Some existential stuff from the leading soccer analytics people today.
First, Chris Anderson asks the simple, and in retrospect obvious, question. What exactly is football analytics?
He argues succinctly that analytics is not simply an educated opinion:
This definition of “analytics” as “reasonable decision making” seemed just a step too far to be useful. You mean, if you use your brain to decide on a trade for a player, for instance, you’re doing analytics? That seems to be stretching things to the point of being useless.
But, Anderson goes on, it’s also important to realize that neither is analytics just a pile of spreadsheets:
At the same time, if all you have is lots of data, no matter how extensive and expensive and sophisticated, then you’re not involved in analytics either. You need to do something with them using a transparent methodology and with a particular goal in mind. It’s about the hard work of “discovery” and “communication” of “meaningful” patterns. Otherwise it’s not analytics – then it’s just data hoarding.
Simon Gleave posted a brief response to Ansderson’s post at Scoreboard Journalism, which was careful to point out that football analytics need not exclude qualitative information in coming to a sensible answer to a sporting problem. He also reiterates the view that it’s better to start with the basics before trying to code the entire sport of football:
The other thing that the word analytics suggests is that people are doing something very complex. I would argue that that is not the case in general and it also doesn’t need to be, at least at this stage. The data needs to be understood first and, crucially, the numeric data needs to inform the qualitative analysis via, for example, video which is the current analysis method of choice at both clubs and in the media. There may be a time for complexity but it isn’t now.
The sport of football is only complex if you want to analyse all 22 players on the field throughout an entire match but who is even attempting this? Will it ever be relevant to do so? The fact is that much of what happens on a football pitch is irrelevant to its outcome. That is why it is important to begin from first principles when trying to understand it. What happens on the pitch that IS important? How big a role does luck – good or bad – play? Many similar questions are there to be answered. These are not complex but I rarely see them being addressed.
It’s not certain that anyone in the field has discovered anything concrete in these areas. But we do know that, at least in other “complex team invasion sports” like basketball, this groundwork has been done to some degree.
Several people have highlighted an excellent post by Grantland’s Zach Lowe on the Toronto Raptors’ analytics team’s data visualization project. I won’t quote at length from it because it needs to be read in full (Lowe was the moderator on the SSAC basketball analytics panel). What the Raptors’ analysis team has done is simply extraordinary. But it was not an overnight project. Nor is it likely representative of where most NBA teams are with regard to using X,Y data in an effective way.
This project involves a lot of the elements of which many in the football analytics field are naturally skeptical. Big data. The confident belief that small areas of play can and should be improved (one of the project’s leaders said to Lowe, “A lot of coaches will say how great it is that analytics confirm what they already see. The fact of the matter is, that’s not really true”).
And yet if the quantitative methods are sound, if the basic understanding of ideal player behaviour in a set number of situations is accurate based on sample size, why not shoot for the stars?
At SSAC, some of the UK data analysts audibly sighed at the X,Y Visualization Panel at the moderator’s blithe aside that “Maybe this will have an application for the Premier League one day”. After all, companies like ProZone were the first to develop the X,Y technology that made the Raptors’ program possible, and others have since followed suit.
But there is nothing so different from football and basketball (minus the very wonky relationship between goals and shots) that precludes something similar from not working in soccer, if not for outright goals than maybe clear cut chances.
Part of that though might have to come from the world’s richest football clubs headhunting some of the bright minds working in North American sports, paying them a handsome salary, and leaving them alone for a while to develop something along these lines. The inherent conservatism in European football can’t last forever. And for soccer analysts: skepticism is healthy, but it’s best applied on a case by case basis. The early pioneers in soccer analysis shouldn’t be afraid of going big.