th December 2016 8:59

As we all know, there are lies, damn lies, and statistics. But while lies and damn lies have been around football for a very long time, statistics have only come to the fore in the past decade. Yet in that short time, they’ve become an invaluable tool for higher-level clubs, many of whom employ analytics personnel to break down the strengths and weaknesses of their own team and their opponents, ever in search of the winning advantage. And there’s this thing called the internet, which gleefully pixelates the numbers and spreads them to the masses, and will continue to do so as long as the power grid holds up. Football stats are here to stay.

Like any radical change in the way we think or live, football stats are loved, hated and misunderstood. I’m firmly in the ‘loved’ category, but there’s no doubt the world of football stats can seem dry, pedantic and downright ugly to those who just want to enjoy the beauty and emotion of the game. Even for those who are interested, the sheer volume of numbers can be a put-off. So, since F365 has very graciously granted me the platform to write about something I love, this is the first of several articles with my take on football stats: what they mean, how they can be used and misused, and why they’re worthy of attention.

The most important thing to remember about statistics of any kind is that they’re just bits of information. Like any other bits of information, they need to be interpreted carefully. We need to know exactly what they tell us, and just as importantly, what they don’t. Context is essential. This is particularly true in football, with its very wide variety of playing styles, formations and positions on the pitch.

An example: as of this writing, Curtis Davies and Michael Keane are tied for the lead in blocked shots, with 27. Nicolas Otamendi is way down the list, with only four. Does this mean Davies and Keane are much better shot-blockers?

Maybe, but context must account for a large part of the difference. Davies plays for Hull City and Keane for Burnley, both in systems that have often involved sitting deep and absorbing pressure. Otamendi plays for Manchester City, in a system that is diametrically opposed. When he played for Valencia in 2014-15, he blocked more than twice as many shots per game. If we look at the list of top ten shot-blockers in the Premier League (Davies, Keane, Mee, Dawson, Cook, Djilobodji, Morgan, Gibson, Vertonghen, Prödl) we find that eight of them play for teams that often sit deep. The exceptions are Steve Cook and Jan Vertonghen, whose presence on the list is therefore remarkable.

This exercise not only shows how context is vital for interpreting the numbers, but also how stats can add to our understanding and enjoyment of the game. Having written that paragraph, I’m now going to pay a lot more attention to the way Cook and Vertonghen block shots. Maybe I’ll figure out what they’re doing right. And incidentally, if you go back to the stats and search for shots blocked per 90 minutes, rather than shots blocked total, you find that neither Davies nor Keane but Michael Dawson is in fact the leader, and by a very large margin. This is no surprise at all to anyone who’s followed his career. It’s also a reminder that sometimes the ‘per 90 minutes’ stat tells you what you need to know.

Another example: last season the futuristic superhuman that is N’Golo Kanté registered 5.2 successful tackles per 90 minutes. This year it’s only 3.1. Has he regressed? No, and you only have to watch him play for Chelsea to see he hasn’t. His numbers have dropped because his role has changed. For Leicester he was the primary holding midfielder, playing for a team that ceded possession and sat deep. For Chelsea, he shares holding duties with Nemanja Matic, and although Chelsea do at times drop deep, they keep possession much more than Leicester, 54.0% to 44.8%. If your team has the ball more, there are fewer opportunities to tackle. Knowing this, I want to look at how Kanté and Matic share tackling duties, how Chelsea maintain possession, and how deep they may or may not sit.

Let’s take a slightly more complex case. Ashley Williams’ pass completion percentages for his last five years at Swansea City were 85.3, 86.3, 86.4, 85.4, and 85. Remarkably consistent. This year at Everton, though, there’s a striking drop, to 80.1. Decline? Maybe, but Everton play much more direct than Swansea did, and so Williams may have made more long passes, which are harder to complete. So we look up detailed passing stats, and surprisingly find that in fact he’s making only a few more long passes for Everton than he did for Swansea: 9.7 per 90 minutes, to 9.4 the previous year. But wait – for Everton he’s sending lots of aerial balls to a closely marked central striker, whereas for Swansea he was more likely to send diagonal balls to the feet of less closely marked wide men. I’m guessing that accounts for the difference, but more research is necessary. All sorts of pass maps are out there to help. Even if we don’t want to go that far, we can watch Williams and see how often he passes to head and how often to feet.

A few stats – very few – are ‘pure’, that is, capable of reading without much context beyond the obvious. My personal favorite is aerial duel percentage. Whether you play for Tony Pulis or Jürgen Klopp, an aerial ball is an aerial ball, and you’re either good in the air or you aren’t. It’s no news that among strikers who have played at least half the season’s minutes, Christian Benteke tops the list at 55%. Nor is it a shock to find Virgil van Dijk second in the league among defenders, at 76%. But I was very surprised to see Jan Vertonghen ahead of him, at 78% – just one more thing to check next time I watch Spurs.

At their best, stats are a process of exploration. You get the number, you figure out the context, and you get taken to interesting places. The game is no less wonderful for that: Williams’ accurate diagonal balls are still a pleasure, watching Kanté and Matic share the holding role is fascinating, and there’s nothing like Dawson throwing himself in front of a 100 km/h shot. A powerful header by Benteke or clearance by Van Dijk remains a thrill, too. But through stats we gain in understanding, and for those of us who like that sort of thing, we grow to appreciate and love the game even more.

So that’s what I’ll be up to in the next several weeks. We’ll go by types of statistic, looking at what they tell us and what they don’t, how valuable they are or aren’t, and how best to approach or avoid them. All examples will come from the Premier League, so you can see how the numbers play out with teams and players you know. We’ll start next time with attacking stats for individuals. Hope you find the series worthwhile.

Peter Goldstein

More Related Articles