As Barcelona’s two managers for the 2019/20 season, Ernesto Valverde and Quique Setién, failed to revert the downward trend of the team, we make use of multiple statistics to analyse whose team performed better.
Coming on the back of a disastrous season all around, in every single competition played, this article will try to take a somewhat in-depth look at the underlying statistical ebbs and flows that defined 2019/20 for FC Barcelona.
One of the biggest defining features of the season was, perhaps, the mid-season sacking of the incumbent manager Ernesto Valverde and appointment of a new coach in Quique Setién. As such, it makes a lot of sense to discuss the intricacies of the season by putting the stats of the team under each manager side-by-side.
The sub-topics chosen to discuss here have been motivated the recent series of interactive league reviews released by Statsperform – previously Opta. You can find their La Liga review here. First up, shots, goals and xG.
Shots attempted and conceded, goals scored and conceded
Let’s begin our tour by looking at the shot heat maps and goal location maps under each manager. The data is taken from Understat, and own goals are not considered.
The first important takeaway: Barcelona massively overperformed under Valverde, with a G–xG of around 15, and then spectacularly turned everything around to underperform in the second half, with a G-xG of around -4. On the other hand, Valverde’s Barça conceded more than expected, while Setién’s conceded less than expected.
The shooting and scoring zones remained roughly the same, and perhaps encouragingly, very central and corresponding to very high xG zones. Somewhat worrying is the fact that a lot of shots are conceded from inside the box. Good defensive teams tend to force their opponents to take pot shots from far away, i.e., low xG zones. Barcelona’s defensive endeavours fell flat this season to put it politely, given the number of goals conceded for a top team, and the really dangerous zones the bulk of the shots came from. Both managers were at fault here.
Now, there has been a nagging question in the minds of many: given the xG underperformance, was Setién unlucky? Well, probably, but not by as much as you might think. There are ways to quantify ‘luck’ – one of the ideas is Expected Points (Understat provides the data again). The idea behind expected points is: given the xG of the individual shots, you can simulate match outcomes many many times, and infer the probability of winning, losing or drawing based on those simulated outcomes. fcpython.com provides a nice and gentle intro to this idea if you want to go ahead and do the calculations. For now, let’s just take the Understat numbers directly.
Under Valverde, Barcelona had 40 pts, and so did Real Madrid. Barça were ahead on a head-to-head basis by virtue of the tie at the Camp Nou. Under Quique Setién, the reality is, well, Barcelona lost, but do the xPts tell a different story? Sadly enough, not really. Filtering out by date to get the xPts for the last 19 matches, and it’s 38.53 to 38.33 in Madrid’s favour.
So, even if Setién had gotten lucky with the finishing, it would have barely taken Barcelona tied for the top position. Now, Quique Setién’s team performed better in terms of expected points as compared to Valverde’s, but not enough to have won the league directly.
Defensive activities: pressure
Moving on to the next topic: defensive activities and pressure. Let’s first take a look at pressure percentages by thirds of the pitch, in all games, under both managers.
It’s pretty evident that Ernesto Valverde’s Barcelona focused their pressing in the middle third rather than higher up in the pitch – makes sense since Lionel Messi and Luis Suárez didn’t press a lot.
The story remained the same under Setién. However, one thing that does need to be pointed out here – compared to the rest of La Liga sans Getafe – is that Barça were kings of pressing. Barcelona had the second-best PPDA according to Statsperform. That’s still nowhere near enough the other European elites, but still, not too shabby. Let’s compare the pressing under the two managers now.
The raw pressure numbers were higher under Valverde, but Barça had more success under Setién, with a jump in successful pressing percentage from about 29 to 33. Now, before proceeding, I do want to make the point that across all leagues, a general trend that can be noticed is: the higher the possession percentage a team commands, comparatively fewer are the pressure numbers. Of course, there are fluctuations, especially on a game-by-game basis. But on average, if a team keeps the ball better, they need to press less.
So, one way to make a truer comparison of the pressure numbers is to possession-adjust them, as Statsbomb suggests. Possession-adjustment brings the numbers much closer – roughly around 200 for both:
What this reflects: since Barça had more possession under Setién, they didn’t have to press as much. In my opinion, possession-adjustment makes a comparison of off-the-ball activities a bit fairer. Now, to add to this, successful pressure is good, but not the only way to stop the opponent. Their passing rhythm can be disrupted without necessarily winning the ball back. Here is how the opponents’ passing accuracies looked like: Setién’s Barca disrupted the opposition marginally better.
To get an idea about the intensity of pressing, we can take a look at PPDA. It is a proxy for how frequently Barca disrupted opposition’s passes. Here is the definition from Wyscout:
“PPDA = Number of Passes made by Attacking Team (opponent) / Number of Defensive Actions
The PPDA metric is calculated by dividing the number of passes allowed by the defending team by the total number of defensive actions. Defensive Actions are: possession-winning duels, tackles, interceptions, fouls.
Both values – passes made and defensive actions – will be calculated in opponent’s final 60% of the pitch”
On this occasion, the numbers are taken from Understat. Smaller the number, the better is the pressing / defending intensity, as it signifies the opposition being able to make fewer passes before Barça applies a defensive action.
Once again, pretty comparable numbers here under both coaches. We can also take a look at the time-variation of pressures applied by Barcelona. This data comes from Statsbomb‘ open data set on Lionel Messi. Messi played in a total of 33 league games this season, so this is only a partial set of the full data that is publicly available. Of these 33 matches, 14 were under Valverde while 19 were under Setién. So, to keep comparisons fair, the numbers are converted to per game.
Overall, the intensities seemed to drop as the game wore on, which is natural, but for Barcelona, it was exacerbated by the aging squad and the very disturbing reluctance in taking advantage of substitutions, especially by Setién.
It is fair to say that Barcelona pressed better under Quique Setién. What about the end product of pressing though? How often did turnovers created from pressing translate into something actually constructive?
Presented here are starting locations of “high-turnover sequences” forced by Barcelona under each manager. Once again, there were 14 games under Valverde in the public dataset as compared to 19 under Quique Setién. The takeaway is that the ratio of shots to the total number of such sequences was better under Setién.
Progressive Passing and Touches
First up, a touch zone comparison:
Now, there are some interesting trends here. Despite having more possession and touches, Barcelona didn’t do much in the opposition penalty area under Setién. And while the touches in the final 3rd per 90 increases, as a percentage of total touches it goes down marginally. This has been routinely criticised by several fans: holding on to possession aimlessly. Meandering on and on through the middle 3rd was a massive problem for Barcelona.
Looking at the passes, completed passes per 90 and completion percentage both see an uptick, but the passes to the final 3rd, the penalty area and progressive passes in general only see a minor improvement under Setién.
Quirks of possession sequences and build-up
The final section of this article will have a look at the possession sequences and their characteristics under each coach.
We do see a decrease in directness under Setién. But this is not a criticism. At least not based on these numbers. There is no one way to play football. And there is certainly nothing wrong with being too direct, or otherwise. Next, we take a look at the patterns behind successful and unsuccessful goal kicks from Marc-André ter Stegen.
It is very clear that Barcelona enjoyed most success when using short goal kicks. Long goal kicks to escape pressure failed with spectacular regularity and has umpteen reasons. Luis Suárez, at 34, can no longer provide the same hold-up play as before. Messi cannot be possibly expected to win aerial duels against defenders to try to bring a long ball down.
Short goal kicks during buildup are most often successful, but not without their own issues. With short goal kicks, Barcelona are expected to play out of the press from the back. And this worked fine against most La Liga teams, but got brutally exposed in the Champions League and against better La Liga teams who had organised and structured high press.
The problems were two-fold, in my opinion: a general lack of mobility by the midfield in the build-up phase, and lack of runners out wide who could serve as a target to hit long balls to. Moving forward this season, this is something new coach Ronald Koeman needs to address as soon as possible.
At last, we will look at some typical passes used in build-up. To do this, a clustering analysis was done on all successful passes that originate in different thirds of the pitch. The clustering analysis groups together passes based on their similarity – passes originating and ending roughly from the same zone of the pitch will get clustered together. What is shown below is an average representative of such pass clusters. In fact, five of them for each zone – these are the five most frequently played pass types in each zone.
Something interesting that probably bothers me a bit is the lack of passes coming into the danger zones from wide. The passes in the final third are mostly from the half-spaces to out wide or entering the box, but nothing from out wide enter the box frequently enough.
Anyway, after this really long and arduous journey – if you made it till this far –, I will just simply end with this: Quique Setién did marginally improve Barcelona in certain ways. Fitness-wise, it certainly seemed so. Pressing was much better. However, Barça still didn’t take enough advantage of opposition slip-ups and didn’t create enough progressive actions. And the results sadly enough didn’t justify sticking with Setién.
This is by no means a complete comparison of the two managers. There are probably a million more things one can talk about. If you want to see something else, let it know in the comments below.
Acknowledgements: David Bravo, editor-in-chief of Barça Universal, and Shantanu Das, a friend from pre-historic ages and a fellow football fan
Detailed Analysis: Atletico Madrid 1-0 Barcelona
In collaboration with Anurag Agate.
Ronald Koeman’s Barcelona faced Diego Simeone’s Atletico Madrid at the Wanda Metropolitano. In a game marred by defensive blunders and devastating injuries, Barcelona lost the game 1-0 to fellow title-challengers.
A 1-0 loss to Atletico Madrid in La Liga left Barcelona reeling midtable. This was also the first time Diego Simeone’s side beat Barcelona in the La Liga. Coupled with crucial injuries to Gerard Pique and Sergi Roberto, Barcelona now face a dire path ahead of their UCL game against Dynamo Kyiv.
Barcelona structure and formation
Ronald Koeman went in with his tried and tested 4-2-3-1 formation. Marc Andre Ter Stegen started in goal again. Gerard Pique and Clement Lenglet formed the centre back pairing, flanked by Jordi Alba and Sergi Roberto. In the absence of Sergio Busquets, Miralem Pjanic stepped up to form the double pivot with the ever-present Frenkie de Jong.
Pedri and Ousmane Dembele played on the flanks, with Lionel Messi in the hole and Antoine Griezmann upfront. However, as before, Messi and Griezmann had lots of interchanging positions. Pedri played more in the half-space in possession while Dembele stayed out wide. This often made the team structure a lop-sided 4-4-2. In defensive transitions, it was always a 4-4-2 with Griezmann dropping deeper to defend. Messi restricted his pressing to zones high up the pitch.
Frenkie de Jong had the freedom to push up high in the first half. However, the absence of Ansu Fati meant that the usual overload on the left side did not work in this game. Pedri had a poor game in general. Him moving far too infield to let Alba run down the left did not quite work – the passing was far too restricted by Atleti’s excellent defending. A second-half injury to Pique meant that de Jong had to play 35 minutes roughly as a centre back, which he did very well.
Atletico structure and formation
Atletico were missing some key personnel as well, most notably perhaps, Luis Suarez up top. They also missed a regular left-back Renan Lodi, and Hector Herrera and Lucas Torreira in midfield. They lined up in a highly asymmetric 4-4-2/5-3-2 structure and style.
Stefan Savic and Jose Gimenez formed the centre back partnership. Mario Hermoso played in a hybrid centre-back/extremely defensive full-back role. Kieran Trippier was the more offensive fullback, practically functioning as a wing-back. Yannick Carrasco and Marcos Llorente joined the reliable duo of Koke and Saul Niguez in central midfield as wide midfielders. Carrasco played almost in a hybrid wide midfield/wingback role. Joao Felix and Angel Correa formed the front two.
The hybrid system was particularly evident in the different phases of the game. In attack, Hermoso would push out wide like a full back but stay in more defensive, withdrawn zones. Carrasco had the freedom to stay wide looking for overlapping runs to meet Felix’s clever passes. On the other side, Llorente would shift infield, allowing Trippier to bomb forward.
Felix himself overlapped down the left side several times, trying to create numeric overloads against Roberto and Pique, dragging Pjanic wide in the process. Carrasco’s and Felix’s overlaps on the left, coupled with Saul Niguez moving ball-near side and Correa dropping in to give options – this combination created quite a few problems in the first half. Here is an example – it led to Saul’s shot early on which was saved by ter Stegen.
The game was more or less evenly balanced – neither team were outright dominant than the other in any aspect. Here is the game data at a glance:
Barcelona enjoyed marginally more possession, marginally more shots and shots on target, and a better press than Atletico. Of course, the hosts had the all-deciding goal in their favour. Neither team generated high-quality shots overall, as the shot map and xG flowchart shows :
Barcelona’s possession superiority was pretty stale. Barcelona failed to dominate critical territorial zones, measured by field tilt – which is the percentage share of final third passes of each team. Even though Barcelona had higher field tilt, it was only marginal. What strikes out is that just the goal came when Barcelona were enjoying their best bit of territorial dominance.
Buildup to shots and goals
Next we take a look at some of the shots and the goal. Early on, Barcelona had the chance to score. Dembele burned his marker with pace and sent a cross into the box. It was met by a clever flick by Greizmann. The shot sailed high unfortunately.
Atleti had their chances on the other side as well. Soon after Saul saw his shot saved, the other flank created yet another moment of danger. A brilliant interchange of passing involving Correa and Trippier met Llorente’s clever run into the box. The shot crashed against the bar.
Towards the end of the first half, Barcelona could have scored again. There was a brilliant bit of buildup, a clever run by Griezmann to drag a defender, and then Messi ghosted blindside of the center mids to meet Alba’s nutmeg pass. The angle was too tight and Messi failed to score.
Soon after, Barcelona conceded the goal. Pique stepped up to intercept a long ball. Ideally, that should have been fine, except Pique miscontrolled the pass. That left almost everyone out of position. A simple ball over the top released Carrasco into oceans of space. But the maddening part perhaps was that ter Stegen left his box wildly to tackle the Belgian. He missed; Carrasco did not – he scored into an empty net from distance.
In the second half, Barcelona had chances to equalize. However, Lenglet headed straight at Jan Oblak twice. Greizmann headed straight at Oblak once. Barcelona failed to engineer any better chances than those. The key passes map shows the crosses into the box:
Atletico’s strength lies in engaging from wide areas. In this game, their biggest threats came again from the wide zones. Hermoso, Koke, Saul and Felix regularly released Carrasco and Trippier down the flanks. Trippier would often look for cutbacks or layoff into Llorente upfield.
Barcelona on the other hand tried to create from all possible zones. Frenkie de Jong managed to pull off a wonderful long pass into the box that Greizmann miscontrolled. Dembele single-handedly created chances from the right. The combinations of Alba and Messi created – in subdued amounts – danger from the other side.
For Barcelona, Messi, de Jong, Dembele and Alba were the bulk progressive passers. For Atleti, Koke, Trippier, Hermoso and Savic progressed the ball the most.
Both teams also tried to use width a lot. Surprisingly, Barcelona had more switches of play than Atleti, who have built their game to attack wider areas. For Barcelona, perhaps the reason for frequent switching was that they could not progress a lot directly.
The game data table posted above shows us that neither team pressed a lot. PPDA, which is a proxy for pressing intensity, was around 20 for both teams (low values of PPDA indicate high pressing). Here are the maps showing the defensive activities of both team:
Atleti forbade any progress down the centre. Upfield, they tried to press Pedri and Alba from creating too much danger. Deep in their half, they tried to force Dembele as wide as possible and tried to isolate him. Barcelona pressed all over the pitch in the middle-third. In deeper zones, they had to deal with the wide threats of Carrasco and Felix, and Llorente’s infield runs. The following plot also shows how Atleti forced passes wide and forced mistakes :
Two recurring issues troubled Barcelona yet again. The lack of chemistry and the lack of experience of the youth meant that certain runs went untracked. Atleti’s rapid front line dragged Barcelona into wrong zones, allowing trailing players to ghost into blindside runs. Saul and Llorente’s efforts at goal are perfect examples of this. In the first case, Pjanic was pulled in, leaving Saul free. In the second case, Pedri’s inexperience led to him losing his mark against Llorente completely.
Speed is always an issue that Barcelona has had trouble against. Llorente’s quick underlaps created quite a bit of trouble for Lenglet. Here is yet another example of a run – the pass from Llorente was thankfully cleared.
The goal was a combination of poor positioning and lack of speed, combined with some poor touch and terrible decision-making. Pique was out of position when he made the failed interception. No one in the team was speedy enough to catch up to Carrasco down the left. Ter Stegen should have communicated better with Lenglet and stayed in the box because Lenglet was haring down to secure the centre.
Issues have now been compounded with injuries to Pique and Roberto. If they face lengthy spells away from the pitch, Barcelona are stretched thin in the defence department. De Jong looks set to continue as a centre back for the next game at the very least and Sergino Dest will have to start. Barcelona faces extremely testing times ahead.