Statistics have told a clear story about Manchester United over the last couple of seasons. Their second-place finish last season was largely a mirage, a function of David De Gea playing an astounding season of football. They might have finished with 81 points and only conceded 28 goals, but it was simply unlikely to continue. That seems to have come to pass. Jose Mourinho’s team is currently sitting in eight place and they’ve already conceded 21 goals through just 12 games. The exact reversal that analytics predicted would come to pass. And yet, the story is more complicated.
Three things are all true at the same time. First, Manchester United’s numbers are significantly worse this year than they were last season. Second, Manchester United are no longer posting better results than their numbers indicate they should. Third, David De Gea is, despite that, still playing out of his gosh darn mind. Let’s take them one at a time.
United’s baseline defensive performance is deteriorating. Last season, United’s opponents notched 1.01 expected goals per match. That wasn’t very good by the usual standards of a team near the top of the table, but it wasn’t bad per se. It was fifth in the Premier League behind Manchester City, Liverpool, Tottenham and Chelsea, in that order. The problem was that United needed to be great defensively because their attack was also only fifth best in the league at 1.49 expected goals per match. They were solidly a fifth place team, but because they conceded so many fewer goals than expected, they finished second.
This season, they’re worse on both sides of the ball. Their attack has dropped down to 1.18 expected goals per match, tenth in the league, and their defense has dropped to allowing a scary 1.34 expected goals per match. There are twelve teams who have better defensive underlying numbers than United. Twelve! Last year United played like a fifth place team and finished second. This season, with their negative 0.17 expected goal differential per match they’re playing more like a 12th place team.
United’s numbers have gotten worse, but they’ve also stopped being able to outrun them. Last season United blew their expected goals numbers out of the water on both sides of the ball. They scored 67 non-penalty goals from 56.78 expected goals and conceded only 27 from 38.24. This year the, team is more or less in line with their numbers. They’ve conceded one more goal than they’ve scored, and their expected goal difference is a little over negative two.
How they’ve done it is interesting though. On the attacking side of the ball United, just like last season are on pace to better their numbers. They’ve scored more than their expected goals.
They’ve also conceded more than expected goals thinks they “should” have.
This looks straightforwardly like a team playing an open brand of football, and executing it at a not particularly effective level. That’s startling for Mourinho the dull, but it’s less surprising given the talent he has at his disposal. United have crummy defenders and good attackers, so a team basically playing at the level expected goals expects, while also being skewed towards attack and away from defense makes sense.
This also looks like a team without any noticeable standout goalkeeping. And that’s where things get weird. Because the numbers also show David De Gea once again having an amazing season. Last year it was easy to see De Gea’s contributions. The team faced 38.24 expected goals, and the shots that made it on target to De Gea’s net had a roughly similar value. Their post-shot expected goal value was 38.76. In other words, De Gea’s dominance last year was easily recognizable. The 12.76 goals above average he saved were directly reflected in United’s defensive performance against expected goals.
This year it’s way trickier. While United’s opponents have 14.10 expected goals overall, the set of shots that have reached De Gea have been considerably tougher. Post-shot expected goals shows De Gea as having faced shots worth 24.09 expected goals.
Ok, so what exactly is going on here? In effect what this shot chart is saying is that the while on average the shots United are conceding will lead to a little over 16 goals, in practice opponents have caught the ball quite a bit better than average, leaving De Gea to deal with shots on target that will be scored a little over 24 times. Opponents are hitting the ball way above average, and De Gea is standing on his head just to keep United close to where expected goals thinks they should be.
If we look at this in terms of expected save percentage and actual save percentage it becomes clear that De Gea is basically maintaining last year’s level. Last season, our post-shot expected goal model gave De Gea an expected save percentage 73.2% and he clocked in at 82.4% giving him a goals saved above average percentage (GSAA%) of 9.2%. This season, post-shot xG has him with an expected save percentage of only 64.5% and an actual save percentage of 71.4% for a GSAA% of 6.9%. Despite a severe drop in save percentage De Gea iss still having an excellent, if slightly less superb season than last year. It’s just masked by the fact he’s facing shots that are a lot more difficult to save than a regular xG model predicts they would be on average.
It’s important to be cautious when drawing conclusions about exactly why the xG discrepancy exists. It’s possible that it’s just noise, and that through 12 games United’s opponents have been catching the ball extra pure. Maybe United have been getting really unlucky, and De Gea’s magic has stemmed the bleeding so that it seems like they’re only getting a little unlucky. Football has a lot of moving parts, and the reason expected goals works so well is that even one part being very out of whack tends to be mitigated by everything else. So, De Gea playing great while United get unlucky so that from a distance things look mostly normal, is a reasonable possibility.
It’s also possible this is a further reflection of United’s poor defense. Maybe teams are teeing off on De Gea because United’s shoddy backline, and complete lack of a midfield, is letting them. That too makes some degree of sense. Expected goals against is a fairly accurate predictor of a team’s defensive performance so it wouldn’t be unreasonable to think that what it’s reflecting right now is a unit that consists of poor defenders and a great keeper awkwardly averaging each other out.
It’s also possible that this is a tactical effect. If United are playing more openly this season, and it seems like they are, then maybe that openness is giving opponents a chance to get better than average shots off. It would make sense that above average sets of shots, offsetting each other on the attacking and defending end would lead to a team ending up where they should be according to expected goals while also having exaggerated post-shot expected goals.
The reality is that we simply don’t know enough yet to advance any of these theories with any degree of certitude. What we do know is that our post-shot expected goal numbers show a goalkeeper in De Gea who is still playing his tail off, even as his team underperforms opponent’s expected goals. That in turn allows us to say pretty concretely that United’s disastrous plummet back to earth from their surprisingly lofty finish last year isn’t driven by De Gea’s return to earth. It’s everything else that’s collapsing.
Header image courtesy of the Press Association