Cage match
On swirling build-up play, revamped attacks, White’s fatigue, Raya's mare, the brilliant stubbornness of Fløategaard, discount Rice, and a blossoming strike partnership
Moments of genius. Huge fuck-ups. Last-gasp winners. Unai Emery’s thorough destruction of Man City. A Spurs loss.
This week has been a good month. And it’s only Thursday.
It’s easy (and fun) to critique international breaks, especially when meaningful games are few, yet injuries are aplenty. For once, we have reason to be thankful. It seems Gabriel Jesus and Kai Havertz have returned at their best.
The last few games have showcased some of the subtle changes to Arsenal’s underlying dynamics that were pushed to production after the Chelsea draw. These tweaks, largely in the attacking phases, have come in the same manner that the one character went bankrupt in The Sun Also Rises: “Two ways. Gradually, then suddenly.”
The biggest topics of discussion are all related, and all have names:
Gabriel Jesus looks back at his expansive best
The Kai Havertz Project is coming together
Martin Ødegaard has seen a dramatic (and welcome) shift in role
We’ll dive into these topics more as the article goes on, but make no mistake, they are the primary storylines.
Except Kenilworth Road cares not for your precious narratives. This is not to say it’s an impossible ground to play at, or that Luton are particularly good. They’re not. But they are a determined, well-coached foe.
What is true is that games at The Kenny seem to live in a galaxy all their own. They feel tangibly, and refreshingly, different than the others. The poor camerapeople zoom and wind around, hoping to catch up to the play. The pitch itself is a full 521 square metres smaller than the Emirates. It’s a hall of mirrors.
That feeling extended to the game. For a team that can often feel boringly in control, even in contests against more even foes, this immediately felt like a slightly more hectic, loose, direct, Liverpooly game from the whistle. While Arsenal were guilty of looking lethargic in a few mid-week clashes last year, I’m not sure that was fully the case here, at least not across the teamsheet. The front-6 in particular returned much of the energy and fightiness of the freshly-promoted side, but that last push in the duel often seemed to tip in the direction of the home team.
(I’m sure it was entertaining for a neutral. These days, Arsenal games do not usually contain seven goals.)
The away fans, who entered through the Oak Road stand, surrounded by brick houses and independent shops, got their money’s worth — and much more. The delirious scenes after Declan Rice’s last gasp winner felt shocking, yet shockingly familiar. It was another moment for the season-long compilation video. I yelled a lot.
Throughout, there were plenty of reasons for excitement, confusion, concern, then excitement again. Especially as Man City are looking a little vulnerable, we ask ourselves a question.
What do we make of it all?
👉 Everybody is a midfielder
To get a sense of how cagey and all-action this game was, here’s a stat: Wyscout credits Kai Havertz and Gabriel Jesus with engaging in 65 combined duels.
A high number of loose balls don’t usually portend well for this Arsenal side, which is why Arteta has been trying to clamp that down, at some cost to attacking fluidity. In games where the opponent have 50+ recoveries, Arsenal are 1-2-2. In games where Arsenal have 50+ recoveries, they are still only 2-1-2.
This was directly in line with the Luton gameplan and performance, both of which deserve much credit. Whereas some teams can push an opportunistic but risk-averse press against Arsenal, The Hatters went for it with more conviction.
They pressed in what looked like a high 3-4-3, with a second line looking to solidify the sprints of the energetic three. It kept Arsenal honest.
As the phases progressed, it mostly fell back into a jumpy 5-4-1, at least as far as I can tell. It was hard to get all the players in the same shot with the Kenilworth Kamera Angle, so here’s the best I got:
Arsenal had a series of counter-measures. The first was to go long and play over the press. Raya launched every single one of his goal kicks (5/5).
This is one of the ways in which “press resistance” deserves a broader definition than it often gets. In practice, Kai Havertz can give a silky La Masia graduate a run for their money, because he can often negate a press just as well.
Uh-oh, I’ve made a meme.
But otherwise, Arsenal sought to play out of the back, bait the press, and exploit the spaces they may receive in return. With Zinchenko kept in bubble wrap after Tomiyasu’s devastatingly frustrating injury, Jakub Kiwior got the start, and the team would look to get creative to compensate.
What resulted was the most flexible build-up performance of the season: a jam-packed midfield on a tiny pitch, where width was difficult to find and space was at a premium. Despite people like me clamoring for this kind of thing every game — hey, it’s fun to watch — this approach is not without its drawbacks.
We got a sense of how expansive it could be early. At 15’, here was a look at may conform to our expectations, at least initially. Ødegaard is dropping into his deeper role as Kai vacates and heads up the pitch.
But look. Just 9 seconds later, Ødegaard has gone to the other side of the pivot.
Would it be funny to call him Zinchenk⌀? Not really. But I’ve already typed it, and deleting it would be a whole thing.
Now look again, just six seconds later. Martinelli is jealous of how Ødegaard’s performances have improved after dropping deeper into the midfield, and wants to get in on the action.
…and finally, this. This is the same build-up sequence, and now the double-pivot is Ødegaard and Jesus, with Martinelli in the old Ødegaard spot, and Rice as an auxiliary CB.
So, this is all very fun, and generally effective. We’ll explain later why this can be so impactful, but here’s the spoiler: when things get stagnant in the final third, it’s often because teams are sitting back, and when that happens, rotations can often feel forced and artificial. When the rotations start back here, the players arrive into the final third in interesting spots, angles, and relationships. That’s where a lot of joy is found.
But there were two drawbacks.
Kiwior didn’t look as comfortable: After a great cameo against Lens, not to mention a game at CCB against West Ham where he was probably the team’s best player, there was hope that Kiwior was skipping steps in his development. This was more of a reality check performance. Because of all of these complex and never-ending rotations, he was a little hesitant off-the-ball, trying to understand his place in the washing machine. This means that once he actually got it, he had less space. He was also a press trigger. This led to a contribution of 97 yards of progressive distance, which nets out to about 150 per 90. For comparison, Zinchenko is averaging 352.4 per 90 for Arsenal this year. It’s all good, and he should still be given time and starts like this. It just didn’t work that well here.
Rest defending: With rotations like this, there is a bigger chance of getting caught out on a ball loss. If, say, a pass gets mishit as Rice is rotating backwards, or Martinelli and Havertz are switching, there are generally more spaces for the opponent to exploit. Transitions didn’t feel like they were getting snuffed out with the same cold-bloodedness. It is not without risk.
There were some other things to work on.
👉 White Flag
Benjamin White, probably the team’s most consistent performer last year, looked like a stubborn man who was desperately trying to overcome an exhausted body that wasn’t cooperating. He was the “tis merely a flesh wound” dude from Monty Python.
His overlaps were enormous and his assist was perfectly struck. But going back, he can sometimes struggle on the margins with wide defending — looking a touch too vertical, stiff, and big steppy. He usually overcomes this with dark arts and physicality.
His body didn’t allow that on Tuesday, finishing 2/8 in defensive duels overall. The lead-up to the Ross Barkley goal had two opportunities for an intervention upfield. Then, in the third and final one, White got it wrong. He afforded too much space, he leaned back, and his steps were heavy.
White was likely thinking two things: “Stop him from cutting back to his right,” and “I’m very tired.” Before Barkley even offered a move, White was completely on the heels of his feet (instead of the balls), vertical yet leaning back, and making it too easy for Barkley to spot the heavy step. This surrendered the battle.
His name is “White” and “flag” means “tired,” get it? I’m a word genius.
And then Raya had a mare.
👉 Raya’s mare
Here’s what I tweeted after the game:
My take on Raya is that he had a terrible game and should continue starting indefinitely.
My feelings aren’t much more complicated than that. There were two big fuck-ups. But patience feels especially prudent in a game in which Kai Havertz looked like a star after all the second-guessing he endured earlier in his tenure. Perhaps it’s old fashioned expectation management: when a new player joins, it seems fairer to expect growing pains than seamless integration. Declan Rice (and Jurriën Timber 😢) aren’t the norm.
How do you get past this stage? Minutes and trust. But I repeat myself.
The reason I think that trust is probably warranted is because I watched a whole lot of Raya last year, and also watched a lot of him in the Championship. The mistakes on Tuesday weren’t in line with all that. He had a bad game against Newcastle last year, but other than that, allowed very few low-percentage chances, and yeah, had more saves than anybody in the league. There are some dynamic shifts to consider (Brentford sat back more, and allowed longer shots), but nothing too major. It’s all covered well in this piece, and though I don’t necessarily agree with every point, I’m open to anything.
If you feel that his tally of GA (Goals Against) is due to a fortunate lack of shots against, it’s also worth considering how that comes to be. You limit shots not necessarily through sturdy defending, but through surer possession. It is no small thing to offer a couple hundred yards more of reliable passing per game. The Brentford-to-Lens switch felt night-and-day in this regard.
I also don’t tend to put stock in the idea that he’s suffering due to looking over his shoulder. It seems fairly clear to me that the job is his.
I’d lean towards Occam's Razor with this one. It’s early. He’s mostly been good. He’s been fairly transformative to the in-possession game. He’s also had a couple of mistakes, and just hasn’t looked at his best as a shot stopper.
Maybe … he’s just settling in.
👉 Cornered
Arsenal have gotten praised for their set piece defending, and there’s merit to that. Even after Luton, their embiggening lineup has still conceded less xG on set pieces than anyone in the league.
(While we’re on the topic, Jake Fox has an incredibly detailed piece that goes through 141 Arsenal corners, if you’re a sicko like me. You’re reading this, so I think I know the answer.)
But there’s a sure-fire way to avoid conceding on set pieces: don’t allow the corner in the first place. Arsenal have done a better job of that than anyone in the league. Here was the leaderboard before the game.
They gave up 3 corner kicks in this one, which isn’t a lot, but is when you consider the pace. They’ve given up 5 in the last 6 games total.
All told, Arsenal have now allowed 21 shots on corners all year. For Tottenham, that number is 63 as of this writing. For Chelsea, that number is 62.
👉 A strike partnership
Setting the stage for the possible dynamics up front, here’s what I wrote back in July, comparing and contrasting it to Man City’s needs:
It is natural, and human, to seek a more rounded 8 in this role. Nobody suggesting it is silly or wrong (and I’d still like somebody who can play over there for matchups and depth, in fact). Especially prior the Havertz links, I was game to give Rice an adjustment year in both spots before likely taking over at the 6 from there.
[…]
But Haaland and Jesus are fundamentally different players, and this impacts the responsibilities of those around them. While all the focus is on the gap in their finishing ability, not every advantage is in Haaland’s court. Jesus got 46.2 touches per 90 last year; Haaland had 24.8. Jesus offers about 80 more yards of progression over Haaland. He also has three times as many recoveries, and a huge advantage in terms of raw defensive output: literally ten times as much (1.92 tackles+interceptions per 90 to .19).
What does this mean? First, Jesus fills much of the gap in build-up touches of a more expected profile in the role. Second, Arteta is building something different, and it must be understood as such, because his model will continue to deviate from Pep’s.
Put better: Jesus is a striker who is kind of a midfielder, which helps Havertz be a midfielder who is kind of a striker. This is not haphazard — this is the best use of their respective abilities.
This has come into full bloom in the last two games.
The main factor is just that Jesus has been playing out of his damn mind. Let’s look at five consecutive Arsenal goals. Warning: this will be less an analysis, and more a celebration.
For the first one, Jesus does hold-up play for Saka on the right, expertly using his body to gain leverage and create a beautifully-worked chance.
For the next one, he is telepathic with his old buddy Zinchenko, playing his ball into space just as his long-time teammate runs into it. Goal.
Next, he hounds Thomas Kaminski into clearing it into a dangerous spot, then rushes to take the snap throw-in himself. We’ve talked about how throw-ins should either be “long, fast, or clever.” This was the second category.
This frees up Saka, who — after Havertz pulls away the Luton defenders to the near-post — delivers it to Martinelli for a goal.
Next, our strikery midfielder pulls two defenders into the near-post with his gravity, freeing Jesus to feel out the space and pound this in with his head:
…and finally, Jesus immediately manufactures the leveling third goal out of thin air, muscling for position before Havertz floats into the open space for a tap-in.
That’s a fucking ridiculous run. It should give you some sense of the impact he has wrought. But the most promising thing of all may be the mutually beneficial partnership with Havertz.
Rewatching the game really hammered home how much sustained pressure they kept up together. This wasn’t an instance of Havertz taking a shot and the bounce finally going his way. He took five, and one went in. And the means of creating those shots look sustainable.
The following is one very, very small moment, but hopefully instructive. You can see how many players that Ødegaard is able to discard with one pass — literally four. But more than that, as soon as Jesus swings to accept, Havertz is right there to fill the zone. This happened over and over.
It’s also worth looking back at the Jesus goal where Havertz was able to vacuum up two defenders to give Jesus a free header. How did that come about?
It goes back to the build-up fluidity we covered before. Here, Havertz drops deep and hits a wall pass out to a free Rice. This is probably the most-used way of beating a press.
…and next Rice carries it up, so Jesus drops down on the left to accept. This frees up the “9” spot for Havertz to rush into.
…and ta-da, you have Havertz rushing into the box, with Jesus in behind. Just another bit of sensory overload to unsettle the defenders.
The team was also using a reliable method for going long that you heard me crying for in the early parts of the season. Havertz switched up to become the furthest attacker, and Jesus played off of him. This is a nightmare for a centre-back: you have to try and win the header, then watch the bounce, then track Martinelli and Saka marauding around you, and Jesus (and often Rice) pushing forwards as elite winners of second balls. It’s just a lot to manage.
The Havertz stocks are on a steady climb. He had the winner against Brentford. He had the best game of the season against Lens. Then, he topped that against Luton.
This feels like a fairly meaningful step in his progress.
Here’s what Arteta said after Arsenal lost 2-0 to Manchester United in the preseason, and the media and fanbase had some concerns:
“We have to find the right balance but a year ago Granit could never play as an attacking midfielder. Let’s try to do the work,” said Arteta. “It’s the first time we played with four players inside and three of them had never played together, ever. So ‘tranquilo’.
“Day by day, keep doing the right things and give people the opportunity to settle in and we will be fine.
“You use the word ‘physical’ but if I put Granit and Kai (together) who is more physical? He’s (Havertz) 1.91m (tall). It depends what physical is.
“Physical is to run, to run in behind, run forward, to tackle, to defend. He will fit in with the qualities we have for sure and the good thing is that we have options in midfield. When we want a game to become more physical we have the options to be very, very physical.”
After my long thread on “The Havertz Hitch” — about how his new circumstances and his time at Chelsea had conspired to make him second-guess his pristine instincts — I feel proud to say that the “hitch” has said “fuckity bye” for several weeks in a row now.
He has been attacking space. If the hitch stays gone, that’s good for Arsenal’s chances.
👉 Fløategaard is here to stay
We covered this at length last time, and Friend of the Newsletter Lewis Ambrose put together a great piece today on the captain’s updated role. This shift has the potential to be the most meaningful tactical story of the Arsenal season.
We can see the evolution in real-time.
It’s not all that complicated. If you have a player who is brilliant at passing and moving the ball forward, it’s usually an advantage to get them on the ball a lot. I had an out-of-body experience watching Pedri:
Earlier this year I watched a Barcelona/Getafe game that ended 0-0. Barça attempted 767 passes. But Pedri was pinning the backline, and only contributed 35 — 8th on the team. My simple brain thought “maybe they should have the best passer ... pass more?"
Anyway, Martin Ødegaard.
Against Luton, Ødegaard was absolutely relentless. Instead of awaiting service like a striker, he becomes the protagonist — dropping deep, dictating play, curling and feinting runs, then delivering great balls (often from the half-space!!!).
It strikes me as a big deal.
👉 Wrap it up, Billy
This was a weird and fun game. In many ways it should be kept in a box all its own. Luton had their fewest shots of the year (6) but matched their high for goals (3). Arsenal dominated in so many ways, but the scoreline didn’t feel totally unfair, given the mistakes. It was also perhaps a helpful warm-up to the next match against Emery’s Aston Villa, one that is likely to be similarly hotly-contested, pressy, and action-packed — but against a team with more talent.
What didn’t I cover?
Oh yeah, Saka is the second-best right-winger in the world.
Martinelli is good.
And the great thing about Rice is that it’s delicious, cheap (only £105 million!), and goes with anything.
Have a beautiful one.
— Billy
Incredible Rice analogy, worth the sub price alone.