Previously, I wrote that I’ll especially be watching Dortmund’s Ilkay Gündoğan this season, observing how he continues developing into his crucial role as the attacking-minded defensive midfielder (DMF) in BvB’s 4-2-3-1 system. I also argued that we need to stop expecting Gündoğan to become Nuri Sahin 2.0, since the two have such different styles of play, meaning there is no one-to-one correspondence in replacement value between the two Turkish German midfielders.
So now that we’re nine weeks into the Bundesliga campaign, it’s time to check in with Illy and Dortmund to see what what’s happenin in the Schwarzgelben midfield. Below are the big numbers through seven starts and two other appearances in both the Bundesliga and the Champion’s League (number in parentheses are 2011-2012 final averages). Note: no two sources provide exactly the same numbers; Bundesliga.de contains only league stats, while WhoScored.com contains some CL stats. Additionally, WhoScored.com tends to count fewer passing attempts and more longballs, for example. Regardless, here are Gündoğan’s stats in both the Bundesliga and Champion’s League:
- 1 goal, 1 assist.
- 18 scoring chances created, 0.6 key passes per game (vs. 1.2).
- 1.7 shots per game (vs. 1.3).
- 86% pass completion rate (vs. 85%), thanks Freiburg snow!
- 46.4 average passes per game (vs. 48).
- 4.8 accurate longballs per game (5.1).
- 83% of longballs completed (vs. 83%)
- 2 dribbles per game (vs. 2.3).
- 1.7 times dispossessed per game (1.8).
- 1.3 interceptions per game (vs. 2.3).
- 56% tackles won, 2 tackles per game (vs. 2.8).
- 54% of challenges won.
- 3 yellow cards.
- 7.3 average rating in all competitions, according to WhoScored.com’s system (vs. 7.24).
There you go. The Gündoğan trading card contents. (Aside: does anyone else love those Marco Reus Topps trading card commercials at half time?)
Frankly, I was surprised to see what looks like a slightly significant statistical down tick in a few stats:
- key passes.
The difference in positioning isn’t drastic, but enough to illustrate Klopp’s use of his DMFs. Finally, Kehl’s role as a passer is different from Gündoğan’s. A majority of the former’s passes tend to be lateral or angled back to Hummels and Subotic, while a majority of the latter’s passes tend to be angled up the pitch through various midfield channels.
Let’s address the other two stats that have dipped down from last season for Gündoğan: key passes and longballs. Frankly, I don’t hold much stock in the key passes stat WhoScored.com uses, simply because there is a coding consistency issue in labeling passes as “key,” which begs the larger issue of what exactly is a key pass (hat tip to Bundesliga Fanatic reader). Perhaps Illy needs to hit the woodwork more often, then hope for a deflection a teammate can try to whip into the goalmouth? Ta da! Key pass achieved.
Next, the longballs, which are defined as passes of 25+ yards. Illy has connected on 44 of 53 longballs, averaging 4.6 a game, for a completion rate of 83%. Although Gündoğan’s per game average isn’t as sparkling as Europe’s leader, AC Milan’s Riccardo Montolivo (12.4 per game), or Andrea Pirlo (10 per game), or even the Bundesliga’s own Bastian Schweinsteiger’s 9 per game, Gündoğan’s average is still among the best for midfielders in the Bundesliga. However, Illy’s completion rate is better than Pirlo (80%) and equal to Montolivo and Schweinsteiger. Besides, I would assert that the quality of longballs trumps quantity.
Longball accuracy is one of Gündoğan’s chief assets, since it allows him to accelerate and stretch the BvB attack. His week 6 longball wizardry against Gladbach illustrates just how far his passing stretches the pitch:
I’m fond of this chalkboard with its fascinating passing distribution – the passes clustered like twin cities along the banks of a wide river, which winds through the center of the pitch, being crossed here and there with bridges formed of Gündoğan passes. In this match, Illy made friends and influenced people. For example, take his passing distribution (ordered from most to least, passes received from Gündoğan in parentheses):
- Schmelzer (10)
- Kehl (9)
- Götze (8), Piszczek (8)
- Blaszczykowski (7)
- Reus (6), Hummels (6), Subotic (6)
However, “sprint” chalkboards alone don’t account for his movement; for example, Bundesliga.de tracks “intense runs” numerically, but doesn’t plot these events on its chalkboard matrix. So I would argue that a key facet of Gündoğan’s value on the pitch for BvB isn’t even accounted for by the available metrics. I’m not arguing he’s got the clichéd intangibles, those elusive athletic goods of myth, rather it’s that he’s doing something we don’t entirely quantify yet, besides the smörgåsbord of heatmaps, sprints, and passing chalkboards. Anyhow, these tools don’t quite capture the quality of the movement.
So what I’m left with, in terms of evaluating Gündoğan’s value to BvB, is the “optics test,” that is, what I see when I watch the match. Here’s my tableaux image: the camera is filming the middle of the pitch. From outside the frame, Gündoğan darts into the camera shot to relieve a pressured teammate by taking a pass, shifting around a bit with smooth dribbling, then threading a ground ball somewhere close to the box. This tableaux repeats. The point of this image is that Gündoğan always breaks into the frame, already in motion – coming from somewhere, going somewhere – allowing his teammates to generate the ramifying effects. Because of Illy’s quickness, defenders have to mark him, or he’ll cause chaos around the box. By contrast, neither Leitner nor Bender provide this threatening motion as replacements when Gündoğan is absent. (Nor do they produce highlights like this.)
But my tableaux is only as good as its function: a generalization to clarify essential movement. The danger is that it could trigger something like the confirmation bias when I look at chalkboards and the stats columns. But even if I try to avoid this possible bias, the other problem in writing cause-and-effect analysis pieces like this is attributing meaning to what simply could be random effect.
I first began this piece trying to think of a hypothesis to test, but ran into some trouble. Obvious attempts are too vague, for example: “Dortmund is a better team with Gündoğan on the pitch.” Okay, but what does this statement actually mean? Better how? In what regard? How can you isolate Gündoğan variables to test the hypothesis? Getting more precise doesn’t help much either: “Dortmund’s midfield is more effective with Gündoğan on the pitch.” Still problematic. Effective in terms of what – ball possession? Pass completion? Scoring chances? What? Besides, even if Dortmund has better stats when Gündoğan is on the pitch, each metric’s value is relative, requiring more contextual information to be helpful – what was the weather? What formation did Dortmund face? Who was the starting XI? Was something Götze was doing freeing up Gündoğan? The possible variables march on.
Ideally, I would want to answer this question about Gündoğan: how many points is added to Dortmund’s table position by having him on the pitch? What I’m asking for here is a footy version of the wins above replacement level stat (or WAR; see this infographic example) that baseball saberheads know well. Perhaps a footy WAR is already being generated. (If so, please tell me about it in the comments section.)
However, given the current availability of soccer data, I don’t think I would trust this stat right now, because the kinds of numbers of would it be composed of – say, average of goals scored per touch / per minute, pass completion rates, yellow cards per minutes played, etc. – don’t get you an absolute value. I’m guessing that averages would be made from too many random outliers, rendering a statistically insignificant number, or parts of the formula would consist of stats so context specific, they’d be meaningless in a larger sense. For example, Xavi’s ball touches and pass completion rate can be valuable only within the ball control system that Barça plays, but not so much at Real Madrid.
Another method I tried was examining Dortmund’s performance with Gündoğan on the bench (a sort of games control group). However, it gets messy here, too. First, the sample size is pretty small (four games, one of which featured Klopp going crazy with his 3-4-3 formation). Second, Dortmund’s passing stats were pretty similar in these Illy-less games anyway. One difference stood out, however: the Schwarzgelben were outrun by Hamburg in their record-ending loss in week 4. A rare situation indeed. Yet the most damning evidence is that without Gündoğan, Dortmund lost two games and drew one in Bundesliga play. However, I fill not proclaim causation from this correlation.
You’d be silly to argue that Dortmund is not a better team with Gündoğan on the pitch. Our current metrics just render the rationale murky. But I won’t give up on my cause-and-effect quest. My eyes will stay on Illy Gündoğan for the rest of the campaign – my bet is that his play makes more friends and influences more people, completing the exorcism of Sahin’s ghost.