Wojciech Szczesny had an almost unique job as keeper for Arsenal in the 2011-2012 season. As part of a possession based team, he was expected to play an integral role passing the ball both short and long at a high percentage. And since the opposition decided that Arsenal were vulnerable on set pieces they attempted to exploit that perceived weakness with an aggressive aerial game forcing Szczesny to be commanding in his area. As if that weren’t enough, the Arsenal system last season also featured a hawkish attacking style which often left a lot of space between keeper and last defender and thus required Szczesny to come out and claim the ball in an almost sweeper-like role. When he was unable to claim, Szczesny was forced into a large number of what Opta calls “big chances” for the attacking player; one-on-ones and other scoring opportunities for which you think the attacking player ought to score.*
Given all these duties, Szczesny excelled at some and struggled with others as Arsenal allowed 49 League goals in 2011-2012, the most ever in the Wenger era. So, which areas did Szczesny do well in and where would we expect improvement this season? Remembering of course, that it’s a team sport, in part 1 we will look at what he and his teammates did well and tomorrow in part 2 we will look at where he and his teammates will look to improve this season.
A word of caution that no one will read: this is one season’s data and while it is an enormous amount of data gathered over that one season Arsenal already look as if they have already changed some of the fundamental problems from last year. Not only that, but the Opta data set I was given is missing some key information (especially about these “big chances” such as “big chances saved” and “big chances saved in the box”, etc.) which could shed further light on where Szczesny placed on an individual basis. And finally, this is a team sport but unfortunately it’s often the defenders (and especially keepers) who get the blame for goals allowed. Is it Szczesny’s fault for every goal? No. And maybe even more radically, is Szczesny a terrible keeper because, say, his saves stats are low? NO.
As for the methodology, I was given most of Opta’s dataset for the 2011-2012 season. Each player, each game, and every one of more than 200 events is logged by Opta, which I can comfortably say is more data than you can shake a stick at. In fact, it would take 100 kids shaking 100 sticks at 60 shakes per minute 17 hours to shake a stick at all this data.
This is the biggest data set I have ever played with and more than once, I lost my work and had to start over. That is why today’s post is so late and why this post has taken me so long to write overall. What’s even more frightening is that this data set is just a slice of the data that Opta are recording. For example, for every event in the data set I have, there’s position data which I don’t have: where, exactly, on the field events take place. Which is just one of many other events that I am confident Opta collect. Given the enormity of the task I undertook, there is the very distinct possibility that I’ve gotten something wrong. If so, please point it out.
I took all of the players in the League who played keeper more than once and filtered them by the 20 players who played more than half of their team’s games as keeper. This resulted in one keeper for every team in the League:
Reina
Szczesny
Schwarzer
Vorm
Cech
Friedel
Al-Habsi
Hart
Howard
Ruddy
Begovic
De Gea
Foster
Kenny
Given
Hennessey
Krul
Bogdan
Mignolet
Robinson
I then removed superfluous events from the data set, such as shots and shots on goal, and focused on the keeper’s main contributions to the team: saves, claims, penalty stops, distribution, etc. I will say here that in stats that don’t have much meaning, Szczesny led all keepers with 3 successful dribbles. Al Habsi, Cech, Howard, and Foster all had 1, and no other keeper had a successful dribble. Meanwhile, Begovich and Reina were the only two to have an unsuccessful dribble attempt. There was also the anomaly of Howard’s long range goal and Robinson winning a penalty for Blackburn. All of those I made note of for their “gee-wiz” factor and moved on.
In terms of passing the ball and “Goalkeeper distribution”, which are in the Arsenal system a form of defense, Szczesny’s numbers are exactly where you expect them to be.
Among goal keepers, Szczesny was second in the League in passing percentage with 70% and 2nd in total passes completed. That’s just behind Swansea’s Vorm at 71% and ahead of Liverpool’s Reina at 70%. However, if you remove free kicks, Vorm was the hands-down most prolific passing goal keeper last season with 706 passes, 209 more than 2nd place Cech of Chelsea. Szczesny comes in at a respectable 487 non-dead ball passes, good enough for fourth. I expect to see this pure passing number climb for Szczesny now that Arteta is playing in the deep lying midfield position, much like Brendan Rodgers had Joe Allen do for the Swans last term.
In the final third, Szcz only misplaced 13 total passes all season, again, the same number as Vorm and with a completed passing rate of 97% it doesn’t really matter whether it’s tops in the League or not, just so long as it’s not 84% like Paul Robinson. Just to give an illustration of how important that number is to Arsenal: if Szczesny had only completed 84% of the passes in his own final third, and given the number of passes that he attempted, that would have meant 74 misplaced passes in his own area. In fact, accounting for all passes in his own half, Szczesny passed the ball at a 91% rate which was second best in the League and meant that his defense didn’t have to worry about the keeper’s distribution behind them.
I actually shouldn’t use “distribution” in that last sentence because it’s a term of art for keepers and refers to their rolling the ball out, punting it away, or goal kicks. Again we see a familiar cast of characters topping the list in terms of distribution percentage; Reina is number 1 at 79%, Szczesny is 2 at 75%, and Vorm is tied for 3rd with 72%.
In his sweeper-keeper role, Szczesny led the League in keeper clearances with 49 last season and won the second most number of duels (tackles and aerial headers) behind Vorm. All of the keepers who play for the top teams are forced into duels and Szczesny did well in all of his attempts, winning 3 headed clearances (2nd behind Begovich), 1 last man tackle (tied for first with Begovich and Reina), and only conceded 21 corners all season. A number which is good enough for 7th best on the list.
And finally, Szczesny was outstanding in what people often refer to as “commanding his area”. This is typically measured in how often a keeper comes out and claims a cross or high ball off the opposition’s corner or dead ball situation. Szczesny led the league in successful punches with 27 and was 9th overall in total catches with 50. Once he does try to catch the ball, he’s very sure handed with just two dropped balls all last season. Arsenal weren’t tested the most off crosses and set plays into the box, that was Wolves’ Hennessey with 2.59 per game but Szczesny did have to deal with over 2 per game which puts him nearly 25% more per game than De Gea and Hart. And considering that possession numbers for Man City were similar to Arsenal that’s a significant challenge to the young Pole. One which did well to deal with managing an overall 3rd best 93% claim rate.
Wojciech Szczesny is very much the type of keeper that Arsene Wenger’s system demands: someone who fits into the possession type offense and can recycle the ball very efficiently in his own half (91%). Szczesny also does the job as last defender (clearing the ball and making last ditch tackles, etc) when Arsenal press for the winning goal — though some will say he was called upon to do this too often last term, a point we will examine in the next article. Where Szczesny has started to endear himself to the Arsenal faithful is in the way he controls the 6 yard box on corners and other set plays. Arsenal grew a reputation for a soft touch on set pieces over the last few years which will take time to overcome. Based on Szczesny’s performance last season, and some of the defensive organization this season, the club are well down that road. West Ham on October 6th will surely test that supposition.
Tomorrow we will look at where Arsenal’s defense needs improvement and where Szczesny could help push the club in the right direction.
*What we refer to as “pants-shitting moments” here on 7amkickoff.
Splendid job Tim..
I don’t realize that the possession football team are the team who put keeper “distribution” as one of the core function. I’m a bit surprised that Szczesny has a good “distribution” ability. That is because I always heard that Szczesny has a bad “distribution” or more specifically a bad kicking accuracy. And I do notice the difference when there is Sagna and no Sagna.
If we put the definition of GK distribution as rolling the ball, passes, punting it away, and goal kick, I’m slightly wonder why a possession football team who rarely get attack by the opposition use GK a lot. I always though that the one who top the “distribution” is the team who defend a lot like Stoke perhaps.
I can’t wait to see the second part of the post. I wonder which part of Szczesny, as a young GK can improve because from this number bar the conceding part, his very good.
If by “distribution” you mean “hoofing” then yes, Stoke lead the league.
Their percentages are obviously going to be well down on teams like the Arsenal, who play out short passes from the back unless under pressure — at which point Szcz booting it upfield would count as a clearance, I believe?
The main thing to remember is the lad’s age, too. His numbers are respectable next to Cech’s, but he’s about 8 years younger. I’m bloody excited to see how good he can really get.
All this time I thought Cech was closer to 40 years old
Well, although it’s true about the ‘hoofing’ part, I mean to say how a team who get attacks a lot will going to use their GK more. I thought with GK being use more, he could have the ball more and so distribute the ball more.
Tim about the GK, Is there any data that can describe his positional sense? At least how much did he was in the wrong place by moving toward the opposition too soon like Almunia used to do. I do remember that some of the goal that Szczesny concede is come from his rash decision, like the goal from Swansea if I’m not wrong.
Nope. And frankly, until I learn how to do more database programming, I have no desire to have a bigger database.
Hehe. You said “database.” Hehe.
Got it Tim. Then I will enjoy everything that you have came up to. Thanks for your hard work for us reader.
Interesting stats tough I would say that I wasn’t impressed by what I saw from Scz. last season. Of course, I just watched a few games, so I cannot have an accurate overall impression of the Pole’s performance. It would be even more interesting to see his save ratio per shot on target, as well as his save ratio per shot from outside of the box, both compared with other keepers in the league. I have seen other keepers save clear cut chances regularly but have some feelings that Scz. is not decisive enough.
for all the arguments you can raise against Fabianski, I remember that 14 games with some man-of-the-match performances he had (when he finally gained his confidence) just before the back injury that forced him to the sidelines and allowed Scz. to become number one. For me, that was the best strings of performances I have seen from an Arsenal GK for some years now. For now I can only dream of a GK who could reproduce such performances on a regular basis. For sure people will remember his blunders more than that.
If it makes you feel better, I remember this period and also lament that Fabianski wasn’t able to continue that form in the long term. With that said, I’m nowhere near as pessimistic on Chesney as you are.
People’s perceptions of Fabianski are based largely on when he was a backup who only occasionally played. With a run of games, he looked an excellent keeper indeed, making barely any mistakes. He doesn’t have the aura of confidence that Chesney has, but he was very competent before he got injured. I think he was actually our best keeper that season.
But when he has featured as a fill-in since then, he’s looked more like the Fabianski of old. Without a run of games to build consistency, he is flappy and doesn’t look like he really fits.
Interesting read indeed. Who said statistics were useless.:D
If you are referring to me, I have neither said stats are useless, nor even implied that. What I believe is, that they are of assistance rather than defining. For instance, an actuary can tell you how many people per 100000 will die in a fatal traffic accident, great, but not who.
You put a hell of a lot of work into this, fair play to you.
Slightly off topic, but amusing. Utd bought a goalie some years ago. In his first game, through complete panic made some great saves, to get MOTM. He disappeared shortly after that.
Nice work Tim. Some people criticize Woj, this sure is one hell of an answer!
What I would like to see though is him improving his save rate. Perhaps it was just that last season the defence let too many chances for the goalkeeper to save, but I am sure he needs to improve in that area.
Tomorrow.
Thanks for all the hard work. Really appreciate it!
Regarding Szczesny’s perceived poor distribution, I think it does have to do with his long kicks. My perception, and I have no data to back this up, is that Woj lacks both accuracy and distance on long kicks. My perception is that we retain the ball poorly, compared to say Stoke or any Allardyce coached team, when he kicks it long. I also understand that advancing the ball and establishing possession in the attacking third from long goalkeeper kicks also depends on having forwards and midfielders who can win headers and fight for the second ball. Without those players, it doesn’t matter how good Szczesny gets at long kicks, it’s not going to be a good way for us to establish possession in the opposition half. Is there any data to confirm this perception? And if the data confirm this, why do even bother? We’re certainly a possession based team who try to build attacks patiently from the back so why go away from that just because the keeper’s got the ball in his hands?
Quick question Tim…
As some one who have seen the MCFC data and Opta Data..
Can you please tell me in which extent Opta data is richer??
Does opta data give pass by pass co-ordinates? Does opta give kms run etc, which I think was missing in MCFC data..Also in which half, howmany forward, backward, sidewise etc?
Here is something interesting stat that I found, which puts a perspective on the defensive system in front of Sczz..and also the oft criticised Arsenal style of iiiiline..
The top 5 defenders in the league in terms of Last man tackles..
Koscielny 9
Vermaelen 5
Coloccini 4
Kaboul 4
Mertesacker 3
Another one is Error leading to goal
Note that people were pissed off we didnt bring Reina or Schwarzer
Hennessey 6
Robinson 5
Schwarzer 5
Bogdan 4
Reina 4
Begovic 4
Szczesny 4
Szc though leads in Errors leading to attempt…which is where I assume the first stat that I mentioned kicks in
Szczesny 3
Mignolet 3
Sørensen 3
Bogdan 2
Begovic 2
Reina 1
Lastly….Error leading to goal and Error leading to attempt for defenders…
Terry 3 Koscielny 3
Figueroa 2 Ferdinand 3
Vermaelen 2 Assou-Ekotto 3
Cuéllar 2 Hangeland 3
Mertesacker2 Martin 2
and I thought EBJT cant do no wrong.
My guess is this time the defensive organisation is to reduce the last man tackles by reducing such situations..where the only one available to tackle is last man.
Sczny as suggested is ideal for the wenger system…just that the defence before him also need to be consistently available. I think we have almost 15+ combinations of last 4 in the league.. Guess all contributes to the GK stat…
I have two minds to go to Opta..catch them by the collar..and mouth…”Gimme the data Lebowski…”
Opta provides (for a massive fee) an XML stream of all live games which includes all of the things you’re looking for (except distance covered, I think). So, yes, (x,y) coordinates for every single action on the pitch. If you’ve seen the 442 app or the Grauniad Chalkboard then you know what I mean — that’s how they get their data.
Pingback: