Believe it or not, I had this idea before Stan Collymore started his own brand in ‘goal scored by a cross’ alerts. I was wondering whether the stats community’s general disdain for crosses, built up as a reaction against the ‘chuck it in the mixer’ cliché, was a little unfair. That doesn’t mean ‘we’ think you should never cross, just that there can often be better options and maybe, y’know, we should be thinking about what we’re doing a bit on the ball.
Collymore in his tweets on the subject mentioned things like cut-backs, which stats guys (I use the term as a loose generalisation) like, but I particularly wanted to look at what I called ‘traditional crosses’. These would be balls into the box from wide areas, the kind of thing that generally comes to mind when people think of crosses.
The way I see it, crosses are statistically ineffective because they can often the result of frustration or boredom at not breaking down a defence. Whereas with most passes you weigh up the possibility of completing it, ‘traditional crosses’ are often basically thoughtless, purely hopeful that someone will get on the end of it.
Originally I wanted to look at positions where strikers and defenders were when the cross came in, but I soon realised that would be very difficult to do, particularly in a decently sized sample, so in the end I just marked down how many (outfield) players on each team were in the box when the cross was hit. There are some more technical details on my criteria below the results.
|Difference in numbers in the box between crossers’ team + opponents|
(Count% = % of all crosses for each difference; %X = % of all crosses for that difference that became a touch/shot/goal; cnv = how many shots in that difference were scored)
The most crucial things are the %shot and %goal rows, that’s to say the amount of crosses that turned into a shot and goal. These percentages are slightly skewed by the fact that all the crosses I noted had to beat the first man (for reasons explained in the technical section) – if you included blocked crosses then these figures would be even lower.
Now, I only noted just over 500 crosses in total (it’s a very tedious process) so at the extremes, times where there are 4 or 5 fewer attackers than defenders, or more attackers than defenders, are few on the ground. This means that discrepancies can occur which probably wouldn’t if I’d looked at 5000 instead of 500.
For example, at -4 (four fewer attackers than defenders), one Jamie Vardy shot was enough amidst only 7 examples to make the percentage of crosses turned into shots go above the amounts for both -3 and -2. There were also no goals in the 7 instances of attackers outnumbering defenders, however the comparatively huge percentage that turned into shots suggests that, over many more occasions, this percentage would increase.
Clearly, having more guys in the box compared to the number of defenders helps, and this is probably something everyone would have guessed. However, now there’s actual figures for it, and it could (hypothetically) be used to help make players think more about when they’re crossing the ball.
If you can see out of the corner of your eye that your striker’s outnumbered by the opponents’ entire back four there is only a VERY slim chance your cross will even result in a shot. If you’re a team that likes, for whatever reason, to cross on the other hand, maybe consider putting some extra bodies in the box.
Of course, it will also depend a lot on the quality and the cross and the quality of the players in the box. Over the course of this (some 40 odd matches), I’ve seen some truly abysmal crosses and it is definitely an area that can be improved upon.
The slightly, but not very, more boring, but also quite important stuff:
– I didn’t use set pieces in this, as these are different kinds of situations where both sides have a lot of time (compared to open-play situations) to prepare and organise themselves, and generally there’s a hell of a lot of bodies in the box, which there aren’t for open play crosses. However, I included instances where the original set piece had been headed back out, or passed (as opposed to nudged a yard) from a corner for a different angle, as I consider these to essentially be a new ‘phase’ of open play.
– There was some fluidity in how I counted the number of defenders and attackers in the box. Sometimes there would be players in a near corner of the area who simply cannot be the target of a cross – if the ball would go to them it would be a five yard pass instead. Therefore, there will be some instances where I might have classed something as -2 where someone else might have thought it was -3, but I figured these would equal themselves out largely, and this isn’t exactly a rigorous academic study. If it was I’d have looked at more than 40 games and 500 crosses.
– They had to be from wide areas, as opposed to half-spaces (if the term half-spaces is new to you, it’s essentially what the English know as the channels – there’s an article here on them, the third diagram being the most relevant one). This was partly to stick to my original idea about ‘traditional crosses’, and also to help me classify what counted as one.
– They had to get past the first man cleanly, meaning they couldn’t be blocked or deflected. This was because I was primarily looking at how the number of players in the box to aim at affected success, and included blocked or deflected crosses would affect this. Looking at how many crosses even got past the first man cleanly is another key aspect of pass selection relating to crosses, and another area for people to study. For similar reasons, they also wouldn’t be counted if they went out of play before going into the box, which a couple I saw did.
– The games I did these for came from over several European leagues and the Champions League, but primarily the Premier League. I figured as I wasn’t looking at quality of the crosses themselves per se, but rather a general average (which is probably another issue) team or league style wouldn’t affect this too much.
Further observations from this as a reward for you if you’ve got this far:
– Certain teams try very hard to play into the half-spaces, and I think sometimes this is a little to their detriment. Generally ‘top’ teams like Arsenal, Man City, Spurs, once the ball is in a wide space players will drive or cut inside, and rarely cross. I think if you’re due to play them and you notice this then it may help you defend against them, as your defence can be more orientated to defending against these half-space areas instead of being worried about having to deal with a bunch of crosses. I imagine this is something akin to what Leicester have been doing defensively, as it’s been highlighted that they tend to let teams have the ball wide, probably because they know these are fairly un-dangerous areas.
– Some teams look to be taking advantage of the kind of principle I’ve highlighted, namely that if you’re going to cross the ball it should be with players in the box. Liverpool are one that I’ve noticed flood the box, and West Ham have done it too. West Ham’s crosses, quality and consistency wise, also looked better than the average team’s judged by the 3 or 4 games of theirs I watched.