I am admittedly biased toward analytics pieces that involve some form of media criticism, so I was intrigued this past week by Ted Knutson’s Statsbomb post on the received wisdom—repeated ad nauseam by almost every English colour commentator in the game—that teams should always post defenders on the posts in corners.
Knutson argues that putting defenders on the posts is like an extremely expensive insurance premium for a fairly low probability event. You’re taking two guys out of the defensive mix and, presumably, you make them less able to help facilitate an effective counter.
But what irks him most is that no one in football media is willing to take the time to question why so many clubs today don’t do it, if it is so obviously effective:
So yeah, when you ask questions about how someone “knows” a thing, and maybe question the validity of that knowledge, you can cause problems. But the fact of the matter is, we should be doing this constantly inside of clubs because it leads to valuable research that can change behavior and develops more effective styles of play.
A goal in the Premier League is worth something like £2M. How many of those do we leave on the table because someone’s knowledge is outdated or just plain wrong?
I am, at the moment, still working my way through Michael Lewis’ book The Undoing Project, on the pioneering work of behavioural psychologists Daniel Kahneman and Amos Tversky. It’s a kind of more narrative driven summary of the ideas that you should almost certainly read first in Kahneman’s popular masterpiece, Thinking: Fast & Slow, but there are a few highlights, and one of them is the story of the development of Prospect Theory.
The theory essentially holds that we tend to overestimate low probability events and underestimate high probability events in making choices between situations that involve an economic loss or a gain (there is a very simple summary of the theory here). Moreover, we don’t view potential gains the same way we view potential losses; we would rather not lose $20 from our wallet than find a random $20 in the street.
So with that in mind, here is an entirely untested theory I have on why many coaches would rather post two guys on the posts during corners.
Let’s say the corner comes when the score is 0-0; we all know this is a vital moment in a football match. Goals are rare in football, and conceding a goal will entail your team must score two to have the greater likelihood of earning all 3 points. So conceding a goal at this point carries a strong negative psychological burden, particularly when it comes to set pieces in your defensive final third.
I’m also willing to bet that when most of us visualise a goal scored from a corner, we see in our minds a towering central defender or striker rising majestically in the air to head the ball into the top right or left hand corner. Something like this maybe:
It’s natural that, when we see this kind of goal, we think “If they would have only put defenders on the posts, they could have headed those chances away.” When people think about defending against corners, it’s this kind of thing they’re worried about most.
What those coaches and commentators can’t see, at least without the benefit of data, is the actual frequency of these towering free headers, and whether that frequency goes up or down if you take the two guys standing around at the posts and involve them in a more active defensive set up. Instead, they have a very powerful mental image of a conceded goal from a set piece, along with all the bad stuff that comes with it.
And so, perhaps out of fear and familiarity, coaches not only vastly overestimate the likelihood of these kinds of chances occurring, and in doing so leave themselves more open to less obvious scoring chances, perhaps from a second or third ball, than they would if they had more active defenders.
I want to use this example to make maybe one of the most important points about soccer analytics—sports analytics, even—that I can, especially with all the near constant talk about the importance of communication between analysts and coaches/players/media.
?also applies to all that expensive tech you might want to buy. https://t.co/sKO6UB63Lo
— Rob Carroll (@thevideoanalyst) March 22, 2017
Stats people will know that producing a study of the effect of using defenders on posts for corners is not difficult, relatively speaking. What is difficult is convincing a risk/loss averse coach or commentator that they should ignore their instincts and trust a pile of data. These are strong, predictable psychological biases that are difficult to simply waive away in the face of even the strongest evidence.
This is why, in my opinion, the job of analyst is not merely to be a good scientist, or a good designer of compelling visualizations. They also have to be good lawyers.
Good lawyers know that it’s not not enough to simply present the facts of the case to convince a judge and jury; they need to make their argument with persuasive rhetoric, repeated emphasis on key points, and a strong, plausible narrative (this also explain why some of the more successful sports analysts I know are former lawyers).
If I were to make the case for taking men off the posts for example, I would want to work closely with my performance analyst to come up with a series of videos, convincing instances where teams that posted players on the posts conceded sloppy goals, or failed to initiate an effective counter. I would want to package my case as part of an effective, top to bottom powerpoint presentation.
And, in making my case, I would want to work with prospect theory, not against it. You have to convince your coach not only that posting guys on corners doesn’t prevent as many goals as you think, but that posting guys on the corner is the riskiest defensive maneuver on the books.
Sure, your dataset will be the foundation of that argument. But it doesn’t speak for itself (and neither does a linear regression or pretty viz). You need to make a case.
On a happier note, not everyone who works in English football media is as obtuse as our imaginary co-commentator. This, by Jonathan Liew, is one of the best examples I have seen recently of a journalist making effective use of data to make a very compelling point about Tony Pulis’ alarming habit of shutting up shop at 40 points. Well worth your time. We need more Liews and Ingles plugging away.