What kind of difference does a Goalkeeper make?
Of all of the clamour that came with Brentford’s long-promised promotion to the PL, the one that interested me most concerned their ‘keeper, David Raya. He had been linked to Arsenal, had been mentioned regularly as the new spawn of the libero-GK hybrids everyone loves now, and had become as entrenched into the notion of Brentford as the phrase ‘Moneyball’. In his promise, in his commitment to modern football, and in his bottling of the play-off final a year earlier, Raya was purest Brentford.
Then he came up, made some spectacular saves, then got injured coincidentally as the Brentford project slightly wobbled in the top flight. In a way of almost underlining his importance, Raya was replaced twice by the club, first through Alvaro Fernández and secondly through Jonas Lössl (via affiliate FC Midgetland) — and I think it’s fair to say neither have carried the same weight in either skill or reputation. After Raya’s injury in October 2021, Brentford have won just 3 PL games and accumulated a grim 12 points in 17 games — compared to the same points in half the games (9, don’t get all anal about the maths please and thank you) with him fully fit, albeit in easier match-ups. I’m neglecting the games since his return because frankly it’s a bit unfair to expect a star player to be as good after that sort of time out (see VVD notably), but WOW the drop off is startling.
This led me to consider how much damage a keeper switch-out can do to a team and their style and more importantly if this reflects in goalkeeping statistics generally — I need it to, because I cannot bring myself to watch Brentford rest-defence and build-ups for hours on end, and I certainly am not choosing to watch Wolverhampton “If anyone scores we’re not doing our job” Wanderers for any extended amount of time (this applies for the Nuno and Lage variants too).
Admittedly this look does have some notable holes — the natural best place to look, especially when establishing replacement GK comparisons, would be the best teams like Liverpool and City. Here the keepers are so intrinsically linked to playstyle that any change will surely propagate largest, as both the keepers are also in that Top 5 in the league bracket for their function; whether Ederson is actually a good keeper is up in the air but still. Unfortunately, both teams have long-term purchases that don’t need replacing, nor do their keepers get injured for significant spells such that their backups can get statistically-conclusive minutes. Instead, we’ll try look at teams which have a) 2 goalkeepers with more than 10 full 90s in the last 2 seasons b) teams which have interesting quirks in their situation regarding ‘vibe’ and c) not Newcastle because who wants to read about Karl Darlow.
Also this study is definitely holistic — whilst a new goalkeeper could force a new style, a new style can develop without that consideration and more generally. New defenders can change style, as can a new coach, so I tried to keep consideration as player-centric as it could be — these discussions are limited to PL only, and over the last two seasons so that most players will be operating under the same tactical set-up. Whether we consider back-ups or replacements or even both (exciting!), we’ll note as we go.
Brentford
Starting with Brentford then, we compare Raya with Fernández over this season — 12 games for Fernández and 11 for Raya.
I mean, its fairly conclusive: Raya despite his reduced ability in shot-stopping is a damn noticeable improvement on his counterpart.
His PSxG-GA, which looks at Post-Shot xG calculations to determine the quality of shots a goalkeeper faces and its differential to the goals they actually concede, is a slight problem falling below the league average and notably below Fernández’s +0.2 (one of the best in the league mind you), but everything else is such a positive that you can kinda ignore that?
His role as a sweeper here is the key and everything links fairly nicely. Sweeping actions are up, therefore so is average position of defensive actions, which naturally improves him box dominance a whole load too — stopping 9.9% of crosses which is genuinely very good (Alisson levels here). This box dominance and confidence in attacking chances earlier will also lead to a reduction in his PSxG/SoT (looking at quality of shot per every shot on target he faces) — compared to Fernández he’s having a frankly lovely time of it with fundamentally the same defence; these two are quite literally at opposite ends of the scale in terms of shots faced. Raya either sweeps his chances before they occur, or has such a grasp of his box and defenders within it that the quality of chances diminish massively compared to under Fernández’s tutelage.
The case moving away from that end of the field is also an interesting one. Raya does far more long passes (over 40m) and has a far longer average goal kick distance (The distance might not be totally accurate visually but they’re on the same scale so its fineeeee). His longer passes are also more accurate, which means distribution wise he’s a lot better for how Brentford want to play — 0 build up, move it to Toney QUICK, CHRIST (for reference, only Newcastle, Everton, Watford and Burnley have less 10+ passes sequences in open play, and they have the 3 rdhighest Direct Speed in their attacks too per The Analyst). His short passes are also down proportionally, meaning the majority of passes are bypassing the 15m mark and going straight down field to at least the midfield. Basically the ball is not staying in Brentford’s half for long either in possession or running towards Raya’s goal with significant danger.
Fernández’s wheelhouse seems to be an on-the-line goalkeeper, weighted towards shot-stopping about all else, which is pretty typical for lower-table teams and is not a bad thing really. Ish. But his massive reduction in outfield actions is a huge hamper on his side’s ability to limit damage against them, and completely deads the side stylistically — not to mention Brentford don’t fancy themselves as a relegation battler which is admirable and maybe slightly arrogant but good on them.
You can see why Jonas Lössl was pursued in January, and his style seems far more fitting with Raya… but his ability to save the ball is questionable so far.
Fair play sports media though, Raya is Brentford.
Arsenal
This is a VERY similar story to Brentford, and it makes a lot of sense as to why Arsenal were linked with Raya for so long. The Other GK (Leno) is worse at sweeping, plays fewer long balls less successfully, relies far more on short passes, keeps his goal-kicks closer to his end of the pitch, and as such has a much higher PSxG/SoT. This all makes sense when you consider that Ramsdale’s proactivity and distribution are his key attributes, and have been the main changes Arsenal have enjoyed this year. The difference with Leno here is that he Is actually a fairly bad shot-stopper in recent times (mainly haemorrhaged by his 3-game stint at the start of this season). Whilst his heights were fairly brilliant (19/20 particularly), his performance last season was just fine placing him 10 thin the league. He is statistically facing some of the hardest shots in the league whilst he plays, and whilst some of this is due to squad changes and style changes under Arteta, a fair amount should be weighted versus his own ability to snuff out attacks.
Even if we remove the awful 3 game run at the start of this season (which showed a lot of Leno’s faults bare-faced), his stats were still not as impressive as Ramsdale — crosses are still more frequently bypassing him, his passes are still short, and the chances are still far far more dangerous even if he is saving them more. He is still a competent goalkeeper but he has never looked the part as a sweeper, and by the looks of things Ramsdale is doing what Arteta wants and was trying to do last season, but now has the distribution to allow for tangible stylistic difference.
Leno’s case is interesting because it allows the question of if a goalkeeper is worth keeping if they prompt such an abandonment of a team’s stylistic principles. Brentford are fairly obviously in the No camp and I’d be inclined to agree. Having a keeper who discourages dominance is not great in a proactive team — he either plays vs lesser teams in cups where you should dominate, or he plays when the primary keeper is out, reducing the ceiling of the side. It’s a Lose-Lose, not to mention unfair on the player by showing off his weaknesses — it feels like a ‘cut your losses’ kind of situation, where it’s a profile that has been surpassed by the club’s projection of itself.
…But equally, playing Runarsson is unfair on the fans, so…. go buy Raya Arsenal.
Chelsea
In another case of a bought-in replacement, Chelsea are another modern success story after Liverpool and City. But in a really weird way.
So, let’s try not to bully a certain keeper. Kepa is a sweeper with higher engagement and longer passing generally, but with the exact same distance of goal kicks and less cross claims. (All pretty fine actually Hm) Despite this, he faces harder shots — a notable trend away from the previous two cases and a Kepa-shaped red flag considering these have basically all been under Tuchel who loves defence more than Sir Alex Ferguson loved bribi… Intimidating referees. (Ah there we go). Mendy is doing nothing spectacular but he is providing a confident platform for the defence and with the regimented organisational defence established under Tommy you don’t need much else; Mendy fits the external no-nonsense presentation pretty functionally. Don’t make mistakes, don’t be too rash, make sure the line in front of you limits the shots your way. Simple (or make it look that way in Mendy’s case).
This is more than probably a case of stats not telling all, and vibe-based inference is doing a lot of heavy lifting here. Kepa is a mistake-ridden nightmare, or at least definitely was under Lampard, and that sorta thing changes a defence’s mindset and the style they adopt. He might communicate poorly, or the defenders may be compensating so much it hurts, or he may just want to do his own thing which Tuchel doesn’t frankly ‘fuck with’. Whichever way, something is definitely off here that we can’t show, and that PSxG/SoT stat is a pointer hidden in a lot of generally fairly solid data that doesn’t demonstrate a horrid option. Beyond this general vibe, Kepa seems a fair fit for the substitute keeper role — Chelsea don’t seem to adapt too much too him offensively, and whilst defensively there is an ironing out of individuality needed until the change is seamless, he has some real positives here.
I still don’t buy it, but Kepa is… fine? STATISTICALLY
Manchester United
Another quirky case, this time where the secondary keeper is the new boy in town, but also a case where the new boy in town is representative of a potential culture shift in the clubs future.
Now I’ll be honest, I’m aggressively anti De Gea. I think he represents the concept of ‘resting on your laurels’ and whilst he was a prospect of modern goalkeeping with his novel technique of ‘kicking the ball’ (more accurate than it sounds), he has failed to even slightly consider developing beyond this one strength. He is one of the purest forms of ‘shot-stopper’, but when his fundamentals run cold (as it has basically every season since 17/18), what is the point of him? This season is an impressive exception, but the clear signs here are that much like Leno, De Gea performs best when he has to be glued to his line because the team (therefore defensive structure) is worse. The reactive far suits him more than proactive; pulling off a wonder save is more his vibe than stopping the need for it. An ego problem? Maybe. A reverberating structural problem? Ooooo boy yes. He’s behind the times by about 6 years now. And his numbers compared to Henderson for PSxG-GA over the last two seasons are not worth the downturn in everything else.
Considering the rest, the passing isn’t anything special. Henderson passes short more which is also a sign of his ability to play out from the back (which De Gea notably sucks ass at), but that might be a reach without looking at the press they both face*. This is perhaps quite nicely supported by Henderson’s shorter average goal-kicks.
Sweeping and box-domination are Henderson’s shit compared to De Gea though, being much more aggressive in every metric and having the supporting PSxG/SoT numbers to boot — De Gea’s numbers are awful here mind you, facing some of the hardest shots and being the second worst sweeper this season (and likely not far off last year given his average).
Man Utd would do well to move on from De Gea and catch-up to their modernisation process under Rangnick et al. Henderson is an improvement and is definitely a functional back-up at least, but what is worrying here is the total distinction of style between 1 stand 2 ndchoice. There is a real lack of tactical cohesion between the two, which is maybe when Henderson hasn’t played this year, but if anything I feel this is the wrong way round. De Gea needs to be replaced by a more modern goalkeeper if Man Utd want to keep up with the big(ger) boys, or else he risks becoming a very blatant illustration of the club’s biggest issue: claiming to be the best in the world, but not doing much to prove it since 2013.
*Again this is a problem of holistics — De Gea might be pressed more because he’s bad at short passing, so does it less, but also might pass short more because of general team instructions under Solksjaer which was generally attack very fast and… uh… yeah? Solksjaer for you.
Wolves
The final example is maybe a little note to Man Utd, and is a club following (maybe?) from Man City and Liverpool’s example by making a noted stylistic choice for their goalkeeper (at far less expense).
José Sá has been a statistical revelation in the league, and is a large part of why Wolves have been dodging xG like it’s COVID. Sá is the highest rated goalkeeper for PSxG-GA in the league p/90, and second total behind De Gea, and his stat means that over the course of the season he alone saves Wolves between 12 and 14 goals a season. What the fuck. This is a 17 goal differential to Patrício last season, with Sá also contributing in the PSxG/SoT side too with a 0.04 drop per shot (which is a 26% drop weighted to the metric limits in the PL).
His sweeping is wayyyyyy up, as is his cross collection and positioning. Patricio is old before his time looking at his passivity, and supremely old-fashioned in that sense too. He’s showing every red flag under the sun at Roma so props to Wolves for getting rid of the rot quick. Sá might just be running very hot, look at De Gea for instance, but even then his underlying work in sweeping and cross claims are up there at the very top end of the league so this looks like an improvement any way you spin it. His passing is weird compared to Patrício but then Wolves are weird too — looks like a lot of medium range passes at like 30–40m (~40 yards) with reduced extremes above and below, which lines up with his average pass length. His volume is low but a notable increase on Patricio who was actually the worst in the league for pass volume last year (albeit with an improved completion). This one needs some eye test to see his influence in style, but damn it’s a very visually impressive win for their scouting, and has allowed Lage to forge an equally impressive unit once everyone settled down. If he can maintain this, Wolves have a serious platform for an adaptational style — Box-controller when they’re penned in, and a hell of a sweeper when they look to control themselves. As they continue to establish themselves as an upper-mid-table force, a keeper like Sá that can do whatever they need. Whilst there is no back-up at Wolves, they have identified their #1 in a style properly akin to Alisson and Ederson. How it changes when they are forced into change is a discussion like Chelsea’s, but right now it’s all coming up Wolverhampton.
Conclusion
Man Utd need to take note. In this discussion they are being left behind pretty impressively, with sides all around the league making a commitment to a keeper who matches the modern game, and how they see themselves within it. Brentford have very clearly identified its importance, and are willing to make dramatic moves no matter the optics, and Arsenal and Wolves doing their bit to improve their defensive baseline in the summer, albeit without the appropriate replacement in the wings. Chelsea are I think trying to adapt Kepa to their new style, which is commendable and perhaps a result of how goddamn expensive he was, but Man Utd? Their number 1 is outdated, and their number 2 is not at that elite level they want. I didn’t want this to be a hit-piece, but sometimes this club makes it so easy. Goalkeepers are vitally important, stylistically and for club performance as a whole. Man Utd should stop trying to let everyone know they’re Manchester United. Sometimes it’s just a matter of being a little bit more like Brentford.
As an aside, if you’re looking for actual statistical and smart discussion of goalkeeping, check out John Harrison’s stuff, guy’s a legitimate genius
All data via Fbref unless stated.
Originally published at http://jakewfox.wordpress.com on February 25, 2022.