Has Arsenal been unlucky with the low probability chances turning into goals allowed?

Most of the goals allowed recently have been low xG, does that mean that Arsenal have been unlucky?

When you look at the xG of the recent goals that Arsenal have allowed it is pretty wild. Just two of the seven goals allowed would be considered “good” scoring chances with seven of them being rated as 9% xG chances or worse.

Here is a handy list of the goals and the xG of them that has been going around social media:

It is a pretty crazy list to see.

There is a part of the tweet that is part of the analysis that I think deserves taking a bit more time with. Does the fact that the majority of the chances being conceded as goals mean that the team has been unlucky?

It is interesting, I don’t think that it logically follows however that the team has been unlucky here.

Sticking with the above end points from the example tweet (not wanting to pick on Monty here, it just happened to be the tweet that went viral and got things started) Arsenal have allowed 101 shots, with 33.7% on target, they have been worth 8.5 xG (9.6 from shot placement xG), and have turned into 9 goals.

Taking the shots and putting them into a simulator you get the following distribution of potential goals scored from the xG ratings for the shots:

The 9 goals is basically right on the expected value here. From these you would expect to be around 6 to 11 goals allowed, centered between 8-9 allowed. About half the time you would expect to have seen fewer goals scored against Arsenal, and about a third of the time you would expect even more goals to have been let in.

This is a bit above the raw xG number but with the error bars around the estimate, it is really hard to make a case that this is specifically a problem or an extreme outcome.

One of the big factors this ignores is that Arsenal just don’t concede many really good chances.

In the sample above Arsenal have allowed just 18 shots that were rated 20% or higher, of those only 8 were rated as 30% or higher xG. This trend holds true for the season as a whole for Arsenal. This is what the spread of xG broken down into buckets looks like compared to the Premier League average numbers:

Arsenal have a disproportionate number of their shots in the 10% or below range 82% over the last 12 matches sample and 80% in the Premier League, compared 68% for the Premier League average.

When you look at 7 of 9 goals in that range it is 77.8% and again basically matches the spread of the shots by bucket here. The total xG is a bit out of what with 5 goals allowed against 1.7 xG in the 5% range but it is made up with fewer goals in all the other categories. For the season as a whole it is more in line with the total xG of the buckets.

For the whole this season in the Premier League Arsenal have allowed 18 goals from 18.1 xG. Breaking it down into the buckets doesn’t perfectly line up with expected but that’s not uncommon given the sample sizes and how randomness happens.

TL;DR: The skew of goals is interesting and notable. Arsenal have an unusual skew compared to other teams for allowing fewer high-quality chances. Arsenal have not looked like they are especially extreme compared to the mix of chances allowed and goals conceded.