Just How Similar are Manchester City to Leeds United?
In this piece, Jon Mackenzie wonders what the similarities between Leeds and Manchester City can tell us about Leeds’ return to the Premier League…
A few weeks ago, when I was fretting about Leeds’ run-in at the end of the season, a good friend of mine sat me down and said to me, “Leeds are as dominant in their league as Manchester City are in theirs. You don’t need to worry. You’ve got this.” Of course, given that Manchester City ended up finishing 18 points behind Liverpool in the league, this hardly helped. But it got me thinking about the similarities between the two sides.
The more you put your mind to it, the more there are to find: Manchester City’s manager Pep Guardiola is often held up as an acolyte of Marcelo Bielsa; both are adherents of positional play; they are both purveyors of possession-based football; they even play a similar formation; both sides deserved to win the league per expected points; both teams dominate in terms of possession; the list goes on and on…
This graphic from Jonny Cooper’s season review shows how many matches Leeds recorded more than 50% possession in the last two seasons compared to other sides in the top two leagues.
These two graphics (the second from Ben Mayhew) show how Leeds finished 23 points below their expected points tally. City, on the other hand, finished a little below their tally but Liverpool wildly overperformed theirs to win the league.
As a result of these musings, I got thinking. Leeds are an interesting test case as a promoted side. There aren’t many sides who get out of the Championship by playing a style of football that sees a side dominating in the Premier League. Will this play in Leeds’ favour come next season? Or could it be their undoing? And just how deep do these comparisons go?
Structural Similarities
We’ve already mentioned the structural similarities between Leeds and Manchester City. While most people would call Leeds formation a 4–1–4–1 and City’s a 4–3–3, the crossovers are obvious: a back four with full backs who advance; a defensive midfielder; two “free” eights; a lone striker flanked by two attacking players.
These graphics from Twenty3 show the average positions of both sides from last season:
As you can see, there is a marked asymmetricality to Leed’s structure which isn’t there in City’s case. We talk a lot on this channel about how Leeds build up on one side of the pitch in a bid to create an isolation on the other side of the field. Manchester City don’t do this. Their build-up play happens on both sides of the field in a much more symmetrical manner.
This shows up better by adding in the passing networks to the picture:
Where City have two clusters of three players who help in the build-up phase on either side of the field, Leeds push their left central midfielder and their striker over to the right to help out there. As a result, the left winger is isolated and is largely supported by the left back. Note also how City’s full back’s get more advanced than Leeds (of which more later).
We’ve already mentioned how Bielsa and Guardiola are committed to what is known as “positional play”. In its simplest terms, positional play is a form of football tactic that seeks to use a team’s structure to create what are called “superiorities”: basically, situations where your players are better positioned than their opponents.
One way of doing this is to exploit the spaces that open up in standard formations. Take a look at this graphic from Nathan Clark’s article on the resurgence of back threes:
With a back four, there are five obvious weaknesses: the three spaces between the defenders and the two spaces between the full backs and the touchline.
For advocates of positional play, you want to exploit these five spaces and so, against a back four, you will look to get five players into each channel that is opened up by them. As Guardiola himself said, “I’m being converted into a manager of five forwards. It’s a curious phenomenon that I’ve learned here and I will always owe it to Germany.” This is what it looks like in practice:
Return to looking at the average position plots for Leeds and City this season and you can see that both sides are structured in such a way as to be able to fill these five channels in the attacking phase.
For City, the two clusters of three on either side of the striker make up seven potential players who can fill these five channels against a back four. Similarly, Leeds use the same seven players in an attacking moment to look for those five channels, albeit with their full backs often getting forward from a deeper position. The principles, though, are the same.
One final comment on structure. Another foible of Guardiola’s that we saw earlier in his time at Manchester City was the use of an inverted fullback. An inverted full back is simply a full back who pushes inside to help out with build up in the midfield areas.
Although this graphic shows both full backs inverting, Guardiola tended to invert Fabian Delph on the left (although Kyle Walker is inverted some of the time) with the remaining defenders forming a back three.
Interestingly, if you look at the average positions map for the second half of the season for Leeds United, you can see a similar inversion has taken place with Stuart Dallas often taking up a more central position to help out in the build-up phase:
Curiously, he was often pushed over as far as the right-hand side, leaving the left open with the defensive midfielder and the left winger tasked with the responsibility of keeping it solid defensively.
As to why Bielsa decided to do this in the second half of the season, I suspect it came down to Leeds’ dominance meaning that he felt a back four was largely wasted. By inverting Dallas, the benefit was shifted to another area and giving Leeds more of an attacking edge without too much defensive sacrifice.
The Attacking Phase
When it comes to the attacking phase, however, differences do start to emerge. Taking a look at the passing event data for both sides, you can see that the bulk of Manchester City’s passing play takes place in the wide midfield areas where for Leeds it occurs slightly different in the full back areas:
This would imply that Leeds do more of their slower build-up in their own half, beating the opposition press there before moving it a little more directly into the opponents half. (Of course, this could be because teams press Leeds higher than City’s opponents press them.) They then do a lot more passing work around their opponent's corner flags than City do.
Where Leeds go wide, City go central. Most of their build-up takes place across the pitch just inside the opponent’s half with the central spaces more important to their possession than the wide areas in their own half and deep into their opponent’s. This difference shows up well in Twenty3’s Passing Flow graphic:
As you can see, City funnel their play into the middle where Leeds go a little more direct down the flanks. This is hardly surprising when City’s two eights are David Silva and Kevin De Bruyne. Their creativity, then, comes from incisive passes in the central areas where Leeds’ comes through wide areas through their wingers, Jack Harrison and Helder Costa. If we compare the radars of Silva and De Bruyne with Mateusz Klich and Pablo Hernandez, you’ll notice a couple of things:
Firstly, you’ll notice how similar Pablo Hernandez is to the two City players in terms of his output. For his radar, we’ve limited the data to the second half of the season where he played as one of the central midfielders rather than as a wide player as he has been for much of his time under Marcelo Bielsa. Leeds are a better side with Hernandez on the field because he is a much more creative option than any of our other players. This doesn’t change when he’s in the middle compared to playing out wide.
For Mateusz Klich, though, you’ll see his radar is much more rounded than the two City players. Yes, his attacking output isn’t as high. But he’s expected to do more defensive work. Compare this with Adam Forshaw’s radar, who played alongside Klich in Leeds’ most dominant team: the XI who played the first seven games of the season before Forshaw picked up his injury:
In this initial approach, the two eights were less directly creative, tasked with facilitating the transition down the field into the wide areas. After Forshaw’s injury, Hernandez moved into a central area and Leeds used two out and out wingers and had a more creative option through the middle.
Compare the assist locations for the first and second half of the season and you’ll see that Leeds’ chances came a lot more centrally in the second half:
Notice also how much more compact the shot locations are in the second half of the second compared to the first.
For comparison, here is Manchester City’s Premier League assists map for the season:
Leeds’ assist map for the second half looks much more similar to Manchester City’s map than the assist map for the first half of the season.
In the attacking phase, Leeds and Manchester City are quite different in terms of the areas in which they build-up. However, in the second half of the season, the movement of Pablo Hernandez inside has meant that Leeds have become much more creative centrally. That notwithstanding, the bulk of their production comes from wider areas than from central areas, unlike Manchester City who are more dangerous from central areas.
Defensive Phase
Defensively, it’s a little hard to compare Leeds and Manchester City, not least because the defensive data (particularly pressing data) is much harder to come by.
As possession-based sides, City and Leeds are both vulnerable to counter-attacks in defensive transition. In principle, possession-based sides want to win the ball back quickly when they lose possession and then maintain possession as the best form of defence.
For Leeds, this manifests itself in a high pressing defence where they try to win back the ball as quickly as possible in the event of a turnover. Where they are unique is that they utilise a man-oriented press which sees each player go 1-vs-1 at certain points in the defensive phase.
One of the best ways of measuring how heavily a team press is to look at their Passes Per Defensive Action numbers. This number shows how many opposition passes are generally allowed before a defensive action is taken. The lower the number in this case, the better. This season, Leeds’ PPDA was 5.28 compared to City’s 8.17. This shows that City allow more passes from their opponents before they take a defensive action.
Leeds’ PPDA is the lowest figure across the two top flights of English football and is well represented in the following Statsbomb graphic:
Interestingly, only Barnsley (who you will remember pressed Leeds into submission late in the season) come close to matching Leeds in terms of PPDA. They produce a PPDA of 5.93 last season.
Compare this graphic with the comparative graphic for Manchester City:
As you can see, City press hard in the opponent’s half but not so much in their own half. This is the result of a tactical decision by the two managers respectively. In the event of a turnover, Pep Guardiola uses a high press to win the ball back. If that doesn’t happen, City will fall back into a zone-oriented marking system or, if they are worried about a good chance being created by the opposition, they may even commit a tactical foul.
Leeds, however, continue their press into their own half in their man-oriented system. This is considered an ‘old fashioned’ approach and one that is difficult to orchestrate simply because it requires an impressive level of coordination which, if it breaks down, could give up chances to the opponent quite easily. If it works, though, it allows them a remarkable frugality at the back.
This table from Jamie Kemp of LUFC Blog shows just how impressive Leeds’ defence has been this season. In each instance, Leeds are being compared to the last 15 sides promoted from the Championship and the then-current top three of West Bromwich Albion, Leeds and Brentford:
Despite the fact that, per goals conceded, Leeds were a little unlucky, their performance in preventing opposition chances is remarkable. Just over 5 box shots per game across the course of a season and just under 3 shots on target per game are impressive figures.
Because of this approach, Leeds have a deeper Average Defensive Action Distance to City. Measured by Statsbomb, Average Defensive Action Distance records the average distance from the goal line in metres at which a successful defensive action is made. City recorded an ADAD of 49.80 metres this season to Leeds’ 44.46m. However, Leeds’ figure will be slightly deeper because they make more pressing actions in their own half than City do.
Fundamental to both Leeds’ and City’s defensive systems is the defensive midfielder. For the last two seasons, Leeds have been able to rely on the services of Kalvin Phillips, a player reinvented by Marcelo Bielsa. During this time, City have largely used two players in the position: Fernandinho and Rodri.
Comparing Phillips to these two players is interesting. Firstly, looking at Fernandinho, you can see that Phillips is probably more of a defensively-minded player than the Brazilian:
At 2.27 defensive duels won per 90 minutes, Phillips is a very strong defensive player who overshadows both Fernandinho and Rodri. With 11.12 ball recoveries per 90 minutes, he’s also beating Fernandinho on that score too. The City midfielder, though, makes more passes than Phillips.
Rodri, who was brought in at the beginning of last season, is a clear upgrade on Fernandinho defensively:
Not only does his make more successful defensive duels per 90 minutes (1.51 to 1.18) he also pips Phillips to the ball recoveries completed stat (11.41 per 90 to 11.12). Add to that his better passes completed figure per 90 minutes than Fernandinho’s (78.02 to 66.35) and you can see why he was brought in by Pep Guardiola.
If you compare Phillips’ passing network map with his counterparts at City, you’ll also see why Leeds’ build-up phase takes place deeper than City’s:
(Notice that Fernandinho’s pass map is from 2018/19 season. He played more regularly at centre back this season.)
As you can see, in the defensive phase, Phillips is tasked with winning the ball back and then feeding it back to his full backs or his centre backs who can then cycle through to more advanced areas through the eights.
For City, though, the central defensive midfielder is responsible for feeding the ball directly into the eights after they have won it back, giving explanation for the two City defensive midfielders’ superior passing numbers.
Much like in the attacking phase, Leeds and City are markedly dissimilar in their approaches. Where both teams press high, Leeds also press deep, using an unusual man-oriented press. The defensive midfielder is pivotal in both systems. However, where Phillips is more of a defensive fulcrum, winning the ball back and feeding it into his defenders, City’s defensive midfielders are more progressive, involved in getting the ball to their central midfielders and performing less of a defensive role than Phillips plays.
Conclusion
With Leeds in the Premier League next season, we’re going to see an interesting variation of a possession-based positional system that aligns closely with Manchester City’s. However, there are systematic differences between these two teams. We have seen Marcelo Bielsa’s football working well at the Championship level and Pep Guardiola’s football is clearly successful in the Premier League. Does that guarantee that Leeds system will prove a success in England’s top division? Only time will tell. One thing is certain: it’s going to be exciting to watch either way.
All the data visualisations in this article except for the two xG plots and the scatter graph come courtest of Twenty3. All data is from Wyscout.
You can follow Jon Mackenzie on Twitter @Jon_Mackenzie.
If you enjoy this content and want to help us create more and better pieces, why not consider signing up to our Patreon page to get bonus material?
If you have any interest in contributing to this blog, get in touch with us @AllStatsArentWe.
