The goalkeeper position is still a footballing lacuna to me. I try to fill this gap in the football fabric with my own imaginings, but I just can’t get there. The phenomena attached to playing this position professionally seems magnitudes removed from the experience of playing “keeper” (heh) in rec play. So I don’t claim to understand the differences between a good or bad keeper with the exception, perhaps, of Manuel Neuer (yay!) or Roman Weidenfeller (ugh!) these past couple seasons.
Of course, the fact that literati like Vladimir Nabokov, Albert Camus, and Pope John Paul II played the position only widens and deepens the mystery of this particular lacuna. The position is odd in what is requires from its members and in how they spend their time during a match. It’s all so odd that books been written on the topic, e.g. Jonathan Wilson’s The Outsider — a book, whose title ostensibly signals the keeper’s weird status. In fact, I might call the position of the keeper, and, the structure itself of playing the position, a parallel sport of its own within the sport of football.
Need I mention that these outsiders get scapegoated almost every time a goal is conceded?
All of which is to say that I’m usually at a loss for what to say about goalkeeping and specific keepers, especially after viewing “data” related to keepers (save tallies, interceptions, etc.). In fact, the only keeper data I usually look at is passing numbers.
However, I don’t mean to say that goalkeeping isn’t understandable in the same way that other footballing phenomena can be to viewers and supporters. Just that I don’t get much out of trying to figure it out.
But to make a start, I’ve recently begun checking the expected goals numbers for keepers that Alex Rathke has been compiling. Rathke brings the much vaunted “Expected Goals” (or xG or ExG) statistical concept into the mix in evaluating keeper performance. To learn about the concept itself — one that we work with here in the Bundesliga Fanatic — you can read Rathke’s own primer on the topic where he explains his methodology. You can also check out Paul Riley’s primer and model. In fact, I could encourage you to do so before reading further ahead in this article.
Here’s Rathke’s table, featuring goalkeeper goals conceded and expected goal conceded per 90 minutes, or two rates basically:
(Click here to see the Tableau page of this dataset with a larger image!)
The yellowish bars = goals conceded per 90 minutes for each keeper listed, while the blue dots = the expected goals per 90 minutes that keeper was expected to conceded in the same situations.
Obviously, large discrepancies between the two rates should grab your attention, as well as keepers who simply over- or under-perform based on their expected goals rates.
Notice that the data measures rates (on a per 90 minute basis), not raw count stats, which makes sense since some keeper on this list play significantly more minutes than others, who only make occasional appearances.
What’s nice about expected goals (ExG) conceded rates, is that we can isolate keeper performance (mostly!) without worrying about complicating variables, like defenders screwing up, flow of play, etc. For example, in certain doomsday scenarios, like being exposed one-on-one with an attacker, even Neuer will concede goals. Thinking about goalkeeping (and goal scoring!) this way is very similar to how stats like FIP in baseball, which isolate pitching performance from fielding performances, allow us think about a single position.
And in the lush and lovely fluid dynamics, constant motion, and chaos of football, goal shots and attempted keeper saves is about as isolated as you can get in terms of identifying discrete events. As I was recently reminded in a piece by the former player Jeb Brovsky, so much of what’s probably meaningful happens in off-ball moments during a football match — except, perhaps, goal shots and keeper save attempts. These events are something like what I’ll coin “terminal events” in that they happen in such a way that dissolves away prior- and post-context. After all, goals are the telos, the end, or Das Ziel in football. Like a switch, goal-related moments relay a match from and into new states.
Okay, now we’re ready to make some observations.
Our “scrooges” are keepers who stop more goals per 90 than they’re expected to conceded. In this regard, here’s the list of winning keepers, listed from fewest to most goals conceded per 90 with goals conceded in bold and expected goals conceded in bold red.
- Manuel Neuer (Bayern, 25 apps): 0.63 and 0.77.
- Timo Horn (Köln, 11 apps): 0.67 and 1.27.
- Lukas Hradecky (Eintracht, 24 apps): 0.82 and 1.24.
- Oliver Baumann (Hoffenheim, 25 apps): 0.95 and 1.42.
- Peter Gulacsi (Leipzig, 25 apps): 0.84 and 1.01.
- Thomas Kessler (Köln, 13 apps): 0.89 and 1.19.
- Martin Hansen (Ingolstadt, 13 apps): 0.93 and 1.30.
- Ralf Fährmann (Schalke, 25 apps): 1.00 and 1.11.
- Rune Jarstein (Hertha, 25 apps): 1.05 and 1.26.
- Marwin Hitz (Augsburg, 25 apps): 1.05 and 1.19.
- Diego Benaglio (Wolfsburg, 14 apps): 1.27 and 1.50.
- Jaroslav Drobny (HSV, 10 apps): 1.38 and 1.42.
Ever met any “Neuer truthers”? You know, folks who snipe that Neuer is overrated because Bayern possesses so much of the ball and stifle its opponents’ shot attempts so that he doesn’t really do anything? Well, you can concede that point that Bayern do stifle its opponents, then point out that still Neuer saves more shots than expected. Still. Moreover, pairing Neuer with the Bundesliga’s (and Europe’s?) stingiest defense is a domineering combination, offering much by way of explaining Bayern’s 4 Bundesliga titles in a row. Playing keep away is still the Bavarian way.
But speaking of star scrooges, man, Köln’s Horn and Hoffenheim’s Baumann’s have been spectacular this season. Each keeper is almost single-handedly saving his side an extra half goal per 90. If you haven’t done so yet, you need to add Horn and Baumann to the very top-tier of keepers in the Bundesliga. And you better believe the summer transfer market will be following suit in a couple of months. Hopefully, both Horn and Baumann will become German keeper commonplace upon the lips of football commentator around Europe after this season.
However, Köln seems full of keeper talent, the weird place. When the veteran Kessler filled in for an injured Horn earlier this season, he performed fantastically. Perhaps Stöger changed up his side’s defensive strategy a bit in these matches — or the Billygoats just got plain luckier? — notice that Köln’s ExG conceded rate was slightly lower when Kessler was keeper. However, given Kessler’s age (31), it’s unlikely he offers more than a season or two of coverage for the FC in case Horn is scooped up with Premier League money.
Finally, Jarstein’s inclusion on this list is a nice testament to the Norwegian’s growth as a keeper this season. And where would Eintracht be without Hradecky, seriously? Or Wolfsburg without the over-performing Benaglio? Yes, the Swiss keeper is actually having a nice season, thank you very much.
Our “Givers” are simply keepers who let in more goals than they’re expected to let in, based on the ExG rates. So here’s the list of “giver” keepers listed from fewest to most goals conceded per 90, again, with goals conceded in bold and expected goals conceded in bold red.
- Yann Sommer (Gladbach, 25 apps): 1.32 and 1.16.
- Koen Casteels (Wolfsburg, 11 apps): 1.38 and 1.22.
- Christian Mathenia (HSV, 8 apps): 1.38 and 1.28.
- Bernd Leno (Leverkusen, 25 apps): 1.53 and 1.17.
- Roman Weidenfeller (BVB, 7 apps): 1.57 and 1.00.
- Alexander Schwolow (Freiburg, 25 apps): 1.58 and 1.30.
- Jonas Lössl (Mainz, 25 apps): 1.74 and 1.38.
- Orjan Nyland (Ingolstadt, 9 apps): 1.82 and 1.19.
- Michael Esser (Darmstadt, 22 apps): 1.88 and 1.55.
- Rene Adler (HSV, 17 apps): 1.91 and 1.81.
- Felix Wiedwald (Werder, 15 apps): 2.52 and 1.74.
First, the surprises: Sommer and Leno. On reputation alone, these keepers are imagined probably to be in the first group, rather this group.
Yet here they are.
Sommer’s rate is probably a reflection of his poor start this season, while Leno … well, I dunno. Once upon a season, Leno was the heir apparent to Neuer. Now? Well, he’s been usurped by Horn and Baumann. With 1.17 expected goals conceded, you can see that Leverkusen’s overall defense has been above average this season and that Leno has actually cost them table points, surely. Anyhow, watch this spot — I’m intrigued to see how Leno recovers from what’s been a disappointing season for himself and club.
Meanwhile, over in the realm of the beleaguered, you’ll find old pro Rene Adler, who can’t buy a break. His HSV has the Bundesliga’s worst expected goals conceded per 90 rate (1.81), while conceding 13 shots a match to opponents. So I’d be hesitant to scapegoat Alder for anything that happens at HSV. I’m sure there’s some keeper psychological index number for shell shock we can run here. Right?
However, I don’t mind if you cast aspersions on either Freiburg’s Schwolow or Mainz’s Lössl. These guys have cost their clubs points, surely, which is a big deal, since both clubs heavily rely on developing unnoticed talent (whether in-house youth prospects or from other smaller clubs) to stay afloat. Having an underperforming keeper is a “luxury” neither club can afford. However, Schwolow is relatively young (24) for a keeper, meaning he could have a much higher ceiling of improvement than Lössl (28).
Then we have the just-not-good-enough keepers, a list populated by VfL’s Casteels (alas! did we think he was past this?), D98’s Esser (salient effort though), and FCI’s Nyland (yeeeaaah). These three are supreme givers of ironic goal gifts. I don’t have much else to say about them.
“What!!!!” you protest, “No Wiedwald in the paragraph above? We all know Werder’s keeper is inferior stuff.” You gesture toward the discrepancy between his 2.52 per 90 goals conceded and his 1.74 ExG conceded rates. Putrid, right? Well, it was. Wiedwald could have surely won our worst keeper during the Hinrunde campaign, but he’s finally improving. In fact, he’s kept — errr, his club has kept! — a couple clean sheets, and Werder has only conceded 2 goals in its last 5 (!) matches. An incredible turnaround for an infamously generous Bundesliga defensive unit. To be fair, however, this run of stinginess has occurred while facing some goal-challenged sides (Mainz, VfL, Darmstadt, current form Leverkusen, and current form RB Leipzig). Nevertheless, Weidwald has clearly helped Werder immensely during this successful run of results — a run that definitely is saving Werder’s season.
That One Club …
By which I mean Borussia Dortmund. First, we can safely concluded that Germany’s “Second Club” doesn’t have elite goalkeeping. On one hand, BVB has a prolific veteran “giver,” and on the other hand, BVB has the curious Roman Bürki, whose rates perhaps define “replacement level” goal keeper, to borrow an American baseballism.
First, poor ol Weidenfeller. The former Mannschafter at 36 is the Bundesliga’s most generous giver. Weidenfeller ExG conceded is 1.00 goal per 90, yet he concedes 1.50 goals per 90. Uggghhhhh. For BVB, he’s clearly a bit of a liability when he’s keeping between the posts. So pray to all your entities above and below that Bürki stays healthy!
As for Bürki, his ExG conceded goals rate (0.84 goals per 90) almost exactly matches his actual goals concede rate (0.83 goals per 90). Somehow, BVB are a slightly better defensive side when he’s on the pitch. Perhaps Tuchel adjusts his tactics a bit to compensate for Weidenfeller’s eroding skills, or this phenomenon could be pure randomness.
Latest posts by Travis Timmons (see all)
- Wutrede Hall of Fame: Giovanni Trapattoni, the Original - January 13, 2018
- The Cephalopod Principle in the Bundesliga - November 29, 2017
- Exclusive Interview with Schalke 04 Knappenschmiede Coach Sam Farokhi - October 11, 2017