the square ball week: pure numbersBack
Part of loving football is being curious about what other people love about it; curious, and judgemental.
We all love the same game, but the gap of understanding between, say, a Leeds United fan and a Manchester United fan is the gap between appreciating Lucas Radebe or Jaap Stam. How can any sane, rational follower of association football consciously resist the charms of Radebe in favour of the bullet sternness of Stam, just because he wore a red shirt rather than a white one?
It doesn’t make sense, and much of what our fellow travellers in soccer do doesn’t make sense. From the songsheets and barbecues of MLS, to the choreographed pyros of middle and eastern Europe, to whatever is going on with the fans watching newly minted gazillionaires dropping into their provincial sides in China, we look at each other askance with a mixture of curiousity and disdain. Those Icelandic thunderchants were pretty cool. We’d never do them here. Orchestrate your ultras with a drum? Fine in Italy, but percussion is a scourge over here. Barbecued hotdogs? Only if Massimo is buying.
Zoologists have studied this along tribal lines, and found that the tribes don’t align, as they might have expected, along the same lines as the teams fans support. There is no mode of thinking of ‘The Leeds fan.’ In fact there are probably few less unified centres of thought than Elland Road. Heroes here are lambasted; Marko Viduka and Jermaine Beckford stand out in the modern era as players who could have reasonably turned to half the stadium and asked, what more, exactly, do you want?
Which is an interesting question. What do fans want out of football? To see their team win is the obvious answer, but even that splits into sub-sects and splinters. To win with style, or to win like Warnock? Victory at all costs, or do goals only count in large amounts? Is one goal enough if the football was beautiful? And who defines what is beautiful? (Luciano Becchio does, but that’s a whole other conversation.)
Trying to understand what other people love about the game I love is something I try to do, because I’m much in favour of loving things more, and if I can understand what other people get from the way they appreciate football, perhaps it will widen my loving net.
Which is why I look at today’s wave of stat-obsessed fans and sigh deeply at my own ignorance, wondering what they’re seeing that I’m not; and why I keep looking, seeking to share their appreciation. These good people aren’t like the stattos of old, who could reel off the scoring record of any designated player from the 1950s, or tell you the reserve team debut of the most obscure player you can think of. Today’s football statistician is mathematical of mind and data obsessed, producing charts and visualisations that show — well, what?
Plenty, if you can get your head round it. The player performance radars produced by Ted Knutson on Twitter fascinate me; [this is Robbie Keane’s](https://twitter.com/mixedknuts/status/768928847739756544), at first glance showing the decline of an aging striker, but when you look closer, revealing the development of an increasingly conservative but clinical style of play. Robbie Keane in 2016 had far fewer touches of the ball in the penalty area, and far fewer shots, than the 2014 version; but his expected goals per shot was much increased by 2016, suggesting that at 36 Keane had a lifelong striker’s eye for getting on good chances.
Missing from those radars, though, are actual goals scored. ‘Expected goals’, there instead, are like the entry-level system for those of us curious about new football stats, and encapsulate both the insight and frustration I find in these methods. xG assesses the quality of a chance, and whether a striker would be expected to score it: as [Michael Caley writes](http://cartilagefreecaptain.sbnation.com/2015/4/10/8381071/football-statistics-expected-goals-michael-caley-deadspin):
“Where was the shot attempted from? What sort of pass assisted the shot? With what body part was the shot taken? Did the attacker dribble past his defender before trying the shot? How fast was the attacking move that led to the shot? Was the shot off a rebound or from a set play? All of these factors clearly influence the likelihood of scoring a goal. By aggregating this information into a model, I can estimate the likelihood of scoring different shooting chances in a match or over a season.”
That has implications for looking at football on an individual and team level. If a striker is getting high xG ratings, but not scoring many goals, he’s missing a lot of easy chances and needs banjoing. A low xG but lots of goals is the mark of a striker with a taste for a blockbuster half-chance. High team xGs indicate a side with effective playmakers, who should be racking up the assists; while defensively, if a side is constantly giving its opponents a high xG, it probably has Giuseppe Bellusci in it.
This is truly interesting stuff for what it reveals. A sequence of 0-0 draws might indicate a drab, George Graham inspired side, but if they come with high tallies of expected goals then something else is going on; either a Steve Morison inspired forward line and a Nigel Martyn in goal, or an anti-goal hex afflicting your stadium.
A curse wouldn’t fit the model, though, and that’s where my view of football veers from the statistical. What is the mathematical factor of romance? For me, it’s high, introducing a high level of subjectivity that would make a statistician shudder. But when I look back at, say, the Howard Wilkinson era at Leeds, I can’t help but be drawn towards the Sheffield factor as a romantic influence upon games.
Wilko drew heavily on his Wednesday old boys for his Leeds title teams, and that meant the matches with Sheffield United had an extra edge; and if you don’t believe me, find a video of Carl Shutt after he scores the winner against the Blades at Elland Road in 1991, running to the South Stand to celebrate with the Leeds fans then turning to point at the visiting supporters, as big an up yours as Mel Sterland’s opening freekick had been. Remember, too, that wasn’t Sterland’s only freekick against Sheffield United; he hated them, and loved scoring against them for Leeds, and didn’t care who knew it.
There’s something subtly more lovely, to me, about a goal like Shutt’s. It was presented to him through a lousy backpass that left him one-on-one with Simon Tracey in the Blades goal, with a twenty yard headstart on their nearest defender; probably a high xG if ever there was one. But how much more romantic to imagine the way Shutt felt, with a gift of a chance against the team he hates, and how to account for the mixture of nerves and skill required to round the keeper and score? That’s before we even think about Shutt in the Nou Camp. And if you haven’t for a while, just go and think about Shutt in the Nou Camp. You won’t regret it.
That is, to me, where the beauty is often found in football; in the longer arc of a story that takes a player from Hillsborough to Elland Road to the Nou Camp, via Ashton Gate, culminating in a tearful moment and a goal he’ll never forget. Oh, and it was the day before his birthday, too.
Such things are not mutually exclusive from stats. Expected goals ratings are, ultimately, subjective anyway; there are competing models for how to calculate the ‘quality’ of a chance, of varying mathematical complexity, and in a way xG ratings too reflect the personality of the rater. Who knows that there isn’t a cheeky statistician out there tweaking his xG model to give more value for ex-Wednesday players in the Sheffield United penalty area? Part of the fun for the people doing this work, I imagine, is testing their models against others, refining the maths to find something that represents a game we’ve seen better than the final scoreline or a match report, that can even predict future scorelines with the kind of low margin for error that could make a brave gambler very rich. There’s something charmingly quixotic about that side that appeals to me.
There’s also something wonderful about living in an era when, within a couple of hours of the end of Leeds United vs Reading, I could go on the internet and find out that Ali Al Habsi had played fifty passes in the game, and see them drawn on a chart. Heatmaps get a lot of heat, but there’s something great about seeing that Charlie Taylor played most of the game up around the halfway line and thinking about how that tallies with the game I saw, the defensive line Leeds were holding.
Which is perhaps the crucial point of all this: that it tallies with the game I saw, and the two together add up to an understanding. “You can tell something about how they are as human beings, to be fair,” Jaap Stam said about the Leeds fans who jeered and laughed their way through his side’s pass-map obsessed performance on Tuesday night. “It’s a way of thinking. Because if you think that this is boring I think they need to have a look at their own team and how they play.
“They can say ‘we won’ and of course at the end it’s about the result – but if you need to play like that then I don’t want to be a manager. Because I don’t want to play and just wait and wait and wait. That’s not my type of playing, my type of tactics.
“You can say congratulations to Garry Monk, he got the result and everyone wants results. But in terms of way of playing, it’s not my cup of tea.”
Stam’s is the kind of talk that gives stat fans a bad name; the view that hails xG over G, pass percentage over final score. The final match stats will record that Reading dominated the ball and were secure in possession; they had 90% pass completion, which is extraordinary in isolation.
What the spreadsheets don’t record, though, is that the 634.5 passes Reading completed were completed to a background of uproarious laughter and sarcastic olés; that Stam’s side were seen straight through by the Elland Road crowd for the anti-football they were attempting; that there was incredulous outcry as Reading stuck rigidly to whatever it was they were trying to achieve. The emperor was naked, and the emperor was Jaap Stam, a very ugly man.
You’ll still find Reading fans willing to defend their side’s game, though, either through blind loyalty to the Royals, or blind faith in the stats: their centre-halves completed more passes than the entire Leeds team, which apparently is a good thing. Which is a danger inherent in relying on stats to tell the story of a game, when there’s so much about football, that numbers can’t tell you, that’s so important.
Or maybe the numbers can tell you, and the stats guys just haven’t found the model for it yet. That must be what motivates them; the perfect visualisation. If someone ever does crack that, I’ll be ready and willing to welcome it, because football’s a broad church of people doing weird seeming things, but we’re all inspired by the same love. People kicking a ball: it’s ace. Unless you do it too much, Jaap; or maybe that’s just something I don’t understand either.