What went wrong in the second half at Leeds, and how to stop it from happening again
In last week’s preview of what to expect in the match at Elland Road, we went through the various ways that Arsenal might attack Leeds: from navigating their press, to exploiting their ball-hawking tendencies, to forcing them into standard buildup, to threaded passes.
The final note was to recognize the battle of attrition. Particularly in recent weeks, and most acutely against Palace, Leeds has shown a tendency to burn too brightly in the early minutes and lose steam from there. (I believe that’s what they call a mixed metaphor). Half of their goals have been conceded after minute 61.
So much for that. As we wrapped up the preview, we drank responsibly and added a note of caution:
If one were dabbling in the trade of expectation management, this does have a few of the characteristics of a trap game. A rested, aggressive, mistake-inducing opponent is hungry for a result at home — while Arsenal is coming off a midweek trip to the Arctic which featured their most lethargic performance of the year, albeit one with every caveat imaginable.
Anyway.
The first half
This period didn’t defy any reasonable expectations. It featured a fast-paced, exciting game with Arsenal’s talent edge resulting in a 1-0 lead. Leeds had energy and chances, but one Rodrigo mistake (and a Sicko Saka finish) cost them dearly.
There was one point of intrigue, though. By their standards, Leeds was actually giving the Arsenal backline space to operate. The high line, usually so frantic for early stretches, was decidedly more calm: their 16.56 first half PPDA (passes per defensive action) marked their lowest-intensity line of the year.
Below is an example of Leeds in defense, which can be in the shape a 4–2–4 or even 2–4–4 at times, but was often more of a narrow 4–3–3 in the first half. After starting a little more aggressive as a team, Aaronson often dropped back, they allowed possession in the Arsenal third, and then had relatively relaxed pressing triggers (occasionally turning it up if there was a heavy touch, or the ball was played out wide).
As aggressive as it often felt, that was the Leeds version of calm. The storm was next.
The second half
As Patrick Bamford came on for the second half, things immediately felt different in ways that his ruffled appearance could have never foretold. The press turned up instead of down, a disallowed goal followed, and then so did chance after chance after penalty after chance, with precious little flowing in the opposite direction.
As you’ll see below, the starting shape (three up front, one forward dropping back a bit) was familiar, but the positioning was higher, and the exhausting and tireless Aaronson played a more integral role.
Pivotally, the Leeds press triggers had more permissive rules of engagement.
Saliba and Partey play and back-and-forth 1–2…
…and that passing is Aaronson’s trigger. He is now all-out-blitzing the ball, and Bamford is watching close for his key to unlock...
…and there it is, Saliba passes it, and Bamford is now running all-out at Gabriel…
…who has to pass it right back to Ramsdale, who Bamford also hounds into a clearance.
This kept up for most of the half, as Arsenal fans need not be reminded. The decision by Jesse Marsch to subvert tendencies and calmly hold position in the first half, and max-press a tired opponent once 45 minutes had already passed, turned the game on its head.
It sure felt like the worst Arsenal half of the year, and as you’ll see below, numbers provide evidence to that claim. It was a stressful watch, and the league leaders were fortunate to escape with three points. But the good teams tend to be the fortunate ones.
It’s human nature to point to the obvious. As much as we’d like to chalk this up to a tired team facing a rested one, it’s not the last we’ve seen of that situation, and is something that’ll have to get figured out.
Arsenal should have pushed the ball more, Arsenal should have passed more, Arsenal should have shot more. But those are ultimately lagging indicators. The real problem lie in the inability of the team to sustain its #identity of competent, confident possession — which it did against Liverpool, but not against Leeds.
Why was that? Let’s dive in.
Comparing this second half to a baseline
What was different? Oh, almost everything. To understand the difference between this second half and all those that preceded it, I pulled two sets of numbers:
A baseline: this is an aggregate of all second halves to date, in and out of league. This goes from Palace to Bodø/Glimt, and runs the gamut from battling high lines (Liverpool), to low lines (Tottenham), to Arsenal’s own parked buses (2H at Bournemouth), to the loss. After the first half, the average score is 1.23 to 0.31 — so roughly the same differential as the 1–0 lead at Leeds.
Arsenal v Leeds: this contains all the numbers from Sunday. From there, there are two delta columns that show the difference between the baseline (i.e. all second halves before this) and this game. I highlighted some of the ones I found interesting.
Okay, here’s the big, ugly spreadsheet I created:
What will jump out is… all of it. Most notable are the deltas in xG, possession %, and passing. Arsenal passed 89 fewer times than they usually would in the second half, and had 5.45 fewer shots.
What is surprising is some of the efficiency stats: Arsenal was even more efficient in defensive duels than usual. They just couldn’t get out of their own third, losing the ball in dangerous spots. As the shady realtor once said, it’s all about location, location, location.
So, why did Arsenal struggle so much?
From here, we'll be focusing less on defensive mistakes in the box — because “don’t swat the ball with your hand, Saliba” doesn’t make for an interesting read—and more on why Arsenal couldn’t effectively maintain its identity in possession.
Let's go through five factors.
🍗 1. More ball losses in defensive third
As you see above, the overall number of ball losses was not aberrational. The team’s 54 second half ball losses were fewer than they had against Zurich, for example.
Where it was bad was the location of those losses. Arsenal lost possession in their own third 20 times, which is twice as much as a usual half.
Here’s an example.
Arsenal looks to play out the back. Ramsdale plays it to Partey, who one-touches it to Saliba…
…and Saliba probably should have just played it back to Ramsdale. Instead, Sinisterra does a really good job of going wide to cut off the angle to White, and Bamford cuts off the angle to Partey (who is not actively seeking position)…
...and the ball trickles off his foot. Bamford cleans it up and goes on to shoot it directly into Ramsdale’s ramsdales.
It happened again shortly after.
In another attempt, Jesus drops deep in an attempt to mix up the Leeds midfield press. Saliba feeds him and he’s immediately trapped.
The idea is good: use a new player to drop into the midfield, attract ball-watchers, and then use their attention to break the press. If he one-touches it to White, it’s off to the races, but he goes for an ambitious dribble instead. They all encircle him, Jesus isn’t quite as refined as usual, and a Roca tackle leads to a Bamford opportunity going the other way.
🍗 2. Low recoveries in middle/attacking third
Here’s a map of all second-half recoveries:
To oversimplify, the reason high counter-pressing is so effective is because a) you usually get the ball closer to the goal and b) you usually can attack a defense when they are out of shape. This is especially true against Leeds, who has struggles in normal buildup.
In all, Arsenal only had 9 recoveries in the middle third, and only 1 in the attacking third. On average, they usually pick up the ball in those combined spaces 12 more times per half than they did on Sunday.
Looking through the tape of all these, so many were bouncing balls from off-timing Arsenal passes in attempted buildup: the left-flank having trouble imposing itself, the right-flank awkward. On that…
🍗 3. Failed progressive passes
Arsenal’s attempts at progressive passing were right on par with their normal amount (~34). The problem was the success rate dropping by 14.2%. That’s 4.3 fewer progressive passes that may have led to more stable possession in the advanced thirds.
There were plenty of little moments like this.
White receives a ball from Ødegaard and has nothing but free space in front of him…
…he takes advantage of the real estate, and does the right thing with a progressive dribble. He passes it out on the wing to Saka…
…and puts it behind him. Saka has to gather it and stop the attack. Leeds has possession again within seconds.
Here’s another frustrating one.
Partey had just won possession on a header, and coordinated well with Ødegaard. With open field ahead and a step on his defender, Ødegaard goes to play him into space…
…but Partey hadn’t sensed the same opportunity, and stops running.
These examples of unrefined passing coordination had two effects: missing the direct opportunity to create a chance, but also missing a chance to calm down in the opponent’s side, pass the ball around, probe, and dominate possession like they did in the later stages of the Liverpool game.
Technically speaking, these missed passes were buzzkills.
🍗 4. Slow feet against triangles on the wing
Arsenal got a taste of their own medicine, with multiple chances originating from athletic tight-space dribbling, overlap/underlap, 1-2 stuff on the left-wing.
This created the disallowed Bamford goal within 30 seconds of the half starting. Here’s to hoping this screenshot makes sense.
Aaronson is double-marked, still gets a deflected ball off to Sinisterra, who one-touches it to Struijk, who one-touches back it to Aaronson, who passes it back to Roca, who delivers a cross to the far side of the box.
Other than pressing, this is the area where the energy deficit really reared its head. I could find nothing majorly wrong with the positioning/tactics of anyone on plays like this. The deflected balls just usually bounced towards Leeds and the Arsenal feet where a fraction of a second slow.
🍗 5. Zinchenko was missed
This is more editorial than capital-A Analysis, but watching back, I saw more struggles for Tomi than I caught on the live viewing. While he helped solidify things next to Gabriel, he simply had a worse game than he did against Liverpool.
Right before the half, he got mugged by Aaronson, leading to a Leeds shot.
This may have influenced his play in the second. After 45', he struggled progressing the ball under the press, and lost the ball eight times in the Arsenal half. That was more than any other game this year, and more than the rest of his league games combined.
These weren’t catastrophic losses — under pressure, there were times when a pass was available (like below to Partey) and he’d basically do the “safe” thing and clear it.
To note:
He won half as many duels overall (including 0/4 on offensive duels) and recovered the ball half as much. His world-elite skill in aerial battles and patient, shut-down defending were less advantageous in a game like this.
It impacted Martinelli’s effectiveness. Martinelli could roast the Liverpool LB situation by his lonesome, but Kristensen could largely keep up on Sunday, and it led to some lonely aggressive dribbles into the corner flag with no friends in sight.
Nothing against Tomi. After all, I just wrote about his brilliance in the position ~6 days ago, and there are all kinds of matchups that are better for him. But we’ve got a world-elite press-resistant possessor in that spot, and this would have been a nice game to have him.
Some other (non-Tomi-related) things I noticed upon rewatch:
The Arsenal press strategy was different in ways I’m not sure I’m fully understanding. I have to wonder if Arteta was just giving the guys a break. Instead of sprinting at CBs and LBs alike, Martinelli could be seen at the halfline with a full-back while Jesus was near the keeper. Didn't see much of the 4-1-4-1 stuff we saw against Liverpool.
White let a couple throughballs by him that were uncharacteristic.
🔥 In conclusion🔥
Finally, we’ll do the part I’m least confident about: how to avoid this from happening in the future.
My confidence wanes because this involves acting like I’d know better than Arteta, and he’s a world-class athlete and manager and I have trouble walking up the stairs without spilling my coffee.
But let us not be deterred! If Arsenal were again to facing a packed schedule and a high pressing, high-energy side, how could they avoid a repeat of the Leeds second half?
Attract traps and pass it out in one touch: As we covered in the previews for both Liverpool and Leeds, press-traps can be your friend. Passing it to a player, triggering a trap, and then one-touching it out of there and passing it through the vacated zones is a powerful weapon. Arsenal was good at attracting the trap, and in the first half, were good at one-touching it out of there as well. As the game progressed, they were more hesitant.
Invert Tomi and White earlier in possession, and drop Ødegaard or Xhaka back and out wide: In the second half, there was a lot of by-the-book buildup against a narrow press trap: Tomi and White were often standing on their respective touchlines to stretch play. This led to a lack of tenable options for Saliba, Gabriel, and Partey — Partey in particular had few options in midfield, and the gap between him and Ødegaard felt too big at times. Xhaka was AWOL from build-up. They were more successful in the first half with a full-back inverting, and one of the advanced midfielders dropping back, like when Ødegaard would pop back into White’s positions. His class on the ball can be so helpful in press-resistance.
Slow down in the attacking third: Once Arsenal got around the press, they immediately tried to create chances, which was understandable, but they'd often immediately lose the ball again because they didn’t have the possessional rhythm they’re used to. I would have preferred them to slow down and dominate, like they did in my favorite stretch of the year: the second half at Liverpool.
Earlier subs: Thought Arteta made all the right subs, but with Jesus not looking entirely like himself and the full-backs struggling to progress the ball, I couldn’t help but wonder what Eddie and Tierney would have done in the flow of a game that still had more attacking possibilities to offer. Eddie came on at the same time Arteta went 5-back, at 82' ... in this one, I would have been fine if that was 60'.
But really, here’s what I’d recommend to Arteta:
Make sure Zinchenko doesn’t get injured any more
Stay out of the Arctic Circle in the days immediately preceding a game against a bunch of rested lunatics at home
OK, that’s that.
May you one-touch it out of the press-traps in your life, and may your opponents’ penalties sail wide.
And happy grilling, everyone.
🔥
-------------------------------
Support this work
If you’d like to support this work coming out every week, I just started a Patreon. I am so appreciative of those who have signed up at £2/mo to keep this series rolling, and if those numbers grow, I'll start looking at other mediums (newsletters, video, etc). Thank you all.