clock menu more-arrow no yes

Filed under:

Chelsea have been underperforming xG for the last 2.5 seasons

New, comments

New issues are old issues

Chelsea v Manchester United - Premier League Photo by Clive Rose/Getty Images

Chelsea’s offensive foibles under Thomas Tuchel have come under increasing scrutiny, even as the new head coach remains unbeaten with six wins from his nine matches in charge. But while Chelsea have conceded just two goals during that span, a truly impressive number, we have scored just ten goals, which is also quite impressive, and not in a good way.

Or maybe not, if we have been paying attention over the past couple years.

Frank Lampard had lamented Chelsea’s inability to finish chances throughout his 18-month tenure (even when the team were leaking goals at near-record pace), and before then, Maurizio Sarri’s Chelsea were largely dependent on Eden Hazard conjuring something out of thin air. And even though Hazard pretty much had a career-best season as his farewell gift, Chelsea still underperformed expected goals, as this lovely chart from data analyst Mark Taylor shows.

Taylor compares the 15 always-present Premier League teams since the start of the 2018-19 season by graphing the difference (using a 10-game rolling average) between goal differential and expected-goal differential, in effect showing how good a given team’s finishing was relative to the (expected) average. Blue is good, yellow is bad.

You need a magnifying glass to see the blue on Chelsea’s chart.

Most of the teams performing best in this metric should come as no surprise. Overperforming xG is a hallmark of “good” teams, after all.

Liverpool have consistently overperformed the average on their way to a couple historic seasons and a long-overdue Premier League title. Manchester City have also overperformed, albeit a bit less spectacularly, showing just how much of a well-oiled machine they continue to be. Spurs, powered by Kane & Son & Co are up there as well, even post-Pochettino, while Manchester United have had a bit of a rollercoaster ride.

On the flipside, on the “bad” teams side, we find Chelsea. Perhaps no other team have underperformed as consistently as we have. Brighton come close. Wolves have flatlined a long time ago. Southampton and Everton have had a couple massive dips, but found their way back into the blue. Even Burnley have had notable periods of recovering to the mean. But not Chelsea, not except for a couple spikes of good finishing form.

Chelsea v Luton Town: The Emirates FA Cup Fourth Round Photo by Catherine Ivill/Getty Images

Of course, xG is not the end-all, be-all of football statistics, and it’s just another metric by which we can look back and evaluate the success of a team. But it fits with what we are and have been seeing on the pitch. It’s not revealing anything new, but it is putting it into obvious and visible form. We’ve not been as great at converting chances as we have been in creating them — which if you consider that we created more shots than all but one team last season, does explain some of our troubles during this time.

Shot-Creating Actions 2019-20 Premier League
fbref

(Obviously, creating lots of half-chances can accumulate similar xG sums as creating just a few great chances, but that’s a breakdown beyond the scope of this cursory glance. But it certainly should be noted that while Chelsea created the second most “chances”/shots in the league, we were only about mid-table in terms of shot-conversion metrics.)

Non-penalty goals vs. xG 2019-20 Premier League
fbref

So, what’s the solution?

Sarri had Hazard to bail him out. Lampard sacrificed defensive structure for more chance creation. Tuchel has done basically the opposite. They’ve all searched for the mythical balance in the meantime, and found none ... yet. Can Tuchel do so in the middle of a season? It’s unlikely, which would explain why he’s seemingly set his sights on grinding out a top-four finish — and defense-first approaches can pay massive dividends in Cup and knockout competitions, too.

The solution in the long-term may be a bit more complicated than that, and may require tactical and personnel changes. Making the correct calls, especially in the latter regard, will be if utmost importance.