As the Premier League’s two best sides ready themselves for Sunday’s potentially pivotal battle for English supremacy, we know what’s going to happen, right? We’ve seen it all play out before, as recently as four weeks ago. Manchester City are going to reassert their dizzying superiority over Arsenal once more and with it they will wrestle control of the title race from England’s nearly rans.
And yet… Pep Guardiola was not far from the truth when he asked last week if there was “even one bet” for Manchester City to beat Arsenal in the EFL Cup Final. His side, slipping out of the title race after draws with Nottingham Forest and West Ham, were underdogs and that was how he liked it. Now, City are on the hunt.
According to the bookmakers, City are strong favorites to win at the Etihad Stadium on Sunday. Were they to do so, they would be three points behind the Premier League leaders with a game in hand. Their fate would be in their hands, as it would Arsenal, given that all that would separate these two would be goal difference if they won their remaining games.
The external mood — and if you listen to Arsenal players it is just that — seems to have shifted radically. The question is if this is an overcorrection. Have Arsenal become that bad? Are City that vastly improved? Did the Carabao Cup capture a moment in time or reflect a wider trend?
Even Guardiola seems to know it’s a little bit more of the latter. “I know Mikel. They are going to adjust something and we have to prepare to do it,” he said. “In the end, it is more simple. It is how your players individually win the me against you and how they are able to do it.”
What went wrong at Wembley
Few would deny that a few weeks ago, it was City’s players who won the “me against you,” though even that comes with some caveats. At halftime it was the Gunners who had had the better of the shot count, five to two, and were established as the likely favorites to win an attritional contest. The warning signs were already there for what would be their downfall, but Arsenal’s issues would have to coalesce around City reaching a height that stunned even Guardiola.
The most profound of those struggles was Arsenal’s inability to manipulate the defensive shape of their opponents. Without the ball City set up in a 4-2-4 out of which a forward may burst, but in general, they were more than happy to cede possession to the center backs and, in particular, Kepa Arrizabalaga and challenge them to break through the sky blue wall.
EFL
Though it was hardly the only issue Arteta faced, the base problem from which so many others flowed was this: City let them have the ball in their own third and Arsenal could not progress it to anywhere where they could threaten. The 28 progressive passes they completed were the third lowest in any game this season, and in the other two — a league win at Brighton and the second leg of the EFL Cup semifinal — they had advantages to hunker down and protect from early on. This time Arsenal could not go around or through, and going long, well, more on that in a moment.
There are tactical adjustments that Arsenal could have made then and could make on Sunday. Arsenal congested the extreme central areas too heavily at Wembley. Here, Leandro Trossard’s positioning is all wrong; even if Kepa Arrizabalaga and co. find a way through the first line, Declan Rice and Martin Zubimendi aren’t going to have an option that stretches the pitch, but one that conjests the middle. Rice’s most reliable out ball was to Bukayo Saka holding width on the right, but he lacked the fitness to beat a man. No one else was really standing in the places where City couldn’t get to, which should be numerous when they line up in an aggressive 4-2-4.
Another option, which is admittedly more high-risk, is to break the City defensive shape by carrying rather than passing. This is particularly the case for someone like William Saliba, who averages nearly seven progressive carries per 90 this season. In the EFL Cup Final he completed just three. Is it a risk to charge at a figure as hulking as Erling Haaland? Certainly, but rarely will the reward be quite as significant as it would be for a win on Sunday.
Ultimately, though, a great deal of the issues Arsenal had with getting through City were of personnel. Barring a dramatic late injury, David Raya will start in goal ahead of Kepa, affording his team not just a high-grade shot stopper and sweeper but a ball-player who can break the lines with more accuracy than most.
That is a lock, but three injury boosts would help Arsenal no end too. Ben White put in a credible performance as both an outlet for passes and pure defender, especially given the injury issues he has suffered since putting his knees on the line for the 2-2 draw with City in September 2024. He just is not Jurrien Timber, whose distribution to Saka and Noni Madueke alone accounts for seven percent of the progressive passes Arsenal have made this season.
Riccardo Calafiori is as much outlet as passer, his inverted runs are another way the Gunners might pull apart City’s 4-2-4.
The defenders are doubts as is club captain Martin Odegaard, whose commitment to dropping deep and offering an angle is so missed in this Arsenal team. “We will try again,” Arteta said of the trio. “Some players are quite close, but the turnaround is short, so we will try tomorrow to push everybody, and if they are in good condition, they will be part of us, and if not, they won’t. They haven’t been in the last few weeks, unfortunately.”
The absence of some key players has, Arteta hinted, been part of the reason why Arsenal’s passing has run aground lately. “Part of football,” was how he described Declan Rice’s assertion in midweek that the short passing basics had gone missing from his side. “Part of the moment. When you are missing certain players that the relationship, the equation is a bit different.
“Working on that means sometimes not talking too much about it and taking more honesty, more responsibility and doing it again.”
TruMedia
If there’s one other change it must surely come up front. If Arsenal cannot get through City’s aerial pressure, they need to be able to go over it, but Viktor Gyokeres is just not a striker who can ensure his team emerges with possession when his defenders go long. Kepa could release pressure out to the right but only completed two long passes down the middle, one of which found Gyokeres.
As CBS Sports reported on Monday, the Swede has the highest proportion of possessions lost per touch among Premier League forwards with 900 minutes this season. He is not an outlet for progressive passes and struggled to impose himself on Abdukodir Khusanov last time, let alone Marc Guehi.
CBS Sports
Arsenal’s crisis came in the second half because they could not keep the ball when they survived a burst from a much-improved City attack. After hooking a cross into the box in the 48th minute, Gyokeres’ next meaningful touch was in the 83rd minute of a match long since lost. It was the same story against Sporting on Wednesday, but this time there was an early bath for the No.14, justified by Arteta as “the game [requiring] something else, another kind of nine-man profile so that he could associate us better, so that he could keep the ball, there was no space to run.
“Kai [Havertz] gave us that.” Havertz must surely start on Sunday. He did not set the world alight last time out against City, but that was as an attacking midfielder, a role where he spent too much time close to Gyokeres and perhaps not enough offering a release valve for the Arsenal double pivot. Putting the German up top with either Odegaard or Eberechi Eze off him offers more possession security and probably a more reliable outball in the first place. After all, at the peak of his powers in the 2023-24 and 2024-25 seasons, Havertz won nearly 48% of his aerial duels, far ahead of the 38% average for forwards in the division. This season Gyokeres has won 32%.
What do other games tell us?
Those are the reactions Arsenal might make to a really bad day last time out. And yet in the more recent history of this rivalry, if there’s a game that bucks the trend it’s the most recent one. Of the seven games since the start of the 2023-24 season, Arteta’s side have only lost the EFL Cup Final, winning two, lifting the Community Shield after a penalty shootout win and drawing the rest.
And yet even that run of success defies analysis. These games are just consistently weird. In September, City responded to Haaland’s early opener by lining up in the most outrageous low block of Guardiola’s managerial career, allowing Arsenal to complete 517 passes to their 229. Before that came the 5-1 where everything turned goal for the Gunners in the last 35 minutes, five shots on target worth 0.25 xG resulting in four goals and the most high-value opening spurned. Then you have the ferocious rearguard of the 10 men at the Etihad, the 0-0 draw that felt like a boon in the title race but came to be viewed as a bust for Arsenal.
Spot the pattern. Easier said than done. As such there is only so much preparation that even coaches as meticulous as Arteta and Guardiola can do. “We will try to do something, they will try something. Five minutes later it will change, it will adapt and that’s the flow of the game.
“That’s when the game is there to be adapted to. Because when the game becomes chaotic, adaptation is action. And that’s it, there is nothing else to do.”
Despite what Arteta said there, if there is one trend that does flow these games it is that they tend not to get that chaotic. Ok, last season doesn’t count. But even then… Those games felt like pure spectacle from Leandro Trossard’s red card to Myles Lewis-Skelly apeing Haaland’s celebration via the limbs caused by John Stones, but even those games were probably lower-incident than you remember them. The 5-1 win for the Gunners late last season had 19 shots, the EFL Cup Final 20. That’s a quarter fewer than the average City game over the last two years.
CBS Sports
Key to this, and another of the relatively rare trends from these games, is how Arsenal have proven themselves able to at least somewhat quell Haaland. Of course, the big guy has scored five in eight meetings with Sunday’s opponent, including three of the last four, but compare the difficulties he gave them in the 2022-23 season, and it is like night and day. At 11 vs. 11 over the last three seasons Haaland averages just 1.87 shots and 0.42 xG per 90 against the Gunners. The slight problem Arteta has had is that they often haven’t been able to keep one of those shots from being quite a good shot. Give Haaland one, and he might just score it, or City might exploit his gravitational force to make space for Nico O’Reilly at the back post. Then again, if it is O’Reilly beating you rather than Haaland, sometimes you have to accept that it is not your day.
Is this really a title decider?
So we’ve established that you can’t learn as much as you might think from the most recent City game nor many of the ones before. And for my final trick, I am going to counsel against learning too much from Sunday’s. Now, there is no disputing that this is the biggest remaining game in the title race. If Arsenal were to win it and go nine points clear, albeit with an additional game played, they might reasonably feel they’ve knocked City out of the running. A draw that preserved a cushion of at least three points and a smidge of goal difference advantage would be a powerful positive for Arteta, though not one he has even considered allowing into his thinking. Indeed, he won’t even plan on sitting back on a one-goal lead like City did to his team.
“We’re not going to propose a game like this because we never do,” he said. “Sometimes your opponent is that good, that forces you to be there. In City’s case, you’re going to have to have moments where you’re going to have to defend deep in your box for a period of time. That’s the reality.”
Suppose though, that City do win. That’s the title race, right? Well no. After all, as we said at the top, the outcome of this season would be just as much within Arsenal’s hands as City’s. If both teams won out from the Etihad onwards the Premier League title would be awarded to the team with the best goal difference. Currently, that is the Gunners though one suspects you and I might gravitate towards the team with Haaland if the final day does become a straight shootout.
It might not. What is notable is that projection models such as Opta’s would still make Arsenal favorites even if they lost. The bookmakers would not, a quick browse of current odds would suggest that the general view is that the Gunners are given a 50-60% chance of being crowned champions before they go to the Etihad. You would assume that would be around City’s level if they win. Opta’s projection model begs to differ and would give Arsenal a 69% chance of winning the title even if they lose this weekend. After all, City wouldn’t have the lead and a fixture list that includes Brentford, Bournemouth and Aston Villa looks trickier than Arsenal’s.
Which prediction to believe? Well, that rather depends on how permanent you believe two factors are. First of all, how real is the City of the last three games, the one that burst into life against Arsenal, blitzed Liverpool, and punished Chelsea for their lapse in composure? In each of those matches, you see a team come to life in a flash, win the game in the space of 15 minutes, and settle into a groove. Why couldn’t they be that team against Nottingham Forest or West Ham? Why were they faltering early on against Liverpool and Chelsea?
CBS Sports
What these recent games serve as is a proof of concept of how great City can be, particularly now that Guardiola has found an XI with which he vibes. O’Reilly’s recovery in time for Sunday’s game is a welcome boon and Khusanov is playing so well that Ruben Dias need not be greatly missed. City have proven in flashes that they can go to a level some might have thought beyond them. They also have a coach who has proven that he can get his teams on runs when it matters most. If nothing else, they have shown very recently that they can blitz Arsenal when the circumstances are right for them.
“If we play like the second-half during 95 minutes and they play like the second-half, we are going to win,” said Guardiola on Friday. “Well, maybe not because football is unpredictable, but it will be closer. That is what we want.”
The other question then is whether Arsenal will consistently play like they did in the second half at Wembley and against Bournemouth? As mentioned above, much will depend on the players available to Arteta. “The biggest lesson [for winning the title], April, May, the whole squad available, your best players on the pitch as much as possible. The probability to win it increases dramatically. As simple and as difficult as that.”
There will be no Bukayo Saka, whose Achilles has not quite healed. How they could have done with Mikel Merino in the games when Martin Zubimendi has struggled. If Odegaard, Timber and Calafiori do return, however, that is not a bad situation to be in for the run in. All three can do something to resuscitate a faltering attack.
CBS Sports
Arsenal aren’t the team they were. A 10-game rolling average of their xG that hovered around two per game near Christmas has only just crawled above one and a half. The defense isn’t at maximum efficiency either. And yet, even in a relative slump, they are still not far from an expected goal per game better than their opponents. For most of this season, City’s best level in terms of xG difference has been at or around Arsenal’s current slump. That is why prediction algorithms like them.
Of course, what they cannot account for is the bottle of title contenders. Nothing we’ve laid out before you really can (well that’s 3,000 words wasted). No matter what a cursory view of the last three seasons might have you believe, Arsenal’s nerve has never been tested quite like this. And it’s for that reason above all else that you can throw out so much about what you think you know about this game.
Viewing information
- Date: Sunday, April 19 | Time: 11:30 a.m. ET
- Location: Etihad Stadium — Manchester, United Kingdom
- TV: NBC
- Odds: Manchester City -120; Draw +250; Arsenal +300
