An accurate use of stats this time. Much better.
Now where does that rank in the PL currently?
Without being updated to take his game today into account and adjusted per 90, he's currently 15th out of 21 in the league according to Fbref. Just ahead of a certain Mr. De Gea in 16th. Overall Fbref's model has them as having conceded 2.2 and 2.5 goals more than they "should" have respectively.
How much do you trust the post-shot stats? (I don't fwiw)
I think it's fine for for making general points like "De Gea was in better shot-stopping form in season X than season Y" or "X goalkeeper is generally a good shot-stopper". And obviously if it suits a random argument I'm making on the internet then I'll shamelessly cite it.
But in serious terms there are still a lot of flaws with the different models and we aren't as aware of those flaws as professional analysts would need to be to actually make effective use of that data. There's a reason clubs hire actual professionals and not any random idiot with access to Fbref.
That John Harrison guy gives a few examples of the models misrepresenting different chances in the video below, from about 5.05 min on, which make clear the kind of basic elements that still don't get captured in a lot of the stats we throw around.