Discourse on Chelsea in Fantasy circles seems to be a recycling of the usual “Chelsea defence is bad” takes, and so here is a perspective I have not really seen on FFScout/podcast, till now – a case for Kurt Zouma (£5.0m). Mostly through looking into statistical precedents of Chelsea’s weird 2019/20 season, where the team with the worst defensive away record had the third best defence at home.
The lack of trust in Chelsea looks justified: they conceded (a really bad) 54 goals last season, with 38 goals away (equal worst, level with Norwich) and only kept one clean sheet away (Aston Villa had zero, everyone else had three or more). The only case seems to be for Ben Chilwell (£5.6m) given his attacking potential, or perhaps Reece James (£5.1m) if he can gain enough form to be undroppable. The centre-backs are seemingly not to be considered, given Chelsea’s rotation and uncertainty.
While this is all true, Chelsea’s 2019/20 defence is a lot more intriguing on closer inspection. Chelsea had a solid home record last season: they only conceded 16 goals at home (third) – just Sheffield United (15 goals) and Man City (13 goals) are ahead of them. And Chelsea kept eight clean sheets at home, this is only bettered by Man City (10).
While decent, this is a flimsy case on its own, beyond an expensive rotation. A team that kept seven clean sheets in 12 ‘easier’ home games (everyone from ninth to 20th), but only two clean sheets in the other 26 games.
The picture gets better looking at Expected Goals Against (xGA) (from FBref.com). Chelsea had an unbelievable league-high xGA of 10.7. That is an underperformance of five goals, as explained later mostly due to Kepa Arrizabalaga (£4.9m). The next best was Manchester City, some distance away with an xGA of 15.2. Chelsea’s story is similar for away games: 38 goals conceded for an xGA of 27.3 (10th).
With the signing of Edouard Mendy (£5.0m), this could massively improve. It’s hard to overstate the value of a good goalkeeper, just look at Liverpool with Adrian.
Chelsea ended the season with four clean sheets in six home games. With an xGA of 2.1 in these games, they conceded two goals (a 30-yard De Bruyne free-kick and a freak own goal v Tottenham). It’s hard to manage the dissonance where a team finished the season conceding two goals in six home games, but 16 goals in six away games. But what does that bode for this season?
In recent times, there are only two ‘big six’ teams who had similar records. This would be Liverpool in 2017/18, and Arsenal in 2018/19. I’ll look into both of them to see if they offer any insight (normally I would focus on Liverpool only to save on length, but focusing on two teams helps contextualize it).
Liverpool 2017/18 example
There is a perception that Liverpool were hopeless before Virgil van Dijk (£6.5m), when no lead seemed safe. Statistically, however, their home record was really good. In 2017/18, Liverpool only conceded 10 goals at Anfield, keeping 12 clean sheets. Just Man Utd conceded fewer goals at home, and no-one kept as many clean sheets. Factoring in that they conceded five goals (50%) to Man City and Tottenham alone (four of them after the 80th minute), it was seriously impressive.
With summer additions of a fully-settled Van Djik, Fabinho (£5.5m) and Alisson (£6.0m), Liverpool achieved exactly the same numbers at home in 2018/19 (10 goals conc. home, 12 clean sheets). Curiously enough, Liverpool’s xGA rose slightly from 11.6 to 13.9, but the addition of Alisson kept the actual count at 10. The big jump was away, something to watch at Chelsea this season.
In 2017/18, Liverpool conceded 28 goals away (21.7 xGA) and kept five clean sheets. Then in 2018/19 they improved massively (12 goals conc. away, nine clean sheets). Their xGA was also reduced to 16.7, with Alisson’s impact visible here too.
While Liverpool 2017/18’s away defence wasn’t as bad as Chelsea 2019/20’s, this could perhaps serve as an example that an improvement is possible. Liverpool massively strengthened, kept their home defence good, and made big strides away. While Chelsea’s signings are not the level of Liverpool’s, they are replacing much worse defenders than Liverpool had.
Arsenal 2018/19 example
The other example is Arsenal, from Emery’s first season to his second.
In 2018/19, Arsenal conceded 16 goals at home (fourth), keeping seven clean sheets. Yet they conceded 35 goals away (14th), keeping only one clean sheet. While Arsenal conceded just 16 goals, their xGA was 23.2 (10th). Almost every other team with a similar xGA conceded around 25 goals. Their away defence was not much better: 35 goals conceded with an xGA of 31.3. To contrast, the best home defences in 2018/19 (Chelsea, Man City and Liverpool) had the best xGA as well.
In 2019/20, Arsenal conceded more goals at home but fewer away (an identical concession of 24 goals and keeping five clean sheets both home and away). A difference of 19 goals in their home-away GA became zero: their home xGA spiked to 30.7 (18th), though they outperformed it again (24 goals), but away from home was a slightly better story as Arsenal’s xGA improved to 25.8 (24 goals).
Correlation or No Correlation?
This casts doubt on the correlation of Chelsea and Liverpool. After all, Arsenal’s away defence improved but their home defence worsened right? Overall, Arsenal’s goals conceded went from 51 to 48 compared to Liverpool’s 38 to 22. Is that not likely with Chelsea?
This can be explained through FBRef, which uses Statsbomb’s post-shot xG (PSxG) model. To explain briefly, post-shot xG takes into account the quality of a shot. So while a shot from the same position should have the same xG, a shot in the top-corner is likelier to score than a shot at the middle. Thus, the top corner shot will have a higher PSxG.
It’s well documented how Arsenal have been consistently outshot by opponents for a while (just look at Leno’s total number of saves). Whereas Chelsea only conceded 115 shots on target last season (third), Arsenal conceded 189 (18th). However, Chelsea’s PSxGA/shot is 0.35 (20th), whereas Arsenal’s is 0.27 (first). Thus, while Chelsea concede far fewer chances, they are of a higher quality.
This does not excuse Kepa. Chelsea’s post-shot xGA difference (the difference between PSxG and actual goals) is a league-bottom -9.6. To give context, only two other teams were below -3.0 (Aston Villa and Watford). Meanwhile, Arsenal had a high PSxG difference at +10.0, best in the league.
This indicates that it would be folly to compare them to Chelsea. Arsenal concede a lot of shots, reducing chance quality and relying on Leno to make a lot of saves. Chelsea give away few chances, which tend to be of a relatively higher quality than the norm. Chelsea have very good xGA numbers overall, while having statistically the worst goalkeeper in the league.
With better defenders and a new goalkeeper, Chelsea’s version lines up a lot closer to Liverpool under Karius/Mignolet, rather than Arsenal under Martinez/Leno. Liverpool conceded a similar number of shots on target to Chelsea throughout, and jumped from PSxG/shot of 0.32 (14th best) to 0.28 (best in the league) after the arrival of Van Dijk, while Alisson meant they began overperforming their xGA.
Prospects
Good home defensive records can also become good away defensive records with better signings and good underlying numbers.
While Chelsea have conceded six goals already this season, at least four of these could be tied to clear individual errors from Kepa Arrizabalaga and Marcos Alonso, rather than major tactical failings. And while this does not excuse anything, after all Liverpool have made an individual error leading to a goal in almost every game (every game if Jorginho had scored his penalty), the Chelsea duo are not long-term players.
With Mendy, Chilwell and Thiago Silva (£5.5m) coming in, there is hope Chelsea defenders will stop passing straight to opposition attackers. As the final nail in Kepa’s coffin, Chelsea have recorded a better PSxGA/shot this season (while still conceding few shots on target) but the difference between PSxGA and actual goals conceded (-2.6) is still bottom of the league.
There is serious potential for Chelsea’s defence to have a monster season, particularly once the forwards click and are able to press effectively and pin defences back.
And so we finally arrive at Zouma, the player I predict to benefit in FPL.
The Frenchman established himself as first choice after the restart. Bar the odd exclusion – he was left out of the 18-man squad against West Brom (which strengthened his case) – he has started every game. What’s interesting is that while Silva played as the LCB with Andreas Christensen (£4.9m) and Fikayo Tomori (£4.9m), he returned to RCB in the Crystal Palace game with Zouma staying at LCB.
This seems to indicate that perhaps Lampard is seeing Zouma as his long-term LCB this season. He can speak both French and English, has had solid partnerships with all centre-backs, and is the best header in the squad. While he has the odd mistake in him and could use an organizer like Thiago Silva, he has demonstrated the most consistency and has the best skillset in terms of 1v1 defensive ability, intelligence and physical tools. With goals already from James and Chilwell crosses, and Hakim Ziyech (£7.9m) crosses still to come, Zouma should also bag more goals this season.
He offers a £0.6m value difference on Chilwell (so could fit alongside a premium defender), and offers more security than Reece James. With a relatively easier Champions League draw, Zouma may well be kept for all league games. Frank Lampard said after the game against Palace:
I want to find some sort of stability with the back four, of course I do. Those two (Zouma and Silva) showed how well they can do it, but that needs to be sustained.
– Frank Lampard
As of now, it is Zouma’s place to lose.
Southampton and Man Utd can be tricky, if not must-bench games, but they are followed by Burnley, Sheffield United and Newcastle. This could be an ideal time to jump on – or just keep Chelsea defence on your watchlist and see if they can start to keep away clean sheets.
3 years, 6 months ago
Really interesting take and excellent research, thank you Kroos Kontrol.
Why do you think Zouma was dropped after the Liverpool game and before WBA?
And, what changed between Emery's first season and his second, or is the point that nothing dramatically changed?
My final thought was - I wonder if Kante makes a difference to Chelsea's PSxGA/shot? Shots from distance have a lower xG and PSxG, therefore if you are not conceding many shots from distance it might boost your PSxGA/shot. I haven't researched this, it's a just a thought.
However, I completely buy what your are saying about the goalkeeping situation and I agree that better defenders should lower PSxGA/shot regardless of the amount of shots allowed from distance (I know the StatsBomb/FBRef model includes defender positioning). I currently own James myself, I am holding for now because I want to see how Chelsea line-up after the IB, and I think Southampton is a good fixture for them. I hadn't considered Zouma but I definitely will now.