How many creators does a top team need?

Arsenal's injury-plagued campaign raises interesting questions about depth

The concept of chance creation and whether Arsenal do enough of it has been swirling a lot in my mind this week.

For one, there was that graphic going around (I’m not gonna look it up, but you may know what I’m talking about) showing that Arsenal are tops in the league in terms of overperforming their xG so far this season. Some might interpret that as Arsenal having the best finishing performance in the league to date, but others took another angle: Six clubs in the league have more xG, and that means we don’t create enough chances.

A little later in the week I was reading through comments on our new favorite debate, Šeško vs Ekitike, (it’s inadvisable, I know, but you can’t make me stop) and came across this one:

And so the seeds of thought were sewn. If you follow me on BlueSky, you may know I’ve been saying since Ødegaard’s injury in the fall that this would be an odd year for Arsenal’s chance creation numbers. At first I thought it might take a few weeks with him back for things to equalize a bit, but then Saka tore his hamstring, and now we may never really get a look at what “full strength” Arsenal’s chance creation looks like.

The natural conclusion, unsatisfying though it may be, is that reading too much into the above is an exercise in futility. Why? Because you’re using performance under extreme conditions that you expect not to be replicated (Ødegaard being out, Saka being out, Havertz being out) to evaluate and predict the future. Next season, Arsenal will expect to line up with Ødegaard, Saka and Havertz, so data from games where they weren’t playing is unlikely to be especially predictive.

That’s what bugs me about the chance creation analysis this season: It’s inventing a hurdle for Arsenal that most sides simply couldn’t clear, and by the way, this Arsenal side are doing a commendable job getting close to clearing it, anyway.

As I write this, Arsenal have played 25 Premier League games. Out of a possible 50, their top two chance creators from last season, Bukayo Saka and Martin Ødegaard, have played 29.4 90s, and keep in mind that most of those aren’t even overlapping.

It’s no surprise, then, that Arsenal have dropped from 1.80 npxG per 90 last season (number 4 in the Premier League) to 1.54 npxG per 90 this season (7th). It’s a drop of about 14%, a couple of decent chances per match. It’s definitely not nothing, and it sounds pretty bad!

But before we go calling Arsenal a puttering attack this season, let’s add some context to the discussion. For all the drama and injuries and red cards, the club is still number 20 out of 94 clubs in the top 5 leagues, roughly 78th percentile. More creative than AC Milan, Dortmund or Lille, to name a few.

There is absolutely the natural, emotional response here to the number 20, which is “that’s not good enough.” And I get that. I’ve had people tell me directly that Arsenal, even with these injuries, should have held faster. There should have been more creators available to plug gaps so the drop wasn’t as harsh. So let’s explore that.

Where better to start than with league leaders, Liverpool. Their closest analogues in terms of chance creators to Saka and Ødegaard are Mo Salah and Trent Alexander-Arnold, each in the upper tier for their positions. If those players missed time, top options to step in as creators include:

• Cody Gakpo, 70th percentile chance creator among wingers

• Dominik Szoboszlai, 60th percentile chance creator among attacking midfielders

• Andy Robertson, 85th-90th percentile chance creator among fullbacks

• Alexis Mac Allister, 80th percentile chance creator among central midfielders

Those are pretty good options! But it’s absolutely key to note that AMC + Gakpo tends to equal about one Salah, that’s how significant his footprint is in their attack. Robertson, despite being a leading creator in his position, creates about two-thirds what Trent does.

Those two (Salah and Trent) generate about 42% of Liverpool’s total xAG output, down ever so slightly from 42.9% last season. Arsenal, for the record, got about 41.1% of its xAG output from Ødegaard and Saka last season.

One real-world example we’ve seen this season besides Arsenal is RB Leipzig, who’ve lost their top three creators in terms of xAG from a season ago: Dani Olmo to a sale, and Xavi Simons and David Raum to injuries for parts of this campaign. The result? Dropping from the fourth-best attack in the league to 10th, with about 40% less output! Another great example is last season’s Manchester City, who were led in chances created by Kevin De Bruyne despite the Belgian playing less than 14 full games’ worth of minutes.

The point here is meant to be simply that losing your top two creators for long stretches is a major blow for any club, even the largest in the world (Barca losing Raphinha and Lamal, Real Madrid losing Vini and Jude, PSG losing Hakimi and Dembélé, etc).

Are some better set up than others? I think they are, but I also think maybe some of the talk about how Arsenal are set up is dismissive of how Arteta has his team set up. Case in point:

• Leandro Trossard is in the 78th percentile for xAG among PL wingers this season, having created more xAG per 90 than Bruno Fernandes, Antoine Griezmann or Jamal Musiala.

• Declan Rice is 93rd percentile for xAG in the PL among central midfielders. He’s created more than all of the above as well, and also Anthony Gordon, Cody Gakpo and Alex Iwobi. And as long as they count goals scored from corners, yes this does count.

• Gabriel Martinelli, amid a down season as a creator in which he’s created xAG in just the 40th percentile among PL wingers, is coming off two seasons in which he created in the 80th and 92nd percentile.

Would having a top chance creator at striker solve for this? I don’t especially think so. Kai Havertz this season has been averaging 0.12 xAG per 90, which is pretty middle-of-the-road for a number nine. The top centre-forward with enough minutes to qualify is Nicolas Jackson, at 0.21 xAG, the difference of one “meh” shot created per game. But even a player like Jackson or Ekitike is not making an attack happen on their own - they are surrounded by creators who are enabling them to create more as well. Both Chelsea and Frankfurt are the second-ranked attacks in their respective leagues.

As Arsenal turn down the final leg of the season, it’s possible their xG created could sink even lower due to injuries and fatigue, as well as the slow return to form of Saka. But between a return to fitness for key players and the business that has been planned regardless of what the xG table said this campaign (and that may not include a new “creative midfielder”), there’s plenty of reason to believe the attack is in good health - even if the players currently aren’t.