162802157

The subtitle of Chris Anderson and David Sally’s new book The Numbers Game, which following the inevitable colon that must be attached to all non-fiction titles, is Everything You know About Football Is Wrong. This is obviously a selling tactic.

“Oi, Maggie, this book says everything I know about football is wrong.”

“Does it Frank?”

“Yeah, it does. It can’t be anything more than a game of 11-a-side trying to put a ball in a net, could it?”

“Well, buy it and find out. We have a gift certificate.”

But I kind of wish they’d gone with Everything You Know About Football is Right.

Let me explain.

Yesterday I watched a Tweeter react to the initial excerpt release of the book in the Times yesterday. His basic point made over several 140-character posts was what nothing in the excerpt was particularly revelatory. Of course teams that score a lot of goals and don’t concede a lot of goals do well. Of course possession is important, and of course it matters that team don’t turnover the ball too much. Sure, it’s kind of interesting that corners are by and large a waste of time as a set-piece, but the banal truths on display here is comprehensive proof advanced statistics is a waste of time (this by the way is the corollary to the argument over the alternate approach to football stats, which argues the game is far too complex to analyze and therefore any attempt to learn from it via statistical analysis is also—surprise!—a waste of time).

Perhaps part of the problem was the sense of expectation foisted on readers by the copy-editor. The introduction boldly states that the book presents “a sea change not just in what we think we know about the game, but — as shown here, in this exclusive first extract from the book — how we think we should play it.”

No doubt there will be some eyebrow-raising statistics in here, and some major challenges to our perception of the game. But the key phrase in this excerpt is this:

“We are not concerned with theory. We are concerned with facts.”

In university I was fascinated by the writings of medieval Catholic philosopher Thomas Aquinas, more for the form than the content. His Summa Theologica was written in a set, repeated format. He poses a question, presents counter-arguments, than makes his argument and addresses the objections in turn. Once he’s done, he moves on as if the argument is settled, and builds upon its implications, constructing his castle in the sky brick by brick.

Of course it’s a massively flawed approach that flatters Aquinas’ faculty of reason to the point of absurdity. That’s because when it comes to matters of fact, naked reason is no substitute for empiricism. And that’s where Aquinas’ approach is instructive.

Despite the old joke about lies, damned lies and statistics, the numbers themselves presented by Sally and Anderson, and others doing interesting work in the football analytics field at the moment, simply are what they are. They are not a journalistic cliche, they’re of no bias or clique, they don’t have an agenda. How they’re interpreted can sometimes be a subject of debate, and that’s certainly where the fun lies. But in of themselves, if the method is sound, the numbers are as close to the fact of the matter in football as one can go. At the moment, they point to some broad truths that seem so obvious in retrospect that one wonders why anyone went to the trouble of finding them.

But the key point is that we need to first establish these seemingly obvious truths as fact and not mere conventional opinion if we are to make any significant progress in the field. This is the core of Simon Gleave’s point that I seem to bang out almost every week—say it with me now!—we need to walk before we can run. Some of these truths are obvious, and some are definitely not (the idea that football’s fundamentals transcend tactical preference is a potentially huge shift in our understanding of their effectiveness: while the means may differ, the end result is the same). But they still need settling before we can properly move on.

I don’t want to keep hammering this, but the best example of what I mean can be found this season in the move from looking at raw Total Shots Ratios to analyzing TSR at 0 game states. One small step for an analyst, potentially one giant leap in understanding how to empirically measure what makes a good team ‘good.’ These little steps are made day by day, week by week, month by month, until they become convention. There’s no guarantee this will all lead to a “eureka” moment in football analytics, but a document like Anderson and Sally’s book could potentially go a long way to popularizing this approach to soccer statistics, and gain more converts to a growing discipline.

Comments (1)

  1. “Of course possession is important”

    Is it? The top sides in the Prem play a style of possession football and also produce 40% of that league’s wins, So their characteristics are going to dominate over a simple regression correlation. Yet they also win the possession battle when they infrequently lose a match. The rest of the league are as likely to win a match and still lose the possession battle. Possession more often simply defines a style of play,it isn’t a universal.Take the best handful of teams out of the sample and Europe wide possession doesn’t correlate with outcome.

    “Sure, it’s kind of interesting that corners are by and large a waste of time”

    Based on a limited amount of corners from a third of a season using StatDNA definitions they may appear so. But use Opta definitions over a season and a side on average scores from a corner once every 5 games,not the10 quoted in the extract. A third of matches see at least one goal scored from a corner and for small sample sized irrelevance,the last two major cup finals have been decided by last minute goals from a corner. Corners also matter greatly as a percentage of the total goals scored by certain,usually less gifted teams.Hardly worthy of the “waste of time” soundbite.

    The timings study of when to make a substitution to the exact minute to maximize your chances of winning from a losing position also quoted in The Times extract smacks of overfitting your data.

    It is great that analytics is growing and I’m sure the book will be a stimulating read,but getting into print shouldn’t be taken as an excuse to automatically endorse mere opinion as “broad truths”

    Waldron

Leave a Reply

Your email address will not be published. Required fields are marked *