What on Earth is up with Arsenal?
Can anyone be sure if this team is good or bad? Let’s dig into it.
This month marks two years since Mikel Arteta took over as Arsenal boss. Those two years are some of the strangest in the history of the human race, so it hasn’t been the smoothest of long term integrations. Still, of his eleven most used players this season, only one (Pierre-Emerick Aubameyang) regularly featured under the Unai Emery. This is his team.
He has been backed in the transfer market, though not always in the most coherent manner. Technical director Edu has overseen a mix of solid acquisitions and really bizarre deals. Perhaps this is by accident, or maybe he intended to do it regardless, but it’s led to Arteta relying more heavily on academy talents, which has actually worked out pretty well. After being an “old” team for a few years, Arsenal now have the youngest average age weighted by minutes played in the Premier League. This Arsenal side is new, fresh, and very much moulded in Arteta’s image.
But is it any good? No one seems to be quite sure of that. They sit fifth in the league table, which should never be the destination for a club like Arsenal, but obviously represents an improvement on last year’s eighth-placed finish. But that’s been marred by some terrible performances against big clubs, creating something of a negative mood. What represents Arsenal’s true level? Let’s dig into it.
A note on numbers: stats in this article are from FBRef using data from StatsBomb, except for PPDA, which is from The Analyst using data from Opta.
When people have been arguing about Arsenal this season, it’s really been about distribution of results. The Gunners are in fifth place with a goal difference of -3. Plenty have tried to make this a conversation about expected goals, but in this instance, xG doesn’t actually tell us anything we can’t already see in the results. Against what increasingly feels like the Big Three, Arsenal have scored zero goals and conceded 11. Against the rest of the league, they’ve scored 17 and conceded nine. It’s a mean vs median issue. Arsenal have been good against all but the very best, but those games against the elite have been so terrible as to drag the mean down.
Michael Caley of the Double Pivot Podcast recently did a study looking at the sort of “high volatility” teams like Arsenal, and he didn’t find a great deal of evidence that discarding these sorts of results improves predictability going forward. “There is no correlation between having a high standard deviation early in the season, and having a high standard deviation through the rest of the season”, he explained. “Teams like Arsenal tend to have much more normal results as the season goes along”.
Look, I know what some of you are thinking here. “But Grace”, you’re probably saying, “football isn’t played on spreadsheets. You can’t explain all of football with a few studies”. And you’re not wrong. There are “soft” factors to consider. Arsenal are a young side with little experience of big game pressure. Does that make this group more likely to mentally crumble when things go wrong, even if the general standard isn’t bad?
Well, the data says some things on that. When breaking down xG performance by game state (thanks to James Yorke, Director of Football at StatsBomb), we can see Arsenal have been a bit weird. They’re clearly the worst team in the league when behind (that’s the impact of those defeats to the Big Three), but pretty good when things are level, and league average when ahead.
I’m reluctant to say this is all fine. As Caley says, you can adjust for the poorer games and all you really get is a fairly average team. It’s hard to find evidence in the data that this is a top-four contender. But, quality aside, Arsenal are weird.
Arsenal do many things on the pitch, but one thing they don’t do is press. Only Newcastle allow the opposition to make more passes before attempting to win the ball back. Newcastle have been wildly dysfunctional, so it’s hard to read much intent into what the Magpies have done. But Arteta spent more than two years in a coaching staff responsible for a very aggressive high pressing system at Manchester City. If he wanted Arsenal to play with a coordinated high press, they’d be doing it by now. In his playing and coaching career, Arteta worked with high pressing schemes and low blocks at various points. He’s gone for the low block here deliberately.
This is a near entirely different set of defensive players to those he inherited, so if he wanted to change the style, he would’ve done it. Aaron Ramsdale has really looked the part both as a shot-stopper and with his feet. In front of him, a back four of Takehiro Tomiyasu, Ben White, Gabriel Magalhães and Kieran Tierney (when fit) is emerging as the clear first choice defensive unit. Swapping in Tomiyasu for Héctor Bellerín has made a big stylistic difference. Tomiyasu is making less than half as many touches per 90 in the opposition penalty area than Bellerín last season, and he’s much less of an outlet for the team, receiving about half as many progressive passes. Just about every attacking stat is down compared to what Bellerín was doing. I don’t think this is a bad thing. Arteta has swapped out a modern style attacking full back for a defence-first option.
Is it working? Obviously it didn’t stop them getting steamrolled by a few excellent teams. But beyond that? It’s kind of hard to make the case. Their xG conceded per game has increased, which we all knew. But their median xG conceded has also risen, from 1.1 to 1.3. Ok, ok, let’s for a second tip the scales way in Arsenal’s favour just to see what we find. Let’s look at the xG conceded this season vs last, for every game except those against Man City, Liverpool and Chelsea. I don’t think this is the most statistically rigorous (not least because Liverpool and Chelsea weren’t dominating the league last time), but let’s just see. Can we now find some sort of defensive improvement against “the rest”? Please?
Nope.
Arsenal were conceding 0.96 xG per game against “the rest” last season, and this year it’s 1.01. That’s a pretty small difference, but it’s definitely not a sign of improvement. I’m sorry, but I’m calling it: Arsenal have not improved defensively this year.
Ok, then. What about the attack? What does the fancy nerd bullshit tell us about that?
Last season, Arsenal were generating 1.41 xG per game in attack. This year, it’s fallen back to 1.27.
Let’s repeat the same exercise as last time. The median xG per game?
20/21: 1.2
21/22: 1.3
That’s an improvement! A pretty small improvement, but we’ve finally found a metric where Arsenal look better than last year. Now, to continue the same exercise, what does the mean xG per game look like when we remove the “big three”?
20/21: 1.50
21/22: 1.54
Maths nerds are probably screaming at me about significant figures right now. FBRef only reports xG totals to two significant figures, so I don’t think what I’m saying here is statistically sound, but it at least looks like there’s been a small improvement against less-than-elite sides. I don’t think there’s enough here to say Arsenal have improved going forward with any degree of certainty. It’s well in the range of shit happens.
When they have the ball, Arsenal are patient. About 17% of the passes they complete are considered “progressive”, which ranks as the third-fewest in the league. Once they get into dangerous areas, they haven’t really been chance creation machines. They’re fairly average in terms of touches in the penalty area and then turning those touches into xG. None of it is bad, it’s just not what we’d expect a team with top-four ambitions to look like.
There’s a strong temptation to say that, while this Arsenal team might not be it right now, a youthful side will improve, and they could really have something in a few years (right in time for Jürgen Klopp and Pep Guardiola to pack their bags). To that, I say… probably. Bukayo Saka in particular looks like a star in the making. He leads Arsenal in shot-creating actions, progressive carries and progressive passes received. He looks a very talented modern winger, adept both with the ball at his feet and at running in behind. Emile Smith-Rowe I am less sold on, and it feels like a lot of the hype around him comes from scoring more than double his xG this season. But Arsenal’s best attacker this season (of those playing every week) by expected goals and assists per 90 is still Pierre-Emerick Aubameyang. The median outcome is that Saka and Smith-Rowe will keep improving as Aubameyang declines. That should be a net gain, but the negatives there are not zero.
Two years in and it still feels like we’re waiting to see what Arteta’s Arsenal ultimately becomes. This isn’t an Ole Gunnar Solskjaer situation. In that case, it wasn’t really obvious what his tactical ideas were, and we could’ve been there for another decade without being sure. It’s very clear what Arteta wants this team to be. I do think we’ll see something better emerge over time, provided they sign more players like Aaron Ramsdale and fewer like Cédric Soares. But I don’t think that’s going to happen at the speed Arsenal fans want it to. This, I’m afraid, looks to me like another year of frustration at the Emirates.
not the most erudite comment, but I love the phrase “Fancy Nerd Bullshit” and think you should slap a TM on that immediately