What's in the box

Assorted thoughts on the current moment's aggravations, tactics, and lessons. Then: Kai Havertz and one area that can boost open-play chances: more presence, runners, and risk-taking in the box

“I accept chaos. I'm not sure whether it accepts me.” — Bob Dylan

---

You probably know that F. Scott Fitzgerald quote: “The test of a first-rate intelligence is the ability to hold two opposing ideas in mind at the same time and still retain the ability to function.”

After the torturous draw against Fulham, I’ll ask you to go a step further. Here are twenty or so takes I have about the team at the moment, some of them opposing, some not:

• Arsenal aren’t creating enough from open play to win a title.

• Arsenal have had some shit fortune so far — primarily through injuries, fixture difficulty, and red cards. The fanbase is annoying about the reds; we won’t let them go; this does not make us wrong.

• Missing Ødegaard was one Big Thing that was wrong, perhaps the Main Thing, but absolutely not the Only Thing.

• Across competitions, Arsenal still haven't lost a game Ødegaard started this season. 6 wins, 3 draws, 0 losses; 22 goals scored, 6 goals against.

• There are some legitimate tactical concerns about this team, no matter the individual performance or injury issues. Some of these have persisted for years. In particular, Arteta’s risk assessment in the middle of the pitch feels off.

• The inversion of Partey (RB) and Timber (LB), though driven by necessity in this recent case, is not ideal, as we’ll cover.

• There is no difference in the value of an open play goal and a set play goal. They are both worth one (1). As such, there is no meaningful reason to overly stress that a team shouldn’t rely on a particular type of goal; no team should rely solely on any type of goal — cutbacks! dribbles! worldies! headers! ahhh! — because nothing is permanent, we are but scattered leaves, and a singular mode of scoring may not happen on the day. But if you’ve got an advantage in a department, why shouldn’t you use it? (I, for one, am sick of Liverpool relying on Salah.)

• Arsenal were probably too conservative and ‘disciplined’ with the summer window. We are clearly short a dynamic, explosive attacker.

• If the biggest attacker/midfield swings weren’t available, they were right not to lurch. The wrong big signing can set you back years. The midfield signing (Merino) has played all of ~350 league minutes at full-strength, and has started one (1) time alongside Ødegaard.

• Much of the team’s dynamism hinges on its fullbacks. It’s fair to question whether the “ask” of them is sustainable.

• Signing a brilliant centre-back with some injury concerns, and asking him to move to a famously physical league, then having him increase his intensity and play a more demanding role, however brilliantly, to play ahead of a host of famously-injury-prone players, may have created an illusion of depth.

• Shipping off attacking midfielders, and leaving space for Nwaneri, is justifiable; Man City learned the risks of the alternate path with Palmer. Being conservative with Nwaneri’s rollout, considering his age, is utterly justifiable too. But it will cost you points in the short term. Those could be enormous points.

• On most 11v11 metrics, Arsenal are still elite.

• Even just in general, Arsenal have the third-best expected goal difference in the league (+0.77 per 90) and the lowest xG allowed (0.94 per 90), despite all the fuckery.

• There are very few instances of Arsenal being outplayed at full strength. Newcastle definitely qualifies. There is a batch of others that were even-ish (from memory, you can pick your own selection of maybes out of Inter, Atalanta, Liverpool, Man City, Spurs; I personally remember the Arsenal 2-0 win against Villa as even, too). Bournemouth definitely deserved the win if you considered that red avoidable, as I basically do. But again, that is a short list, with only one-to-two clear “yeses” on it.

• A draw like Fulham is particularly gut-punchy in the context of the season. In other contexts, a single footballing is more easily understood.

• Since the last international break, Arsenal have outscored opponents 16-4. Trophies are still a real possibility, even if the league is uphill.

• Life is pain. Shit happens.

• ...and so on.

The day before Fulham, I said this.

The overall direction is mostly quite positive, but there are still lingering questions: specifically, the current defensive depth, and whether Arsenal can be vibrant and penetrative enough through the middle, especially without Calafiori … Fulham have been one of Arsenal’s trickiest opponents of late. This time, they bring along some more familiar faces. If there’s any silver lining from the early-season travails, it’s this: the mandate couldn’t be any clearer. Three points, every time.

The “current defensive depth” question reared its disgusting head. Three points didn’t happen.

Here’s what I wrote after the Sporting demolition.

But … 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. They are absolutely capable of that.

Fack.

In truth, I was gloomy as soon as the lineup came out. Seeing Zinchenko go out, again, to join a total of five other starting-level defenders on the injury list, forcing Partey to RB and Timber off to LB … this was rough to process.

But don’t fear! It can still get worse. The remaining two, Timber and Partey, both missed training on Tuesday, and may or may not be available for today’s match.

If you can’t complain about having to discuss Martinelli wing-back shifts and starting eighth-choices in the Champions League then when, precisely, can you complain? (I’m talking about us, not the team).

Structurally, the issues we saw against Fulham were nothing new. Many of the conclusions from this piece in March still apply, sadly:

---

---

But we move.

Today, with the Monaco game a-looming, I want to try and defy the odds by writing a quicker one.

Let’s hone in on one area of possible improvement.

---

---

💸 The cost of that Fulham setup

Here was the shape against Newcastle, in what was probably the worst performance of the year.

Outside of missing Ødegaard, we outlined the problems then:

• With Partey inverting deep from right-back, Saka is now all alone. He gets a dose of what Martinelli usually feels like.

• Timber is playing deep “LCB.” With Partey inverting by default, Timber is hanging around in a back three and finding specific moments to push up. But underlapping is tough up there in that situation, and even if he were to overlap Martinelli, he’d be wrong-footed and want to cut back.

• When inverting from right-back, Partey will have to cover wide areas on a counter, and would logically be more risk-averse as a passer as a result.

• Gabriel, not Saliba, is at CCB.

The short version is that neither full-back will help enough with wide combination play, except in spurts.

With few options on the table, this setup was, again, the case against Fulham.

Of note:

• On the left, there was a triangle of three right-footed players who don’t particularly want to stay wide: Trossard, Rice, and Timber.

• On the right, Partey was inverting deep; he was more like a conservative, inverse version of the Zinchenko role.

With Partey moving into the midfield, affordances have to be made so that there are still overloads on that right side. Ben White is no longer out there holding width, so Ødegaard essentially slides in.

Again, using the Zinchenko example, this is all the equivalent of the 22-23 dynamics: Ødegaard (Xhaka) drops to receive wide, and Havertz (Jesus) floats over to make sure there are sufficient numbers.

(If you remember one of the problems with the 22-23 dynamics, you’re already predicting one of my conclusions.)

In all, it resulted in a passmap like this. I’ve warned about the dangers of over-indexing passmaps, but in this case, I think it’s fairly indicative of what we saw.

When my priors are confirmed, passmaps are good:

Go follow @markrstats

---

It’s fairly clear what is lost in this setup. Timber can play a solid left-back, and should definitely be considered for starts over there against, say, Mo Salah. But this kind of right-back play comes from nobody else on the squad, and few others elsewhere.

With less activity around Saka, these outlet balls are easier to jump.

This has been the case across setups, and I think it’s time to start faking these and then playing Saka in-behind. Opponents have gotten too comfortable anticipating them.

See, here’s another.

Here’s what a wide triangle looks like in healthier days.

Over the years, this has reliably been shown to be one of the most potent mechanisms for Arsenal to create goals: Ødegaard drops low, White goes high, and Saka marauds in the half-space. This is how that Bayern goal was scored, for example. Overlapping is important, but it requires certain conditions; this simple ability to be a fluid passer and mover wide is probably even more important.

With inversion coming from the right, that triangle still has to be created, but with different positions filling each point. Generally speaking, this is how it’s been done.

This is not a problem in itself. Saka and Havertz have a great understanding, and the play over here is generally nice and crisp. But there is a cost.

---

👉 Presence wide + in the box

Here’s the opening goal against Sporting. Look who is out wide. And look who is in the box, almost getting his foot on the end of this cross before Martinelli does.

Late against Brentford last year, here is when Havertz started turning his season around. Look where he is (attacking a back-post cross with force) and look what is around him (numbers and other runners).

This season’s scoring opened up with a post-corner phase in which Gabriel, Partey, and others pushed up in the box. Again: Havertz is alive up front, the rest shape is pushing up to support a ball loss, and there are teammates running off him.

You probably see where I’m going by now.

With Havertz doing so much wide work, here’s an early example of what the occupation of the box looked like early against Fulham.

…and here’s another more glaring example.

Now, these are just screenshots, to be fair. Some of them don’t indicate how quickly Havertz is able to regroup and re-enter the box with a head of steam. And many of us had grown tired of that mechanical style of play where every lane is dutifully occupied, but things stagnate. This kind of expansive, snappy interplay has often been a sight for sore eyes. But there’s also a reason why that style of play — the 2-3-5 that goes back, oh, 130 years or so — is a thing. Arriving is nice, but there is a cost to leaving key spaces vacated.

Saka’s passmap tells a pretty clear story of which passes hit on the day, and which didn’t.

Saka’s passmap vs Fulham

Those are the hard ones, to be sure, and several of them turned into corners. But there’s a lot of gray dottedness in a certain area.

You can also see it with Kai’s received passes.

Havertz: passes received vs. Fulham

In all, the team only hit 3 of 19 crosses. Here’s the cross map from CannonStats:

That’s a tidy little narrative:

• The right-back isn’t available to combine as much as is needed in the high-and-wide areas

• …so Havertz floats over there to fill the void

• …and vacates the box as a result

But there’s always nuance.

---

👉 …not too fast

The first bit of nuance is that Havertz is good at that shit, and Saka loves a runner to play with. This one almost hit for a simple reason: with Havertz gone, teammates flooded the box in his absence.

That one actually shows why I like considering Havertz at the LCM against mid-blocks: somebody like Jesus or Trossard can do the wide combination work, then it’s Havertz attacking the far post.

…but my second point is that Havertz was still in the box plenty, and there were still moments where the runs were sufficient, but they were defended well, or just didn’t hit.

Meanwhile, the following play shows the potential and limitations of the current lineups.

Trossard is able to combine quickly, and Havertz is able to go Full Firmino with a pass, but the winger isn’t then able to get the edge. False-nines love to play with aggressively crashing, whip-fast off-ball runners behind them; Son or Vini or Mané get the edge here, as Martinelli may, and Salah is even more rapid and viciously goal-hounding from the right.

The advantage of a false-nine is how helpful they are in creating wide overloads, and how much space they can leave for others if they take the opportunity. Here was a great example at Craven Cottage.

These kinds of runs are available too; they are more readily available in this setup.

As the game progressed, Havertz occupied the box more. In the second half, balls started pouring in from the left; the so-close winning goal didn’t just happen, it happened via attrition.

Things were better for a simple reason: Arsenal took risks and pushed attackers forward. The box occupation started to look like this by default.

On this should-have-been Jesus chance, Merino is right there, offside, drawing defenders forward and simply giving them more to deal with.

In the Saka goal that was ruled out for the build-up offside, it’s most useful to watch Merino.

He drops to receive and plays it forward, but he’s basically interpreting his role as a central false-nine at this point. He sprints up to the striker position right after.

…and now he’s up there, and Arsenal have adequate numbers in the box: four players going after it. Goal, Saka:

If this were a lonely Havertz, or Rice trying to score while Havertz is out wide doing interplay, it is probably safely dealt with. You need quality and quantity.

Whatever can be said about Ange Postecoglou, he coaches some of the best attacking patterns in the world. The corresponding runs and passes are second-nature to his attack; posts are covered, and cutbacks are routinely hit to the penalty spot. This is not without cost. You cannot separate his willingness to fling attackers so far forward with his team’s struggles with hemorrhaging goals.

Arsenal have some advantages here. Arteta is simply better at coaching defensive shapes than Postecoglou. For another, Arsenal are a far more physically robust side, led by a triangle of Rice, Saliba, and Gabriel.

It is time to up the risk profile.

---

📊 What the research says

Carlon Carpenter (no relation) is the Head of Analysis for Houston Dynamo FC. Using tracking data, he wrote one of the best pieces I’ve seen in a while:

“Final Third Movement in the Austrian Bundesliga: Quantifying & Evaluating Off-Ball Threat.”

You should read it.

It’s so good that I took notes throughout so I could remember some of the many key insights. Here are six pertinent ones for the topic at hand.

1. Volume of Runs in Final Third Correlates with Threat: Teams that make more runs in the attacking third create significantly higher xT (Expected Threat).

2. Specific Run Types Are Key: Runs to attack crosses, and runs in behind the line, are the two most valuable types of movement in terms of goal-scoring potential.

3. Aggressive Passing is Not Optional: Teams that attempt low-percentage, aggressive passes into dangerous areas benefit more in terms of xT and scoring opportunities, especially when paired with savvy off-ball movement. It's tempting to call this "risky." You know what else is risky? Not scoring goals.

4. Speed Matters: High-speed runs (at least 5 m/s) are directly correlated with higher xT. Teams averaging around 4.5 high-speed sprints per possession phase (final third) saw xT increase by a third.

5. Tight Team Compactness Supports Attacking Runs: Teams with compact “rest” structures in the attacking phase (e.g., short distances between midfielders and defenders) are better equipped to counter-press and create second-chance opportunities after a ball loss.

6. Rewarding Off-Ball Runs Encourages Productivity: Teams like Hartberg demonstrated the value of consistently targeting players making dangerous runs, regardless of pass completion rates.

There are many more. Again, you should read the whole thing.

---

🔥 In Conclusion

Havertz is his own thing. He is not Firmino, nor Haaland, nor Fellaini, nor Joelinton — he is some strange alchemy of them all, plus something all his own, and should be treated as such.

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.

In the short-term, there will still only be so much to learn. This style of play increases the running load at a time when that is difficult to envision. There is likely to be another makeshift backline tonight, and the team will do their best, and we’ll be talking through a few asterisks afterward, whatever happens. As I’ve shared before, Monaco are one of the most impressive squad-builds anywhere, and defensively astute. It’ll be hard.

Once we accept chaos, we can see whether it accepts us.