Eighteen months ago, Chelsea were coming off their worst finish in the Premier League table since 1996. The season ahead offered no Champions League football – a competition in which the Blues have become a mainstay – and the uncertainty of a new manager attempting to right the ship. Everyone knows how that turned out. Chelsea had the title locked up weeks before the season officially ended. With fans expecting a repeat campaign, hopes were high for potentially a third title in four seasons. However, going into this weekend’s match against Southampton, Chelsea sat in third, 11 points back of leaders Manchester City. So, what happened? Have Chelsea taken a slight step back or have other clubs improved around them?
It’s very easy to say Manchester City got better and it’s true, they did. I say that with confidence, and would point to the table should you need proof. Manchester United improved as well. And also… so did Chelsea. Alvaro Morata has found the net nine times in 14 appearances, and though they missed N’Golo Kante’s presence early on, he’s back in the fold now to man the midfield. Making qualitative observations like this is pretty easy. I’m much more interested in whether there’s statistical evidence of their decline, or of other clubs just leaving them in the dust. So I turned to Understat.com, and most importantly, I turned to Expected Goals (xG).
Now I don’t want to go off defending advanced metrics and making impassioned appeals in attempts to get you to believe the xG stat. What I will say is this; Expected Goals are a generally good barometer to determine team performance. It is one useful piece of a large puzzle. Furthermore, discrepancies between xG and what our eyes tell us when we watch the matches are worth investigating further. A great example of this is this graphic from Michael Caley (@Caley_graphics on Twitter), showing the xG for each shot during the recent Champions League match with Atletico Madrid.
The game ended in a 1-1 draw, with Chelsea’s only score coming by way of a Stefan Savic own goal. The bigger the box on this chart, the greater the expectation that that shot turned into a goal. In this match, Chelsea struggled to finish, even though the Blues had a number of high quality chances. The data supports that claim, and shows that despite a draw on the score sheet, Chelsea played a good match and can expect to win more often than not.
So, looking at xG figures from their Premier League winning season in 2016-17, we can get a feel for how Chelsea were winning games and their quality in comparison to other clubs. Despite finishing atop the table, they finished all the way back in fifth in xG. Not what you’d expect from the league leaders, but as any Chelsea fan will tell you, last year’s club earned their title by stifling their opponents, not outshooting them. Chelsea led the EPL with the lowest expected goals against (xGA) last season. Not only that, but it was the lowest xGA since Understat began charting xG in 2014. The back line was the key to their success and it eventually earned them a trophy.
The stats for this season, however, paint a different picture. Chelsea sit sixth in xG, roughly the same spot as last season, and indicative of the conservative style Antonio Conte prefers to play. The problem, however, has been their statistical drop-off defensively. Instead of pacing the league in xGA, they’ve dropped into third, behind Spurs and league leaders Manchester City. On its own, you can probably explain this away as small sample size, statistical randomness, or some other confluence of factors. Unfortunately, at the top of the Premier League, where the margins are so thin between the top six clubs, that small drop-off has much larger repercussions. We’re not talking about a team falling off a cliff defensively and suddenly being in the relegation zone, mind you. But for a title contender, can a couple more goals let in here or there be the difference between an 11-point deficit in the table, and, say, a five-point deficit? Absolutely.
One other factor to consider is whether Manchester City, the darlings of the league right now, just flat out moved ahead of them in all facets. An improvement of this magnitude by City is not reflected in the data. It should come as no surprise to anyone to know Manchester City lead the EPL in xG by a wide margin. They also have the lowest xGA so far as well. Last season, they were actually runners-up to Chelsea in that stat, and were only behind by two xG despite finishing third in the table. It seems that spending £200 million on fullbacks and a keeper, and getting them to buy into Pep, is enough to move ahead slightly in that regard. But again, they have not changed drastically in the way Chelsea have in either stat. They’re very much – statistically speaking – the same club.
What does this all mean? Well, Chelsea are a club that mounted their last title run on the back of one of the best defences in recent history, combined with an effective but not extraordinary attack. Now that they’ve taken a slight step back in what they do best, it’s been enough to drop them down in the table, and put them in a precarious position in regards to winning the league, not even halfway through the season. Yes, Manchester City have improved, but statistically, only slightly overall. Not enough to account for the massive lead they’ve opened up.
Chelsea still view themselves as title contenders, and it’s still early enough in the season that City could falter. If the Blues are going to make a run, though, they’ll need to recapture their form in the back to do so. Tighten up the back and the climb to the top should be on.