Won’t be a full column this week, but I wanted to point readers to Opta Pro’s interview with Soccernomics co-author Simon Kuper.
I’ve written in the past that the perception of Soccernomics as ‘a book on soccer analytics’ has had unfortunate consequences for the popular understanding of statistics in football, reducing it to a set of curious “freakonomics” style tidbits that have little to do with how teams play but instead how many fans kill themselves during tournaments, or why England doesn’t win World Cups.
But Kuper touches on something I’ve heard before when speaking to people in the analytics field, and it’s something that has been backed up anecdotally in my experience, both by look at the Opta MLS chalkboards and in watching the game:
However, my sense is that the data shows – and in a few years’ time we’ll probably have a much greater understanding of this – is that crosses are not an efficient way to score goals. It’s therefore not an effective strategy and I’m sure this will become clearer as we understand the data better.
This is something Sarah Rudd, vice president of analytics and software development at StatDNA, touched on when I spoke to her several months ago, and the prevalence of crossing in football is one of the few (if only) areas where the data conflicts in a concrete way with how football is played. This year for example, Duncan Fletcher wrote a piece that argued in part Toronto FC’s relative turnaround under Paul Mariner could in part be ascribed to the drop in the number of crosses.
This should make make intuitive sense to anyone who’s watched their fair share of football matches. After all, there is nothing more infuriating than watching a team carefully and intricately pass the ball from their own half to the flanks, with full-backs running the length of the field to help out in attack, and the forwards moving up with the midfield behind in support, only for a player to whip in a cross in the blind hope one of their players will connect with the ball and score.
It’s why we see often see cross conversion rates regularly as low as 1 or 2 connections for every 30 attempts. It’s a lottery with poor returns, in other words. For a cross to pay off, you need a centre-forward who has both height advantage and good positional sense. You need the central defenders to mistime their jump, or lose their marker. You need a pinpoint cross that is low, fast, and just far enough from the keeper to prevent him from catching the ball.
When everything goes your way, your team is awarded a goal. Unfortunately, more often than not, the end result of a cross is loss of possession, vulnerability on the break, and exhaustion as your players must run in the other direction to defend.
Frequent crossing of the ball is in many ways a holdover from the old Charles Reep long-ball school, which (erroneously) promoted the 3-pass optimization rule, the idea that the vast majority of goals came from 3 or fewer passes. Hence, better to whip in the ball as quickly as possible to the area, rather than tease the ball along the edges of the 18-yard-box.
But as formations narrow, and variations on the difficult-but-rewarding variations on possession football remain in vogue, we may see less and less crossing at the highest levels of the sport. Spain for example attempted a mere 13 crosses in their 4-0 defeat of Italy in the Euro 2012 final, and half of them were short and within the 18-yard box. Throughout the tournament, most of their crosses from outside the area came from corners, and they rarely exceed 20 attempts.
Spain plays a 4-3-3 with a “false 9,” meaning the wingers often cut in and pass to a forward coming in from deep. But not everyone can play like Spain or Barcelona; this doesn’t mean however they must whip the ball in the box in blind hope for an Andy Carroll/Darren Mattocks-style towering header. As more and more data on this subject becomes available, we may see a drop off in crosses around Europe.
Crosses are a VERY efficient way to score goals. All it takes is one pass.
no……it takes 30 on average. it might be the first one of thirty sure, but stats is stats
By that logic, having your goal tender shoot from his own penalty area is the most efficient way to score goals (zero passes needed!). i don’t think you understand the context in which “efficiency” is generally applied.
And a good header is one of the prettiest goals you can see.
So you’re saying Liverpool’s Downing to Carroll strategy was doomed from day one? Ah well, 55 million wasted. No biggie. Makes me miss Rafa Benitez’s transfer policy.
its simply cause theres a lack of good crossers. beckham wasnt that good at playing, but he was amazing at crosses. now players strive for the opposite.
Crosses are extremely inefficient when you consider the number of crosses (on average) before a team will score. even if it’s as high a 10%, that’s still 9 times out of 10 you give the ball to the opposition.
They’re popular often because they are so basic. Chuck it in the mixer, give it to the big man. It’s no surprise that they seems to be used more when there is high urgency, especially by teams that don’t have much in the way of creativity (Yes, England, I’m looking at you)
Still one of the most frustrating things in the world is when a player gets into a good position and fails to put in even a vaguely threatening cross – either by hitting the first man, or by sailing it out of play. I’d like to see more driven crosses, too; hit low and hard behind the defence to create a bit of uncertainty.
Crosses also leave a wide midfielder and sometimes overlapping defender out of position defensively – a goalie can snag a cross and launch a very dangerous counterattack before the team can regroup.
I think there’s a lot of information behind those stats that can explain some of the lack of potency.
As people have said, there are a lot of inefficient crosssers of the ball out there. Some teams just seem to want to force the issue trying to put the ball into the mixer and hope for the best.
I would leave it as: crossing is an effective way to score goals, however, continually punching the ball into the box in a blind and predictable manner is not.
Hm. Do we have data on the success rate of attacks through the middle and/or on the ground? WIthout that, saying crosses have a low success rate is really just saying ‘scoring is hard’. The counterargument is that each pass and dribble is a potential for failure when the pass is poor or is cut out by good defense or the dribbler is tackled, so the longball strategy of limiting the number of actions you take (and therefore the number of potential failures) is long-term efficient, especially if your players aren’t skilled enough to win duels at a high rate.
But the “counterargument” is poor, simply because there are discrete actions in the game that come with a lower likelihood of losing possession, and smart passes and short dribbles are among them, as compared to whipped in crosses. One of them is a shorter passing option, or simply passing to a safer option in order to retain possession and then probe for an opening.
I think one of the biggest holes in soccer analytics is the lack of distinction between contested or difficult passes and easy passes. It’s the gap that makes possession stats (or pass density) effectively useless as a metric, because tap around in your defensive end have basically 0 correlation to goals scored (though they may correlate to fewer goals allowed, since the other team can’t score if they don’t have the ball).
Which is to say that it’s entirely possible that the passes that are actually goal dangerous (through balls, short passes in the opposing box, etc) are just as low-probability as crosses, but that’s obscured by the fact that passing data is dominated by other kinds of passes.
Except they DON’T have 0 correlation to goals scored, because teams in possession score goals.
But moving to your argument here, I think you’re complaining about a metric that doesn’t exist. The issue here is that short passes are, as a general rule (not across all possible worlds), safer than long, searching balls, particularly ones whipped in the 18-yard-box.
As I’ve written in previous columns in this space, information from analytics are generally sign posts, and one of them, in this case, points to the ineffectiveness of whipping in crosses come what may in wide areas. Which a lot of coaches and managers and fans are already aware of.
The threat of the cross is vital though. Those through balls in the middle work well because defenses must be designed to cover the entire width of the pitch.
Without that threat of a cross – you have the situation where defenders can narrow, allowing only wide MFs to take the ‘crossing’ areas, putting more defenders into the channels, preventing the scoring opportunities.
Crosses may be an inefficient way to score for a number of teams, but in certain circumstances – where you have an accurate crosser (Beckham?) and favorable/match-ups people in the box – they can be deadly.
The game oscillates to some degree between extremes of short & long passing over time, right now the pendulum swings Spain’s way, as the game morphs we may see crosses come back into vogue.
Ironically one of the trends of Euro 2012 was the high amount of headed goals and here-
http://www.guardian.co.uk/sport/blog/2012/jun/26/euro-2012-the-question-headed-goals
Jonathan Wilson argues that the reason was partly the extra official behind the goal that reduced the number of fouls but also because the ball was of a design that made good crosses easier to hit.
Crosses aren’t innately good or bad (unless you want to argue from a theological point of view) but how and when they are used is certainly open to scrutiny.
Thanks sidereal! Until you do some comparartive analysis, it doesn’t mean much. Scoring is hard from every approach.
One of the biggest headaches soccer analytics suffers is that they spend a lot of time coming up with statistics that try to make the flipping’ obvious sound like some great new discovery. We’ve just been told that the longer the pass, the harder it is to get it right. This is news?
If the players in question aren’t good enough to put in proper crosses, they sure as hell aren’t good enough to make the “smart passes and short dribbles” that maintain possession in the attacking third. You can tell him to cross from the byline, carve up a back line with Xavi-style passes or tell him to go Messi-dribble his way to the net, all you want, but Theo Walcott is still Theo Walcott and will still manage to fuck it up.
Once teams without Spain’s technical ability try to play that style of game, Kuper and the analytics geeks will have new stats that say trying to make short passes and dribbling around the top of the box is an inefficient way to score goals. After all that bean counting and number crunching, we’ve ended up with the conclusion that good players are better than shit players.
Sometimes you just have to put the calculator down and play the game, FFS.
SB
I think you’ll notice that the use of analytics in this case need not have contradicted intuition. The issue here involves teams for whom crossing is an active strategy, as in players in either flank choosing to cross rather than pass and maintain possession until a more clear cut opening appears.
Richard – your point here is well taken. HOWEVER…
That isn’t what is stated by the data Kuper is using. The data does not tell us what the options were (or were not), or the style of the teams crossing the ball.
Analytics can be VERY powerful. However, keep in mind that statistics can be used and cherry-picked much like quotes to prove any point or view. You could pick a team whose strategy is to cross the ball – and provided you pick a team that possesses the right criteria (excellent crossers, tall CFs, favourable match-ups, etc) prove that crosses are a VERY efficient way to score.
Its like the PK problem – analytics tell us the best place to aim is in the middle – which it is – until a majority begin to do it then in loses its advantage.
Its a bit of Game Theory (with Evolutionary Stable Strategies – ESS) that the data-heads are leaving out.
I personally agree with the analytics, however there is a huge caveat.
Crosses may be a poor way to score goals, but if you don’t have to talent to keep possession ad score by going up the middle its still a good option. You need a winger with a decent foot, and a tall centre forward. to score up the middle requires more skilled players to make it work, so it can’t be an option, in the near term at least, for ALL football clubs.