Now that we’re ten games into the new Premier League season I thought it was about time I checked in with the numbers to see if there’s anything interesting happening. I didn’t want to go on too long or look at every team, so I’ve just had a quick look at four clubs who caught my attention when looking at a few different metrics.
I explained the situation on the data front in the intro to ‘Looking For Players In Different Leagues’, in case you’re wondering where these numbers are from or think they seem a bit off.
Manchester City Are… Better?
Yeah, that’s right. After breaking almost every Premier League record out there last season, Manchester City are putting up even stronger numbers this time around.
There are a few caveats, the fact we’re only ten games in is one but that applies to all teams, while they’ve also had a pretty favourable opening ten games – which can be seen in the tweet below. With that being said, they’ve also had three of their five big away games against other teams in the top six, with the point at Anfield probably being the highlight as it’s an improvement on last season’s loss there.
‘#BHAFC #WHUFC & #HTFC have just had the toughest run of 9 fixtures they are likely to face this season. #AFCB #EFC have had the easiest they’ll face. #LFC has had the most difficult 9 game run of the top six. @InfoGolApp pic.twitter.com/swSPXucpAO
— mark taylor (@MarkTaylor0) October 23, 2018
Despite the favourable games, the fact they’ve managed to improve upon last years numbers is still hugely impressive. Their xG for has experienced a decent rise from 2.3 to 2.72 per game, while they’ve reduced their xG against slightly from 0.67 to 0.58 per game. Which obviously makes it no surprise to hear they’re taking more shots and conceding less shots each game too.
What’s interesting is that despite their pressure map looking more focused on the box (which can be seen in a piece by @TheM_L_G talking about City on StatsBomb) the opposition are actually having more passes per possession this season and City are even conceding slightly more counter attacks this season too. Their opposition passes per possession have gone from 3.2 to 4, which is a pretty sizable increase.
Looking at it on an individual game basis, it might be that City are being slightly more conservative for the away games against the other teams in the top six. Looking at the other seven games the average is 3.43, which is much closer to last season’s figure, then comparing the figures of the corresponding fixtures last year shows a similar pattern.
It was noted around the time that City were more conservative at Anfield this time around and the numbers support this. Last season Liverpool averaged 3.39 passes per possession, but this season they averaged 6.07. The games against Tottenham and Arsenal aren’t as different, Arsenal actually managed more passes per possession last season but with their new playing out from the back it makes sense to press them and force them into errors, while Tottenham went from 4.92 passes per possession to 5.23.
The Liverpool game seems to be the big outlier, but the good news for City is that if there is any change, it’s working. They haven’t conceded in any of their three big away games and I’d argue these kinds of figures are more stylistic than performance driven, with the improvement in the underlying xG numbers being the main performance indicator.
It’s going to be interesting to see whether or not City can keep these improved numbers up and come close to or even possibly eclipse what they did last season in the league, while also giving a better showing in the Champions League.
The Evolution Of Klopp’s Liverpool
When Jurgen Klopp joined Liverpool there was a promise of ‘heavy metal football’, referring to a high intensity without the ball and fast paced attacks. However, now entering his third full season at Anfield, the club seems to be transitioning from this style to a slightly more conservative one.
Talk of Liverpool not pressing as much has been discussed before, notably by @WillTGM over on StatsBomb, while @DistanceCovered constantly posts good analysis on Twitter, particularly about Liverpool, and recently wrote about their (change in) pressing over on AnfieldIndex, which is definitely worth reading and highlights their mid-block approach.
Liverpool seem to have gradually slowed down in their high intensity pressing game in most metrics since Klopp took over. The peak of the pressing Liverpool seems to be in 2016/17, which should come as no surprise. It was Klopp’s first full season in charge, so he would have had a summer to properly drill his style into the team, while it was also a year in which they had no European competition, giving a longer recovery time between games.
In 2016/17 Liverpool’s opponents averaged just 3.69 passes per possession (the smallest I’ve seen for a Premier League side so far is 3.2 from Manchester City in 2017/18), while Liverpool conceded just 1.89 counter attacks per game and only 9.5% of these ended in a shot (the league average for 2018/19 so far is 31.38%).
This high intensity football got Liverpool back into the Champions League, but given their problems against low-block sides and lack of squad depth when the fixtures piled up, it seemed a tough task to keep this up in 2017/18.
In 2017/18 Liverpool’s opponents had slightly more passes per possession with 4.07, while 29% of opposition counter attacks ended in a shot – despite conceding just 0.11 more counter attacks per game. While this doesn’t sound like a good thing, overall Liverpool had a higher xG for and a lower xG against, all while getting to the final of the Champions League.
It’s something I slightly touched on in my Window Shopping piece on Liverpool from back in January, but given Liverpool’s struggle against low-block sides it may not be a terrible idea to not press as high, inviting the opposition to bring the ball out of their defence, disrupting their low-block and pressing more in the midfield area – giving the side room to counter attack into.
The numbers support this idea, with Liverpool going from having 2 counter attacks per game game in 2016/17 to 3.82 in 2017/18. Doing so kills two birds with one stone, they can conserve energy by not pressing as intensely – instead choosing the right time to press – while also disrupting teams who play in a low-block.
In 2018/19 so far it seems Liverpool have moved in this direction even more. The opposition has had 5 passes per possession this season, with the league average being 5.24. The opposition are having more counter attacks again this season, with 3 per game, but now 26.7% of these end in a shot which is an improvement on last season and below league average.
On the ball, the benefit of the new, slightly more conservative Liverpool is that their counter attack numbers are up again this season. Liverpool have had 5.7 counter attacks per game, the most in the league, while making the 3rd most passes per possession.
Again, the change seems to be a positive move as Liverpool’s xG per game is up and xG against per game is down again. In 2016/17 Liverpool had an xGD of 0.9, 2017/18 it was 1.02 and for the first ten games of 2018/19 it’s 1.38.
Overall, the evolution of Liverpool under Klopp has been an interesting one to watch. Each year they seem to have distanced themselves a bit more from the high intensity football Klopp’s famous for, becoming a more conservative mid-block team, however, in doing so they’ve improved each year and look as though they should be in the title race this year.
Wolves Have Had A Strong Start, But Need To Start Offering More In Attack
Wolves have had a decent start to their Premier League campaign, getting 15 points from their first ten games. The foundation of the strong start has been a solid defence. Wolves are well drilled in their 3-4-3 shape and it’s paid dividends as they’ve conceded the 3rd lowest xG so far this season – only Manchester City and Liverpool have conceded less.
The West Midlands club are pretty happy to let the opposition have the ball, with Wolves allowing the 7th most passes per possession in the Premier League, before trying to win the ball back in midfield and progress it through the likes of Ruben Neves and Joao Moutinho. This makes it unsurprising that Neves has the 6th most passes into the final third p90 and Moutinho the 20th most passes into the final third p90 so far this season.
This is a great foundation to build on for Wolves but one area that requires improvement is in attack.
Using Understat numbers for a moment, Raul Jimenez seems to be the only Wolves attacker who’s been close to productive. It seems worrying that Jimenez is the only Wolves attacker who’s started regularly with more than 1 xG. After Jimenez their leaders in xG are wing-back Matt Doherty, Leo Bonatini (thanks to a couple of productive substitute appearances), centre-back Willy Boly and wing-back Jonny. It’s only then you get to Helder Costa and Diogo Jota with 0.77 xG and 0.76 xG respectively.
It feels as though the two players just behind the forward should be a lot more productive if you wish to play a counter-attacking 3-4-3. While it’s good that the wing-backs are chipping in, I think most Wolves fans were definitely hoping for more output from the likes of Costa and Jota. The good news is that Jota is only 21-years-old, Adama Traore and Ivan Cavaleiro are waiting in the wings and Helder Costa’s performances have been good, they just haven’t really resulted in goal contributions.
I’d say that the combination of Doherty and Costa on the right hand side might be the best feature of Wolves’ attack this season. It feels as though Doherty can be hard to track with his tendency to make runs inward rather than typical overlapping runs, while Costa too has the ability to go on the outside or inside. If Costa is wide, the full-back can follow him and a gap can be created between the full-back and centre-back, which Doherty tends to wonder into, leading to either a shot or a cutback.
The goal against Burnley shows this, Costa moves wide dragging the full-back with him and creating a space for Doherty to run into. When Doherty gets the ball Ben Mee moves over to close the angle, but then Wolves have a 2v2 in the box, plus Jota moving in free at the back of the box. In the end Doherty picks out Jimenez who finishes well.
There have been games where the opposition’s wingers have tracked Wolves wing-backs well, often making it incredibly difficult for Wolves to build how they want to – with the long switches to the flanks and the 2v1 out wide. One example of this being the home game vs Birmingham in the Championship last season. While Wolves then have the challenge of breaking down what’s effectively a six at the back formation, the positive is that it stifles the opposition attack, making it incredibly difficult to counter when the wingers have to start their runs from the edge of their own box.
I’d say I’m quite optimistic that the goal contribution will start to flow from the right side, however the left side feels like more of a work in progress. This isn’t surprising with Jonny being a new signing, and it’ll take a while before him and Jota start to get a good understanding, but I’m in two minds about it at the moment.
A good way of comparing is by looking at how Boly brings the ball out of defence last season and this season.
Below are a few clips from the home game against Fulham from last season. It’s worth pointing out I haven’t picked particularly good clips, with none of them leading to anything substantial, but you get an idea of the shape. Barry Douglas hugs the touchline and makes the overlapping run, while Jota tends to stay in the half space, dropping into pockets of space to help in the build up. If he receives the ball he can either turn inwards and link up with the forward, or outward and look for the overlapping run of Douglas.
Clips of Jonny moving more inside and Jota being wide from this season can be seen below.
Jota isn’t really like Helder Costa, who tends to operate a bit more like a typical winger, happy to receive the ball out wide, get to the byline and cut it back – giving Doherty the opportunity to move inward – Jota seems to be a bit more like a #10 and winger hybrid who works best in the half spaces. It’s frustrating because both Jonny and Jota are good players and (though it’s only a couple of clips each) I think Wolves’ build-up looks better in the second clip, it’s just that it doesn’t seem beneficial – particularly for Jota – to have them on the same flank.
As a Wolves fan I hope that it’s just a case of the pair getting used to each others games and that it’ll come good soon, although it’s interesting that Nuno opted for Adama Traore over Jota in the game against Brighton, someone who’s definitely more of a typical winger.
Overall, Wolves have had a strong start to the season and if they keep their 1.5 points per game average up until the end of the season it’d be a hugely positive return to the Premier League. Having a solid defence but an attack that needs a bit of work is a decent position to be in, one that I’d take over the reverse as a newly promoted club, but the hope is that the attack improves over the course of the season.
A Change Of Approach For Bournemouth?
Admittedly, I haven’t seen much of Bournemouth this season, but with how well they’re doing it’d feel silly to not give them a mention here. It’s also worth pointing out I tipped them to struggle when doing my transfer review at the start of the season, so feel free to ignore anything I say.
When mid-table clubs start well I’m always skeptical and assume that it’s just because they’re running hot and having a bit of luck at one or both ends of the pitch, however, Bournemouth have strong underlying numbers, suggesting their level of performance is somewhat sustainable. It is worth pointing out that they’ve had the easiest start to the season of all clubs (judging from the tweet in the Manchester City section), but they’ve made the most of this start and given themselves a great platform to build the rest of their season off.
The main difference that I could see in the numbers from this season and last is in counter attacks. Last season Bournemouth had 2.79 counter attacks per game and conceded 4.74 counter attacks per game. This season sees a reverse of this with Bournemouth now having 4.6 counter attacks per game (4th most in the league) and conceding 3.2 counter attacks per game.
With a fairly drastic change in both numbers it feels as though this can’t just be a coincidence, particularly as the numbers change even more when separating the into home and away. At home last season Bournemouth had 2.95 counter attacks per game but conceded 5.26 counter attacks per game. This has seen a big change this season, almost halving the number of counter attacks against to 2.8 per game, while having 4.2 of their own per game.
Away from home is a similar story, they’ve gone from 2.62 counter attacks for and 4.21 counter attacks against per game, to 5 counter attacks for and 3.6 counter attacks against per game.
It’s interesting that there’s been such a big change here, yet a lot of their other numbers are fairly similar – when looking at things like passes, possessions and shots. The only slight difference is that away from home their opponents have made more passes per possession (5.92 vs 5.3), but with just a handful of away games this could just be skewed by the fact they’ve played both Chelsea and Fulham away who tend to make a lot of passes per possession.
Looking at Understat numbers and specifically xG from fast attack shows further improvement from Bournemouth when it comes to counter attacks. Last season they had 0.63 shots per game from ‘fast’ attacks with an xG per shot of 0.15. This season they’ve doubled both the quantity and quality of shots from fast attacks, they’re now having 1.2 shots per game with an average xG of 0.3 per shot. Defensively the shot quantity has stayed around the same, dropping from 0.87 to 0.7 per game, but the quality has taken a big dip from 0.18 to 0.05.
Annoyingly I think the (non Understat) xG data I’m using includes penalties and given Bournemouth have had 5 this season it’s likely skewing it, but the change has seen Bournemouth go from having the 3rd lowest xG per shot to the highest xG per shot in the league while also having their xG per shot against go from the 4th highest to 3rd lowest.
I need to start watching more of them to see what’s changed in more detail, but it’s been a hugely encouraging start for Bournemouth. With four of their next six games being against Manchester United, Arsenal, Manchester City and Liverpool it’ll be interesting to see if they can keep up these good counter attacking and xG numbers, but even if they don’t they’ve taken the pressure off themselves for these games thanks to their strong start.