There are a number of conflicting narratives swirling around Arsenal at the moment. In the mainstream media, Arsenal are now being discussed among the title contenders. Here, for example, is Darren Lewis, talking about them in the Mirror:
Of course they are. With two strikers of the calibre of Aubameyang and Lacazette, allied with the creativity they have behind them, you’d be a fool to count them out.
They started yesterday’s stunning win over Fulham with Ramsey, Aubameyang and the highly-promising Guendouzi on the bench, for goodness’ sake. Torreira is already justifying the hype...
The team has a renewed impetus with the divisive politics of Wenger gone and the new methods of Unai Emery already taking shape.
It’s easy to say their wins have come against teams you’d expect them to beat but had they lost or drawn against those teams the critics would be saying: ‘Same old Arsenal’.
They aren’t the same. Their confidence is back, as underlined by the superb team goal scored by Aaron Ramsey.
Then there are the more statistically inclined writers, who are doing everything in their power to beat back that narrative. Here is Ted Knutson in Statsbomb:
Arsenal’s expected goal difference thus far in the 18-19 Premier League season is right around zero. Their actual goal difference through eight matches is +9. That’s a gap of over a goal per game, which is massive. Arsenal won’t continue to win this many matches unless their underlying performances somehow dramatically change for the better.
Now xG isn’t perfect... However, at this point across many thousands of team seasons in the data set, we have some strong ideas of what levels of overperformance are and are NOT possible.
Scoring on 25% of your shots like Arsenal are doing currently just does not happen across a full season of matches.
The table lies. Especially in the early season. Football has a lot of randomness and a lot of variance, and eight games isn’t remotely enough sample for all the “luck” to shake out.
Looking at the expected points table, which is based on the expected goals that each team produces in each match Arsenal would typically be in 10th place, based on the shots that they have created and allowed, and not 5th. A team with their shot profile would expect to have six points less than what they have currently. That is not good...
Looking at the same fixtures from last year (or the comparable one for Cardiff City) Arsenal have created 3.1 less expected goals and 12 less shots. On defense Arsenal have somehow even gotten worse against the same teams, conceding 1.4 more expected goals and 23 more shots.
Last season was the worst season that Arsenal have had in twenty years and with the change to Unai Emery the results have gotten better but the performances have gotten worse. I want to trust the process, but this makes me very worried.
All of that brings up a number of questions that need to be answered. One of them, and the one I am going to give a bit more information on today, is: does Unai Emery have a track record of his teams beating xG?
This is the operating narrative of managers like Sean Dyche or Lucien Favre - that they have special tactical things that they do that are not captured in the model. I have been asked if Emery also has this profile.
So let’s take a look at what Emery’s teams xG and goals scored have been over the last five years:
Unai Emery xG by Team
Looking at this data, the short answer is that there does not appear to be any long term beating of xG in Emery’s track record. His teams have produced 351 xG and 354 goals. Yes they have beat xG by 3 goals, but that is a difference of less than 1%. Arsenal are currently outperforming their xG by 63%.
The teams that generally outperform will have world class finishers on their team like Lionel Messi (even Cristiano Ronaldo doesn’t move the needle for a team like Messi does, and I don’t entirely buy the argument that Aubameyang* and Lacazette** are at the level to do that either), are far and away the best teams in their League where score effects are huge (Juventus has this happen for them), or do things like block waaaay more shots than expected as a tactic (like Burnley last season, but keep in mind they still got very fortunate on the offensive side of things).
Watching Emery, I don’t think that he fits into any of these categories as a manager, or that these categories fit the way that Arsenal have been built.
Arsenal are winning right now, and that is fun. Arsenal played good attacking football against Fulham, and that was fun. Arsenal are also going through a very hot finishing period, though, and history tells us that this doesn’t last forever.
That doesn’t, however, automatically mean that Arsenal are going to get stone cold to balance things out. Regression doesn’t work that way - it doesn’t mean an equal and opposite negative variance is for sure to happen. “Regression” just means that Arsenal will likely trend towards their more typical shot quality created as the season progresses. I just hope that Arsenal can improve that shot quality, so this statistical settling doesn’t hurt the team’s results.
*Aubameyang is an amazing player, but his skills as a striker come more on getting into great spots to score goals and not from being a “clinical” finisher. He isn’t bad in that regard; he just tends to be pretty close to his expected goals, and not a consistent overperformer.
**Lacazette is a player that has shown an ability to finish chances better than expected. He rates as a very accurate shooter in his career. I also don’t think that his shot volume is enough to drastically raise Arsenal’s overall team numbers. Messi is able to do this because he takes 6+ shots per match, while Lacazette is taking 2.5 per match.