Editors Note: The stats that were collected were as of August 26th, which means the Vancouver Whitecaps match vs the Montreal Impact on the 28th was not counted, so it is not mentioned in the piece, nor is it added to the stats.
What started out as a look at trying to figure out how good Maxime Crepeau has been for the Whitecaps this season, as he has helped turn Vancouver back to a decent defensive team after a porous 2018, ended up turning into a look at 26 goalies around MLS, as we ended up looking at which goalkeepers over and underperformed relative to their teams.
Maxime Crepeau had himself a game last Saturday. While his side dropped a 3-1 result away to San Jose, the Canadian international stole all the headlines, as he put together an MLS record 16 save performance in a match that was not as close as the scoreline would suggest.
Despite the heroics of Crepeau for much of the season, the Whitecaps remain at the bottom of the MLS standings, last in the West and in second-last overall. It led us to wonder: How much worse could it be? After seeing Stefan Marinovic and Brian Rowe struggle in the Whitecaps goal last year, we decided that we would do some statistical analysis to find out how much of a difference Crepeau has truly made for the Caps.
To try and figure out if it was possible to find out how much of a difference Crepeau has made, we collected some of the stats available through American Soccer Analysis, who provide goalkeeper stats (without own goals), as well as Expected Goals (XG) data. We took all 26 goalkeepers that have played at least 10 matches this season, as well as 2018 Stefan Marinovic and Brian Rowe, and made a chart to analyze their performances between the sticks for their respective clubs.
First, we took the simple stats, which is to take the number of shots each goalie faces, then take the saves they have made, and calculate their save percentage. Pretty simple, right? But it seemed incomplete, as it didn’t factor the quality of chances each goalie faces on their respective teams. It meant a top goalkeeper on a bad team would be rated the same as an average one on a top team, despite the fact that the quality of chances each would receive in a game would be completely different from the other, not to mention the number of chances they would face overall would be largely skewed.
So what we did was to take the GA-XGA stat, which takes both the XG (Expected Goals) each goalkeeper was expected to let in based on the chances given up by his team and takes the number of goals they actually conceded, and removes the XG from the GA, making GA-XGA. GA-XGA is either positive or negative, with a negative number meaning that the goalie saves more than he should for the quality of chances he receives (IE: he is actually doing good), and a positive number meaning that he lets in more than he should (IE: he is not doing so hot).
From that, we were able to create an adjusted save percentage, which was calculated by taking the original save percentage and subtracting the GA-XGA from it. So if a goalie was at a 0.66% save percentage, but he has a GA-XGA of -5, his adjusted save percentage would be 0.71%, which better reflects the calibre of goalkeeping performance that he puts in relative to the team he plays in.
Here is the table (it was ordered by the amount of saves in 2019, with the two 2018 Caps goalkeepers inserted at the bottom). Also to note, the average save percentage in MLS this season is 0.67%, and the average adjusted save percentage is also 0.68%, which means anything above those numbers would be considered good.
|GK||SA||Saves||PCT||GA-XGA||Adjusted PCT||GK||SA||Saves||PCT||GA-XGA||Adjusted PCT|
|David Bingham (LAG)||153||115||0.75||-5.79||0.81||Bill Hamid (DC)||145||109||0.75||-3.29||0.78|
|Max Crepeau (VAN)||134||98||0.73||-0.34||0.73||Vito Mannone (MIN)||127||90||0.71||2.81||0.68|
|Stefan Frei (SEA)||122||84||0.69||-0.47||0.69||Tim Melia (SKC)||128||84||0.66||0.26||0.66|
|Brian Rowe (ORL)||116||83||0.71||-1.83||0.73||Daniel Vega (SJE)||118||77||0.65||-4.99||0.70|
|Luis Robles (NYRB)||115||75||0.65||0.29||0.65||Brad Guzan (ATL)||103||73||0.71||-3.54||0.75|
|Joe Willis (HOU)||107||70||0.65||0.93||0.64||Jesse Gonzalez (DAL)||98||65||0.66||-0.10||0.66|
|Tim Howard (COL)||103||65||0.63||0.65||0.62||Evan Bush(MTL)||116||65||0.56||9.82||0.46|
|84||64||0.76||-3.37||0.79||Spencer Richey (CIN)||106||62||0.58||3.49||0.55|
|Nick Rimando (RSL)||84||57||0.68||0.62||0.68||Quentin Westberg (TFC)||87||55||0.63||1.71||0.61|
|Matt Turner (NER)||75||56||0.75||-7.10||0.82||Steve Clark (POR)||68||52||0.76||-6.50||0.83|
|Sean Johnson (NYFC)||77||50||0.65||-0.83||0.66||Andre Blake (PHI)||72||41||0.57||8.29||0.49|
|Jeff Attinella (POR)||64||40||0.63||-0.65||0.64||Zack Steffen (CLB)||56||39||0.70||-1.67||0.72|
|Kenneth Kronholm(CHI)||57||35||0.61||3.38||0.58||David Ousted(CHI)||53||33||0.62||-0.05||0.62|
|2018 Bryan Rowe(VAN)||49||29||0.59||3.40||0.56||2018 Stefan Marinovic (VAN)||109||65||0.60||5.90||0.54|
League Average: Save PCT (0.67%) Adjusted Save PCT (0.68%)
Max Crepeau has put up a good campaign, finding himself 7th in MLS for adjusted save percentage. The other goalkeepers ahead of him? David Bingham (LAG), Bill Hamid (DCU), Brad Guzan (ATL), Tyler Miller (LAFC), Matt Turner (NER), Steve Clark (POR), all goalkeepers on playoff teams, except for Clark, who’s Portland is currently 3 points out of 7th place in the West, but they do have 2 games in hand on the team in 7th, FC Dallas. Another interesting fact about the list is the name after Crepeau, which is Bryan Rowe. After putting up an adjusted save percentage of 0.56 last season, which would have him tied for 23rd in 2019, he now has a 0.73 mark, which is a pretty crazy resurgence all things considered.
Crepeau: Bang for your buck?
After last season, where Marinovic and Rowe each cost the Caps around $130 000, having Crepeau on $90 000 has been huge for the Whitecaps. While he does have a new contract extension that kicked in this summer, one that will likely see him earn something similar to what Rowe and Marinovic earned, if not more, the value he has provided on his own trumps what Rowe and Marinovic ever did as a tandem.
To give an idea of how valuable Crepeau really has been, we decided to see how many goals that his performances this year would have saved on last years Caps, as well as check how many he would have saved on the Impact, who have struggled defensively in part due to Evan Bush’s regression.
Crepeau on 2018 Caps:
Note: Own goals do not factor into these stats, as XG is not influenced by them. We will put them beside the conceded number, but you have to add the OGs to the total to see the true total, I.E: the total that shows up on the standings.
2018 Whitecaps: 162 shots against, 64 conceded (3 OG), 0.60% Save PCT. XG: 53.
2019 Crepeau: 135 shots against, 36 conceded (1 OG), 0.73% Save PCT. XG: 36
2019 Whitecaps: 170 shots against, 45 conceded (2 OG), 0.74% Save PCT, XG: 48
So to adjust Crepeau’s stats to 2018, firstly we would take his save percentage, and plug it into last year’s shot stats.
162 Shots times 0.73= 44 conceded, 118 saved.
We would then take the Whitecaps XG, and calculate Crepeau’s GA-XGA.
44 conceded-53= -9 GA-XGA
Therefore, his adjusted save percentage on last year’s team would have been a mind-boggling 0.82, which truly shows how poor the Caps goalkeepers were for them last season. Max Crepeau’s form this season would have saved the Whitecaps 23 goals more as it stands, and even if his stats fall a bit by the end of the season, anything over a 10 goal swing is pretty impressive, as even just that would have probably meant the Caps in the playoffs, as they only missed by 2 points (as you would figure a 10 goal swing would amount to more than 2 points gained in the standings).
To give an idea of how poor that was, if you plug in an average MLS keeper (0.67 Save %) to last year’s team, they would have conceded 53 goals, which is a 0 GA-XGA. That means that an average MLS goalkeeper would have conceded 14 fewer goals than the 2018 GKs, despite an adjusted save percentage of 0.67%, 1 point below the MLS average. (Which means based on the shots the Caps gave up last year, as well as their XG against, 0.67% is the adjusted percentage they would have needed to stay exactly at their XG against of 53, which is to say: did Rowe and Marinovic ever hold back last years team).
If we were to reverse things, say do the 2018 Caps goalkeepers on the 2019 team, the picture continues to look bleak for last year’s side. Plugging in their combined 0.60%, the Caps would have conceded 68 goals already this year, an astonishing 23 more than they have this season, and that is despite the fact that there are 6 games remaining in 2019. Given that is a +20 GA-XGA (!!), that would have meant an adjusted save percentage of 0.40%, a full 27 points below the MLS average.
While, as we have seen with the resurgence of 2019 Brian Rowe, last year’s Whitecaps goalkeeping performance was historically bad, and that no goalkeeper should realistically be expected to repeat anything like that, so for all you Montreal fans, you can breathe easy (ish) for next year, as Bush is not as bad as he has shown this season, despite his age curve suggesting a decline.
However, because it’s still interesting to note, we did also check 2019 Max Crepeau’s stats on the 2019 Impact, given that they gave him away to Vancouver for $50 000 GAM and a Draft Pick. Given that Evan Bush has played all 28 games up to this point of the campaign, it makes things easier, as we can just plug Crepeau’s stats into Bush’s, making things easier. Before we do that, here is a visual comparison:
2019 Crepeau: 135 shots against, 36 conceded (1 OG), 0.73% Save PCT. XG: 36
2019 Bush: 116 shots against, 51 conceded (1 OG), 0.56% Save PCT. XG: 41
So if you were to plug Crepeau’s stats into Montreal, he would represent a full 20 goal swing, only conceding 31 goals, which if we were to calculate the GA-XGA, is a -10 score (really good). But given Bush has a +10 rating (horrible), even if you were to plug an average MLS goalkeeper (0.66), they still only would have conceded 39 goals, 2 less than Montreal’s XG for a -2 GA-XGA, which shows us that their defence has not done Bush many favours, because if you plug in an average keeper and he still outperforms your XG, you should probably lower that XG, as well as your shots.
What does all of this tell us?
There are a few interesting takeaways from this exercise. While it gives a good idea of who is under and over-performing relative to how their teams defend, it still requires careful interpretation, because A) goalkeepers are volatile and vary from year to year (Brian Rowe) and B) each team is different, so maybe someone like Bingham, who has done well behind an LA Galaxy defence that gives up high XG chances, might struggle in a different set-up, as he may just be someone that likes to be kept busy all game, Instead of not getting much action behind a stingier defensive line.
It does give us an idea of how bad last years goalkeeping really was, as even an average MLS starter would have turned the Caps into a playoff team, as well as how poor Montreal’s keeping has been this season, as they may end up facing a similar fate, despite having some premier attacking talent. It also lends to the theory that paying for international, DP-level goalkeepers is a huge market inefficiency, as many of the top guys are cheap domestics, meaning you’d be better off using your 7-10 international spots on players elsewhere on the pitch, as well as doing the same with your DP and TAM spots.
So to give Marc Dos Santos credit, it’s one part of his roster-building he did nail. Even if Crepeau regresses towards the league average next year, the Caps should still be alright, because their offence should stand to be much better than they have been, and by the same token, the defence should be tighter as well. If Crepeau does regress but the Caps offence continues to struggle, as well as their defence continues to give up a relatively high XG, then it may be time to start asking questions of Dos Santos, but given his pedigree he has shown at all levels, the Caps should improve a lot, and if Crepeau stays close to his current level, watch out.
If you have any similar theories, any rebuttals to this analysis, any reasons that show why this could be largely flawed, feel free to comment, or tweet us, or interact in any way you can, I’d love to read and learn about other things people have to offer on this subject.
All goalkeeping stats provided by American Soccer Analysis. Check them out on twitter, as well as their website americansocceranalysis.com.