After a contest in which so much happened, maybe it seems strange to focus on things that didn’t happen.
But Manchester City’s second-half approach in their 2-2 draw against Arsenal was curiously passive, with Pep Guardiola barely changing his starting approach despite his side’s radically different task after half-time. Guardiola, regularly criticised for meddling, for once could be accused of underthinking things.
To get some things out of the way: the first half was a completely different contest, and played in a more open fashion than expected. Arsenal deserve credit for withstanding such pressure for almost the entire second half. And City did eventually scramble a late equaliser through substitute John Stones.
But, back to the point: why did Guardiola let the second half drift so much?
Here’s the system Guardiola started with. In possession, it was basically a 3-1-5-1. Kyle Walker narrowed his position from right-back, while Josko Gvardiol pushed infield to become City’s inside-left. Rodri played the holding role, two proper wingers stretched the play, and Erling Haaland led the line. Despite finding themselves 2-1 down at the break, because of an outstanding long-range strike from Riccardo Calafiori and another set-piece header from Gabriel, City were the better side in the first half.
Here’s the system Guardiola was playing midway through the second half. With the exception of Mateo Kovacic having replaced Rodri due to injury, it’s identical to the first period, and Guardiola made no further changes until 70 minutes.
And here’s the system in the latter period of the game. By this stage Guardiola had made substitutions. But it’s still that 3-1-5-1 in possession.
But the situation was completely different. Guardiola was facing an Arsenal side who played three very different systems without possession. They started with a compact 4-4-2, much as they played the previous weekend in a 1-0 win at Tottenham Hotspur.
When Mikel Arteta rearranged at half-time after Leandro Trossard’s red card, he sent his side out in a 5-4-0.
If that wasn’t extreme enough, midway through the second half he switched again to a 6-3-0.
It seems odd that Guardiola believed the same formation was the optimal approach against Arsenal’s increasingly desperate attempts to hold onto the lead.
More pertinent than the lack of system change was how cautious — or sluggish — Guardiola was in terms of substitutions. The pattern of the game was obvious within a couple of minutes of the second period — Arsenal were going to make no attempt to hold onto possession, press high or even launch counter-attacks, with only Gabriel Martinelli capable of sprinting forward on the break.
But Guardiola maintained a system that featured four recognised defenders. Manuel Akanji is speedy and technical enough for this type of situation. But Ruben Dias was ill-suited to breaking down a deep defence, taking too long in possession, unable to play penetrative passes and attempting some wayward long-range shots. It felt City often ended up in this static situation, with Dias on the ball, and Arsenal entirely comfortable.
Gvardiol is clearly capable of popping up with a goal, but continuing to use him as City’s inside-left was strange, particularly given Guardiola — as he explained after the game — had explicitly told his players not to simply pepper the box with crosses. As such, Gvardiol’s aerial threat wasn’t a major factor. Walker’s famed recovery pace was useful against Martinelli, but Stones (who was only introduced with 12 minutes remaining) or Rico Lewis (who wasn’t used at all) would have offered more in attack. Besides, Walker had been at least partly responsible for City’s two concessions, so was hardly having a good game.
Guardiola was once famed for eschewing proper centre-backs and turning midfielders — Javier Mascherano, Javi Martinez, Fernandinho — into defenders. For an away match at Crystal Palace in 2019, Guardiola knew his side were going to dominate possession, and would barely have to defend their own penalty box, so he fielded a centre-back pairing of Fernandinho and Rodri, two midfielders, because he knew they would spend the game on the halfway line. They kept a clean sheet in a 2-0 win.
In recent years, Guardiola has taken the opposite approach, often loading up on four proper defenders. That clearly has its benefits, and helped City win their first European Cup in 2023, but it’s difficult to see the justification for keeping four proper defenders on the pitch against a side who were playing no strikers, and not really attacking. City ended the game with Akanji, Dias, Gvardiol and Stones all on the pitch together, and while it was Stones who netted the late equaliser, he was the one centre-back you obviously would want in this situation because of his technical skill, and the player Guardiola took agonisingly long to introduce.
Guardiola took a while to make changes in the second half (Michael Regan/Getty Images)
All this seems particularly strange in the era of five substitutes. Guardiola’s first change had been enforced, but he had the capacity to overhaul his side and create something more technical. Even sticking with that 3-1-5-1 shape, an earlier version of Guardiola might have ended up with something like Akanji, Stones and Gvardiol as his back three, Kovacic in front, then perhaps Savinho, Lewis, Ilkay Gundogan, Bernardo and Phil Foden behind Haaland. That would have offered more passing quality from deep and more technical players in attack, and could still have shifted into a solid shape without possession.
It’s worth remembering that, two seasons ago, Guardiola started Bernardo at left-back away against Arsenal. It wasn’t overwhelmingly successful, but that’s the extent to which he was once willing to use creators in defence. Now he seems to expect defenders to be creators.
It was also notable that City offered minimal goalscoring threat from long range after half-time, for which Guardiola is less responsible. Of the four City players who most regularly shot from outside the box last season, Kevin De Bruyne was out, Foden was only a substitute, Rodri (who often pops up with crucial goals in big games) had gone off injured, and Julian Alvarez was sold to Atletico Madrid in the summer.
Kovacic comes up with the occasional screamer but isn’t a particularly good shooter. Walker and Dias’ regular efforts from distance weren’t very threatening. It feels like Guardiola wants his players to shoot from range in situations like this — more so than in his Barcelona days — because of the possibility of deflections or rebounds. But City were without their best players for the task.
Maybe Guardiola will feel his patience paid off. Stones was thrown on as a bonus centre-forward and popped up with the equaliser, a little reminiscent of how Gerard Pique grabbed a late goal when Guardiola’s Barcelona narrowly failed to overturn Inter’s 3-1 first leg lead in a memorable two-legged Champions League tie way back in 2010. Defenders aren’t just there to defend, after all.
But the goal came in added time at the end of stoppage time, from a rebound after a set piece. It didn’t feel like a grand plan had paid off. It felt more like Guardiola had got away with it.
Michael Cox concentrates on tactical analysis. He is the author of two books – The Mixer, about the tactical evolution of the Premier League, and Zonal Marking, about footballing philosophies across Europe. Follow Michael on Twitter @Zonal_Marking