It feels as if lately I’ve had to more frequently attempt to explain to people how xG works and why their conception of it, what it can, and what it cannot tell us is completely off. That might be because I’m starting to draw more on advanced statistics when I discuss football, but I think it’s also very much a general problem in the sports world and in society at large. We are not good with statistics.
I’ve been talking more about xG when it comes to Arsenal for two main reasons: the team’s performance since Boxing Day and Pierre-Emerick Aubameyang. Arsenal have, for me, been unquestionably better since that Chelsea match. xG bears that out and does so in a way that results maybe do not.
@oh_that_crab I looked at some Arsenal stats on https://t.co/PjzvzxMCxj pre/post Chelsea and comparing to other team’s full season xGD90.— Matt (@themattmyth) April 13, 2021
- Pre (14 games): 12th best xGD90 in the league at -0.12per90
- Post (17 games): 4th best xGD90 at +0.45per90
The advanced stats on Pierre-Emerick Aubameyang are a bit more complicated and nuanced, but basically, they indicate that he really hasn’t been significantly (or disturbingly) worse this season. His production is slightly down, but his production has been going down slightly each season for the past four or five, which is what you’d expect from a striker his age. He hasn’t fallen off a cliff production-wise. His numbers, like those of the entire team from an attacking perspective, took a big hit when the club basically forgot how to score goals for more than a month in the fall, but when he’s played CF, especially in the second half of the season, his xG production has been right around where you’d expect it to be.
So yeah. We’d all do well to learn about more about how statistics work. And in doing so, I think our collective opinion of where Arsenal stand would improve.