State of play
Tactical lessons from Monaco, Everton, and Palace. Why things stagnate; why that second half cooked; the Havertz Hitch; Merino and updates to central access; speed, risk, and what still needs work
“It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of light, it was the season of darkness, it was the spring of hope, it was the winter of despair.”
(It does feel a little basic to start with Tale of Two Cities. But I’m a basic bitch.)
Disappointing matches leave themselves wide open to I Told You Sos.
I partake. The last we spoke, after Fulham, there was ample opportunity. I’d grimaced as soon as the lineup was posted, injury news and all, and then saw some of my pet gripes play out again: Timber played left-back, Partey played right-back, Havertz vacated the box, and the space he left wasn’t properly backfilled. Ugh.
After a crushing Everton draw, the only intellectually honest thing to say was this: Arteta played the exact lineup I would have played; then, he gave those players pretty much the exact same responsibilities I would have given them. As we’ll cover, whatever your hobby horse — Timber at right-back, Martinelli played into space, Saka central, Ødegaard drifting left, Havertz at 9, Havertz at shadow striker, central passing, longer shooting — it likely played out on the day. And yet, no goals.
Not so smart now, eh? Nice fucking newsletter.
In So Back, written after Ødegaard’s return and the Champions League walloping of Sporting, I ended with a note of caution.
The night is dark and full of terrors. The joke of the “we’re so back” meme is how quickly it can return to “it’s so over.” There is a threat of getting footballed at every moment, so the objective of the team is to play with a quality that is beyond the clutches of variance.
Whatever the causes, Arsenal haven’t yet been able to consistently perform “beyond the clutches of variance,” nor have we been particularly fortunate. There have been the circumstances I’ve outlined at length, but there also hasn’t been enough relentlessness in attack — not in creation, nor ideas. Though he’s been hamstrung, Arteta deserves some critique for his employment of certain configurations, which we’ll cover. And the players deserve some critique for their lack of urgency for spells.
But Crystal Palace happened, too. Facing a first-choice Premier League side, it brought the Two Cities of the team in even starker contrast. The respective halves saw a familiar refrain: certain setups clicked, and certain setups didn’t.
Then Jesus røse, and everything was right again.
Let’s trace back some of the evolution over the last two weeks, and see if we can expand on Arsenal’s explosion in that second half, why things have stagnated at other times, and what we can learn moving forward.
🔎 Investigating the approach against Everton
At first glance, the structure was roughly this.
You’ll see:
Lewis-Skelly in a staggered pivot with Rice
A tight back-three, with Timber purposefully closing distances
Merino and Ødegaard free to play high in the pockets or drop below
Saka has already rotated to striker in this example. Havertz was generally occupying that spot.
It’s hard to critique this all too strongly. As I said, I was excited to see this lineup.
We’ve talked about how Ødegaard does countless, selfless, hopeless runs through the middle that are timed to a switch out to Saka.
This pulls a defender inside, and gives Saka a second or two upon receiving. It’s the kind of run that requires genuine comfort with the position that becomes predictive.
On the left, that’s generally been missing. Havertz was timid at LCM for much of last year, and Rice doesn’t do it enough.
In the situation we started with, you’ll see the ball head out to Gabriel.
Merino immediately tracks Young over his shoulder, sees that he’ll be followed, and pushes to drag him down. The key is that this is happening before Gabriel even receives. We talk so much about formations and numbers, but success often comes down to timing, distances, and energy, as we know all too well.
This forces the handoff to Tarkowski, who now has to run with Martinelli.
Stop me if you’ve heard this before: the final action didn’t come off. Martinelli didn’t cut back, and Tarkowski hit it out for a corner.
In general, the dynamics on the left looked healthier, as one would expect with a couple of left-footers over there. This kind of triangle has been too rare in these parts.
In that moment, Merino cranked one into the Ø-zone.
That pass is a good encapsulation of Merino as an associator. There is a simplicity and immediacy in his action: the kind of urgent, ‘risky’ pass is exactly what Arsenal lack these days; his read was intelligent, and more passes should be hit to Ødegaard in this area in any case; but also: he blasted it in there, and he doesn’t always provide the easiest ball to handle. It is what it is.
🧐 The return of the Havertz Hitch
Things on the left, nor elsewhere, didn’t pan out.
Aside from simple misses, good saves, and variance, this next moment shows one of the reasons why a goal didn’t go in.
Here, the structure and movement are again promising, and Gabriel is able to hit his fellow Gabriel over the top. To my eye, this is a direct result of improved dynamics over there.
And now that Martinelli has the edge, this is the kind of opportunity that is too hard to generate against a block like this.
Though there is a lot of blue on the screen, this is essentially a 3v2 in front of the goal. All Havertz has to do is immediately drive at the space between the two defenders, attack the near-post in Branthwaite’s blindspot, and pull the line down to the goal. Then, Saka can go to the far post, and Ødegaard can fall into the soft spot that opens up. A simplified version is this:
These runs can also cross depending on the locations (and blindspots) of the defenders and the timing of the arrivals.
Haaland, for example, often looks to go far-post (because he’s tall, can outjump everyone, and can occupy two defenders). But if there are supporting runs elsewhere, look at the commitment of his near-post run here.
In this example against Luton, you can see how Havertz attacking the near-post and dragging down the line creates the spot for Martinelli to pounce at the penalty spot.
In this moment against Everton, though, the “Havertz Hitch” reappears. He never fully commits to sprinting, and sees if he can hang back and receive the ball at his feet.
…he then sees his error, but it’s too late. Branthwaite uses that moment to hang back, jump the pass, and stop the ball from reaching the Ø-zone again. Watch it play out:
Before the subs that degraded the overall dynamics, it was these little hesitations, small mistakes, and minor issues that kept the scoreline blank.
👉 Subs and stagnation: what went wrong
With a well-rested, well-drilled opponent in Everton, the first goal was always going to be hard to find. The hope was to generate something, anything, and then open the game up to goals from there. Game-state impacts everything.
It’s been annoying this year, too. By Understat xG, when the score is even:
Arsenal have scored 14 goals on 20.4 xG.
Liverpool have scored 15 goals on 14.63 xG.
That goal didn’t come. With the initial lineup, that was largely because:
There wasn’t enough ball speed on the passes.
Ødegaard didn’t look fit, but occupied all the right locations. This meant that he was on the end of a lot of several good opportunities, but biffed them.
There was unhelpful variance and a good shot-stopping performance from Pickford.
On the Ødegaard side, we can see just how much of the chances laid on his foot in about an hour on the pitch.
When the subs came on, the wheels came off. I was good with an Ødegaard-for-Nwaneri sub (as said, Ødegaard didn’t look right to me), but Arteta made a real error in going back to the dynamics with Timber at LB and Partey at RB.
After this shift at 69’, Arsenal only attempted one open-play shot for the rest of the game — a game that sure felt like a must-win, or close to it.
This setup falls short for a few reasons:
This left pocket (Timber, Trossard, and [x]) may remind you of those England teams from the Euros with Trippier up at LB and a bunch of 10s all looking to float in the same zones. The pitch gets artificially squeezed and everything is easier to defend in the middle. The profiles don’t complement one another.
The right side has no real action in the wide areas around Saka. Partey plays a bit like Timber does deep, but then doesn’t follow up with carries, dribbles, underlaps, and overlaps, except in spurts.
This is not really about individual performances. This is about asking players to have characteristics they don’t really have. We can complain about a lack of urgency from the players, but it’s hard to be immediate with your actions when the connections and locations aren’t right.
Everton abandoned any impersonation of an attack. They finished with zero shots on target for the day. The full-time whistle bellowed after countless dead-end, no-advantage situations like this.
💭 A familiar feeling early against Palace
For much of the first half, things felt slowish again versus Palace. This was for slightly different reasons.
For one, Kiwior got uncle’d by Mateta and Arsenal gave up a goal almost immediately. Raya didn’t necessarily cover himself in glory, either.
Kiwior strikes me as fairly partner-dependent, as many centre-halves are. It’s hard to expect much more given his placement on the squad list, but we can also see the gap to the starters in such a moment.
You’ve perhaps heard the “cat-dog” analogy for CB partnerships. The dumb one in my head is always the underworld “boss-enforcer” partnership: the boss as a calm but domineering presence, operating with distance and gravitas and intimidation; the lieutenant as an impatient, hands-dirty-agent-of-direct-action.
Kiwior doesn’t neatly settle into either category, but could fare better as a CB at a midtable side (with less high-lining) or especially on the outside of a back-two-or-three in Spain or Italy. If he finds his groove for the right club and coach, he could be a difference-maker. But here, he’s not quite aggro enough for the up role, or quite physically dominant enough for the down role.
For now, the bar to clear at Arsenal is whether the team could win a title with the player as a starter. If Kiwior is an LCB alongside Saliba and a good supporting cast, I think the answer is still, narrowly, yes. His passing range keeps him there. We can upgrade, but we’ve seen darker days.
Anyway. As we’ve learned all-too-well, things get trickier when the opponent scores an early goal, as they get even deeper and more compact with their approach.
Crystal Palace, to their credit, mostly stuck to their gameplan.
Interestingly, Timber started at RCB, but was inverting into the midfield early. This wasn’t a momentary rotation, but a genuine strategy. Yes, that’s the “Stones Role.”
It’d happen even in more “opt-in” scenarios. This put Partey as the RCB in build-up, and allowed Jesus, Merino, and Nwaneri to float around and look for advantages.
The primary setup was mostly this.
With Crystal Palace going into a 5-4-1 defensive shape, Jorginho could patrol the midfield alone as a mostly-lone #6, freeing up attackers high.
With so many new connections, the setup was a little slow, and Sterling didn’t have a lot of help creating chances on the high-right side. But the configuration was not without its advantages; Nwaneri looked nice, and there were moments of fluidity and skill. It was one of those viewings that was more frustrating in the context of the season than for the actual performance.
Even in the first half, we saw the benefit of having a high-and-wide left-back, comfortable in those winger areas.
With improved dynamics on the right, this advantage would compound in the second half.
👉 What transformed Arsenal’s attack
In a near-perfect A/B test, it only took two tweaks for everything to truly erupt.
Those changes were as follows:
Ødegaard on for Nwaneri. Nwaneri was plenty good in the first half, but may be more of a ceiling-raiser than a floor-raiser at this stage. More importantly, Ødegaard looked like he had his legs back.
Saliba on for Partey. This switched Timber over to right-back. (I’m still worried about overworking him.)
Some of this is simple arithmetic: at this stage, Ødegaard is a better football player than Nwaneri, and Saliba is a better football player than Partey.
But it’s deeper than that.
In So Back, we covered how Timber squeezes Saliba at CCB to increase the space the right-wing (usually Saka) has to receive, and also improve the chances of quick movements through the middle.
He’s done it more since. Against Everton, he kept himself available for quick passes through the middle, and also looked to bait the opposing winger up. He can afford this, because he then is able to sprint to join Saka after the initial ball.
This dramatically improves the conditions for an outball to Saka, as it increases the time for the double-team to arrive as he receives.
But here’s the other reason he crowds the CCB: central access. We see him here at the base of a diamond.
That right-sided improvement can lead directly to improved usage of the left-sided width. One hand washes the other.
Do you remember the graphic earlier about attacking specific zones on a cross? The one that was missed against Everton?
Look here, for the Sterling double-chance.
This setup continued to see improved dynamics. This was a reset (so don’t look at the distances too closely). It was nevertheless interesting to see Jorginho accept the ball in the Pirlo spot, and face a double-pivot of Timber and Merino, with a couple of bonus outlets out wide.
In this case, Timber provided central access. This is what makes him such a potent weapon: the same player doing a Saliba impersonation in the first half is looking a bit like Nwaneri here.
🗂️ Box occupation: a status report
In What’s in the Box, there was a series of takeaways, largely about Havertz:
Whatever his overall impact, this team is just not creating enough shots, and we’d like to see his volume increase. He’s currently at 2.46 shots per 90, and Arsenal are currently in eighth (even with Crystal Palace).
But my piece is not meant to argue that he needs to just stay in the box; his wide combination work is powerful, and I wasn’t arguing against it after Sporting. It’s more multi-faceted than that.
When the right side is reunited (Saka/Ødegaard and either White/Timber), they don’t need as much of his help. When he drops, I’d like to see him drop more centrally, and stay in the box a little more.
As he does, the team should target him with more crosses in general.
When he drops or drifts, the team needs to be more aggressive with crashing (with numbers) behind him. The sheer volume of aggressive, in-behind, central, goal-conscious runs can and should increase. This includes Saka, who is excellent and immediate at them.
The team needs to up the risk in a few areas: more low-percentage passes; more attackers in the box; more “lower efficiency” shots. The feel of “chasing a goal late” shouldn’t be reserved for “chasing a goal late.”
Some of this is the Catch-22 with Rice at LCM: you want him patrolling the middle, winning it back, perhaps a little deeper; but you also need somebody up there regularly crashing the box, too. If Rice is deeper, that falls upon the left-back — which increases their running distance and physical load.
Arsenal have one of the most powerful counterpresses in the world; without more risk in the final third, the power of that counterpress is blunted.
I’ve got some high-level data that suggests a moderate negative correlation with the amount of touches that Arsenal take and the amount of xG generated. In short, that seems to imply that Arsenal should play quicker and riskier.
This is all a case for Merino playing more. Crosses from Saka to Havertz, Merino, and Martinelli/Trossard should cause great fear for opponents.
The TL;DR is that this is how you score on crosses:
…and this is how you ensure you’ll never score on a cross:
Against Fulham, with “right-inversion” being used, Havertz was essentially playing the role of a quasi-fullback. A third point in the triangle was necessary out here, so it was Havertz combining, overlapping, and running. That’s not a problem in itself. The problem is that, between Rice LCM and Timber LB, there wasn’t enough pressure in the box when he left.
This really changed against Monaco, in subtle ways.
For one, Partey stopped playing a fully-inverted, “Right Zinchenko” type role — and played more like “White without the overlaps.” This wasn’t revolutionary to the dynamics, but it gave the wide areas a little bit more support.
Then, on the left, there was an actual left-footer, who could support play out wide while being the primary inverted partner. Myles Lewis-Skelly is awfully exciting, isn’t he?
This combination of left-footers (Myles, Merino) allowed the other M — Martinelli — to get off the touchline and explore the space a little.
(Unfortunately, Martinelli got right back on the touchline against Everton, for reasons that are not fully clear to me)
Against Monaco, this balance allowed Jesus to lean left and right to help, without overdoing it. His presence through the middle led to several chances, and showed something that had been missing of late.
Against Monaco, here were the most prevalent connections for forward passes:
Kiwior to Lewis-Skelly (7)
Kiwior to Merino (6)
Partey to Ødegaard (6)
Merino to Martinelli (6)
It felt nice to see that side doing normal things. Lewis-Skelly generated a huge, memorable moment a half hour into his Champions League career, showcasing his skills in quick succession: control, then riding a challenge, then a little carry, then a line-breaker. But there were three important little things to notice: Merino dragging a defender away, Martinelli central, and Jesus staying just central enough.
MLS then hit the linebreaker through the open lane.
Jesus provided the assist on his left foot.
Here’s what I wrote in July.
If you succeed in rebuilding the left-hand-side, you will succeed in getting Saka some tap-ins. More low crosses, deflections, rebounds, etc. It's about time that man got some easy goals.
In all, Jesus has done really well positionally.
Against Monaco and Everton, Rice served as the base. He’s generally looked great at #6, and has been sneakily progressive throughout the year.
Things didn’t go quite as well against Everton. It’s difficult to decouple instruction from temperament. He can sometimes reflect a team’s tempo instead of dictate it, and is still evolving and improving there.
Still, he played well, and Rice wasn’t really the problem.
In the below setup, but with Rice in the #6, there’s enough vibrancy and ideas around him to really propel things forward. The team is yet to lose an 11v11 game started by either Ødegaard or Calafiori, let alone both. That “floating and fixing and exploiting” role that Ødegaard plays on the right is often taken up by the full-back, not the LCM on the left.
But back to the subject at hand. After being dissatisfied with the box occupation and the number of runners in previous matchups, I kept a keen eye on it against Everton.
… and it was fine.
This wide triangle got plenty of action in different configurations, and there were usually ample runners in the box.
Havertz generally stayed central. When he floated, Merino was usually there to make the run through the middle.
It was most notable here. Havertz floated right and deep, and showed why he shouldn’t be confined to the box. He did a flick that wound up being a full through-ball to get Saka behind the line.
Then, as Saka hit the byline, there were three runners collapsing on the box, just as you’d hope.
Ødegaard just missed.
Havertz looked a little more tired than usual when switching on to attack final balls.
The end.
⬆️ Attacking the middle
This chart, via @dataanalyticEPL, doesn’t make for pretty viewing.
This is clearly, largely, tactical. Arsenal are only above two relegation teams in this.
You’ll notice this stat is not a perfectly clear predictor of success. Man City lead Arsenal in Zone 14 passes by a total of 545-145, but carry a worse expected goal differential (+6.3 to +12.6 at the moment). Arteta overly accounts for the risk of counters in this area, and looks to build attacking patterns out wide.
It feels obvious that this pendulum has swung too far in the direction of caution. Some of that makes sense in light of Arsenal’s backline injuries, but if we’re to win … things … we’ll need to see more creation and shots from here.
This has seen some marginal gains of late for two reasons.
The first is that MLS actively looks to play around in these areas.
Against Everton, Ødegaard was seen popping up in these areas a few times.
But the other reason is some newer structural dynamics. Along with Jesus’ finishing heater, this likely won the game against Palace.
Arteta didn’t seek to shoehorn Tierney or MLS into preconceived roles. By playing Tierney out wide, Trossard essentially became the “left-10,” floating around, and playing a role that was inherently more comfortable to him. Once Ødegaard came on, this meant Merino pushed in even further, occupying a zone that you don’t usually see filled by an Arsenal player.
This had all kinds of impacts. By freeing up Merino to go left, right, and forward, he was driving around to pull man-markers in this scheme. Palace had five on the backline, so this created a 6v5 overload, and forced Palace to keep an eye on him. The improved right side took full advantage.
Here, Merino sees Ødegaard cutting to the left, so he makes a run to pull Maxence Lacroix away from the action.
This run creates the lane that Ødegaard needed to feather it through. A nice body feint from the captain doesn’t hurt, either.
…and we all know how that ended.
We continued to see the value of that high-and-wide help on the right.
As Saka settled in for the last twenty minutes, and Lewis-Skelly came on for Tierney, we saw the continuing value of the updated shape.
Lewis-Skelly continued to play like Tierney, pushing up instead of fully inverting into the midfield, and this enabled a series of cascading rotations: Trossard into the half-space, and most notably, Merino floating through the middle and right to overload. That, to my eye, was the key: Merino is serving the role Havertz would offer — but there’s still a striker up there.
Goal number two.
This is what we continue to see: if we’re able to overload sides, and finally access the middle, and retain a striker presence, the goals follow. Easier said than done.
🔥 Some final reflections
Arsenal are one of the very best teams in the world. As I tire of repeating every week, much of the frustration is for circumstantial reasons: injuries begetting injuries, a play-style requiring intensity, some bad variance, the reds, all that. Meanwhile, there hasn’t been an offsetting scoring explosion from anybody. I went through some tape from last year, and was struck by how many games would look different without those seventeen Trossard goals. On Wednesday, Jesus gave us a reminder of how that feels.
It’s also easy to suggest that everything can be cleared up with a January transfer. There’s some truth there: we sure do seem to be short an attacker with that highest level of dynamism. Saka is doing all he can from a creation standpoint.
But it will be difficult, and it doesn’t mean that Arsenal can’t continue to improve plenty in the meantime.
Here are a few more improvements that can happen in any case.
Tactics and vibes go hand-in-hand. If you want to play with more purpose and urgency, you need to put players in the best position to do so. After doubling down on some recent tactical learnings, the players need to play with more ball speed and risk.
Moving forward, it should take a train to bump Timber out of right-back. Moving him to LB should be viewed as an inherently defensive move. From a balance perspective, I’d likely prefer Timber at RB and one of 5-6 LBs (Calafiori, Zinchenko, Lewis-Skelly, Tomiyasu, Kiwior, and Tierney) over the Timber (LB) and Partey (RB) duo. Exceptions can be made for emergencies or Mo Salah, but I repeat myself.
So much comes down to that left-back position. It may be frustrating to hear that Calafiori was an attack signing, but as I go through some of the underlying stats, his early impact can’t be overstated. He should be kept in bubblewrap. Overlaps are effective, sure, but more than that, a simple comfort in the highest and widest area can stretch the pitch and allow other players to get into more comfortable, less crowded positions — this can include Martinelli/Trossard (pinching in), the striker (more central), Merino (leaning right and central more), and Ødegaard (floating deep more). Lewis-Skelly offers hope, too. It’s all interlocked.
I was really intrigued with all that central floating that Merino did. Ødegaard and Nwaneri can offer that as well, of course.
On the topic of speed: Liverpool’s famous throw-in coach Thomas Grønnemark says that throw-ins should be “long, fast, or clever.” Clever takes some time, and we so often reap the rewards of the extra care put into set pieces. Nothing should change on corners, but especially in even game-states, we need to make a more concerted effort to keep the ball in play and moving otherwise. If you believe your team to be qualitatively superior, you want as many live-ball situations to prove it; otherwise, the variance monster gains its awful strength. Arsenal foul too much (joint-fifth-highest in the league; that number should be near the bottom), and can take too long otherwise; Lewis-Skelly took ~30 seconds in a second-half throw-in against Everton, only to throw it backward so Merino could clear it. Down 1-0 against Palace, Jesus earned a free-kick at 29:14; it was 30:54 by the time of Sterling’s shot. All of this means that Arsenal are down in ninth in live-ball touches. Through strategy, urgency, and game-state, there are four Premier League teams with a lower percentage of possession who have more live-ball touches. We shouldn’t time-waste our own precious time.
We’ve seen the striker stay a bit more central, and when he drops, more other runners rushing in. Now that some of the “box occupation” issues are improving, there needs to be increased usage of that box occupation — more aggressive runs in-behind, more whipped crosses to the likes of Merino and Havertz. I think the coordinated, ‘automatic’ runs still look like they can improve.
I’ve written this a million times, but: trust the rest shape more. You can’t squeeze the opponent and regain the ball in advanced areas if you never lose it yourself. Hit it in there. Be willing to lose it.
Arsenal have the closest average shots in the league (14.6 yards), but fall to seventh in xG per shot (0.12), and are ninth in shots per 90. This can be partially attributed to deep blocks, crowded boxes, set pieces, etc. But also: I think there’s a general ‘prep touch’ problem. Need more first-hit rips. More shots, slightly ‘worse’ shots, quicker shots.
I’d also suggest using Gabriel like Spurs often use Romero: throw him up there in the box for longer periods, and have somebody like Rice rotate down with Saliba to chase down counters. I always believe in doing what the opponent wants you to do least. Having Gabriel up there begging for crosses would scare the shit out of me.
For a team with so many physical advantages, Arsenal are too vulnerable on opponent set pieces: there’s a miscommunication or mistimed jump nearly every time. I chalk most of this to all the backline rotations. But you’d still like to see it improve.
The attackers need to cheat more, like Jesus did for the third goal.
Thank you for reading.
A rematch beckons.
Be good.
🔥❤️
Made my day. Thanks and bless you, Billy
I think Calafiori and Zinchenko are a massive miss. Zinchenko is not perfect - we know this - but he makes things happen and I like that. Cala is awesome but seems to be fragile but his natural ability is quite something...and Zin is his fill in. There lies the problem IMO.
I don't we are 'fine' I think we need more disruptors in our line up.