APBRmetrics

[Harold Almonte]

Provoking a corner is at some point better than the ground play, because you put the ball very near the goal with a lot of off. possibilities, (with the exception of goalkeeper's privileges) there are more random consequences, something like an alleyup or to grab an off. rebound and score. But, teams just set the play where they feel an advantage, and yes defense would prefer just to allow the lateral inside shot rather than to concede the corner. It's a continuos adjustment.

[Carlos]

Hey, doesn't anybody here plays soccer? It's not easy to kick the ball into a defender in a way that ensures that the ball will bounce off the defender and out of the field. The two players need to be really, really close to the lines for this to be possible, for starters.

[schtevie]

The fact that the relative ease of increasing the proportion of attacks that gain a corner kick is another emprical issue about which, apparently, little is known is entirely on point.

It is this apparent general dearth of information which gives me confidence that my supposition - that teams would be well-served to try to induce more corner kicks - is correct. By well-served, I do not mean to say that such a strategy would transform a mediocrity into a champion, only that I would expect a discernable positive return.

Without hard numbers and in a game where scoring success rates are so low (never mind the shorter seasons and the more fluid nature of the game) there is every reason to expect that suboptimal strategies and tactics would have a very long life.

Consider relevant NBA analogies that I enjoy revisiting. Starting with the shot-clock era, it took a quarter century for offenses to arrive at approximately efficient outcomes in terms of maintaining possession and culling the worst shots from the mix, despite enormous counterfactual gains. And this (almost) all took place in the context of there being definitive statistics that were readily available and considered!

Can we write that off as a pre-modern abberation? Not so. There is also the quarter century (and counting) adaptation to the three-point shot. Again, this is an instance where the counterfactual gains were as clear as day.

Informed by such experience, why should we expect that soccer, without the benefit of a body of statistics to benefit from (or ignore), would be played in an optimal way? And in this light, there is my supposition about corners.

[Chicago76]

Schtevie,
There is a thread located here that covers a lot of information related to corners:

http://www.bigsoccer.com/forum/showthread.php?t=118462&highlight=corner

A few key points:
1-The conversion rate on corners isn't particularly high in MLS (abt. 2%). Although as they point out, we're not sure of the exact definition of a goal from corner, i.e., rebounds off keepers that fall to the scorer, improper clearances, etc should count.

2-The number of corners taken by a team is actually negatively correlated with winning. Although, as is pointed out in the thread, teams trailing in the final minutes of matches try to "win" corners in the method you have suggested. So it could simply be a case of losing leads to corners rather than vice versa.

3-The opportunity cost must be considered. Winning a corner may be good, but if the alternative is to have the ball in the penalty area of the other team near the goal line, the conversion rate may actually be lower for corners. The corner taking team has the benefit of executing a set piece, however, the defense also has the benefit of insuring greater numbers in the penalty area. Teams scoring during the run of play normally do so with small numbers of attackers beating small numbers of defenders. It's easier to win a 3 v 3 battle than it is an 8 v 8 battle. 8 v 8 battles permit less space and provide more opportunity for defensive help.

[schtevie]

Chicago76, thanks for the reference. The statistics presented are tantalizing, but unfortunately incomplete. A few thoughts:

(1) I wonder to what extent the completion rate of corners is a function of the data being from the MLS circa 2004. To the extent that set pieces require greater skill and to the extent that our domestic league allegedly had/has less of it that the majors of other countries, are these numbers similar, say, to the premiership? Then again, in such leagues perhaps defensive skill is also greater.

(2) I too wish that the definition of a corner had been stated. To my mind, the most reasonable one should include within its purview the immediate aftermath of the kick. Thus, if the corner goes into the box, is headed out, but in the scramble thereafter a goal is scored, is that success attributed to the corner, or not?

(3) The fact that success in MLS 2004 is "only" 1.5% does not imply that on the margins it is not optimal to try to attempt more of them (nor that there isn't an opportunity to invest in players that might yield a greater likelihood of success - note the full range of 4%). For example, when an offensive player has the ball along the endline AND the opposing defense is completely set, is there a greater than 1.5% chance of scoring by taking a different tack? More generally, when the ball is "near" the opponents goal with no man advantage, what is the average success of scoring in these situations? Given that on average only 1.3 goals were scored per game in MLS in 2004, my guess is that this rate was lower than 1.5%. Maybe not.

Some more data is required to address these issues, but it seems to me not that much more, it wouldn't be terribly expensive to acquire.