Uncharted Territory
Putting Man City's Current Run of Games in Context
On a recent Double Pivot podcast, Michael Caley asked a question which aligned with something I had already started looking at: just how bad this current run of City games is, viewed in context with how the team has performed under Guardiola in the past. I think analytically inclined folks tend to assume that most narratives run too hot or too cold, and the answer invariably lies in the squishy middle ground. It’s rare that you ask a question about something being notably good or unusually bad and discover, “holy crap, this is actually historically weird!”
We know Man City have been poor of late, though debate remains as to how recent of a phenomenon it is - the last seven winless games, the 14 games since Rodri got injured, or the entire 2024-25 campaign? Given the fact that this poor form comes after opening the season with zero losses in 13 games, without digging into data too much (or watching the games) it would be a reasonable hypothesis to suggest this run could just be a case of running cold and getting footballed. '“Good teams sometimes lose games”, you might say. “Pep didn’t care about the Carabao Cup and they’ll progress in Europe anyway, so you can discount some of these losses”, others might chime in. City have suffered before, and Pep always find an adjustment to snap them out of it. Invert this piece, move that one deeper, add MORE control.
Looking at the data is pretty striking though. Below we have plotted the cumulative expected goal difference over all seven game periods since Pep joined the club (due to availability of xG data at fbref we’re only including Premier League and Champions League games, and don’t have data for his debut season at the Etihad). Why seven games? Because Caley mentioned seven as a sensible sample and he’s 100x smarter than I am.
The last seven games include this current run of winless games, plus the 1-0 win over Southampton in the league. Over this period, they have notched 13.7 xG (1.96 per game - seems fine!) but surrendered an astonishing 15.5 xGC (2.2 per game). For context, through 13 games no team in the Premier League has a worse per game defensive record than that (Southampton rank 20th for the season with 2.1 xGC per game). Viewed in context of City’s past results under Pep makes this total even more striking:
We don’t have data for his first season - which included a slow start for Guardiola’s City - but after that, this is the first time under Guardiola that City have posted a negative expected goal difference over a seven-game period, with the last seven games giving City a total of -1.8. The prior down points where notable for City dropping below +4.0 xGD, and even then, they have tended to have important mitigating circumstances.
2017-18 - Mid-Feb to early April. City were romping their way to the title and a 100-point season and this “bad” stretch of games was largely driven by a low-key Champions League second leg against Basel - when the tie was already decided after a 4-0 first leg - and a resounding 3-0 defeat to Liverpool in the Champions League quarter final. It’s notable that City still notched more xG than their opponents in all but two of these games (and on in which they didn’t they still won 4-0).
2020-21 - Start and end of the season. This was possibly City’s weakest title winning team, thought they’d still cantor to victory over Solksjaer’s United. This low xGD total was due to a wacky couple of games to start the season, with City losing 5-2 against Leicester (0.9 to 2.9 xG) and drawing 1-1 with Leeds (1.2 to 2.4 xG). Pundits at the time highlighted a “growing vulnerability” in Guardiola’s side, despite conceding an average of just 1.0 xGC over these “bad” seven games and going on to concede more than a single goal once in the next thirty-one games.
The other down period from that season was the end of the campaign when City fielded weakened sides with the title wrapped up, and a couple of close contests with PSG in the Champions League semi-final.
2022-23 - End of the season. Once again, this period included a couple of dead rubber games after City had wrapped up (Arsenal thrown away?) the Premier League, along with two very tight Champions League games away to Real Madrid in the semi-final and then the final victory over Inter.
In summary then, we have four periods over three seasons where City posted an cumulative xGD which came close - but still materially eclipsed the current run - and in all but one case (the start of the 2020-21 campaign) there are very specific circumstances to explain a large portion of the lower performance, namely difficult final stage Champions League games, or Premier League games that didn’t matter after City had won the league and Jack Grealish was already, ahem, celebrating.
There are potentially important factors this season too, primarily Rodri’s injury, and arguably that being paired with De Bruyne and Kovacic’s limited minutes. However, the former isn’t coming back any time soon, and one wouldn’t bet that peak De Bruyne is going to show up and play 30+ 90s over the remainder of the calendar.
As Mike noted on the podcast, Guardiola gets the most benefit of doubt of all managers to right a ship of course, and the prevailing wisdom is that he will come up with a novel solution to the existing problem.
"His problem now might be that Rodri is out but we all know that he will come up with a solution and their run of form will go again.” (Arne Slot)
We’ll see if that’s the case, but the data shows that this is a new depth to come back from, and mere regression and a few lucky bounces aren’t going to be enough to turn things around.