1) I don't see any correlation between chance of going in and having positive differential. They could have a post-shot xG of 0.5 and the keeper could still over- or underperform. It seems that stat only ranges from 0.28 and 0.36 in the league so it's not like Chelsea are such a huge outlier. Kepa was also ranked 36/38 for post-shot performance last year and 35/50 the year before for Bilbao.
2) You can't just make a definitive statement like that and leave out the expectation of the chances going in. If Chelsea's opponents scored those 11 goals from an xG of say 4, then that would be very much on the goalkeeper, wouldn't it? I haven't seen all of those goals but I do remember corners where Kepa was definitely at fault, akin to the clip of the van Dijk chance earlier.
1. 46% correlation between the two metrics this season. Last year was close to 0, 39% in the season before. Just think about it - If I'm facing tougher quality chances, it's more likely that I'd concede more goals. And in xG terms 0.08 is a pretty big difference at a per shot level.
2. Set piece goals are in 90% cases down to defensive tactics. I remember bashing Mou and then Ole for this and the season for that is this can be solved. Either clearly define responsibilities of each player or more sessions on that.
You're interpreting this stat wrong - you're looking at post shot xG per shot on target. This is a way to measure the extent to which opposition finishing had an impact. This metric is more or less independent of the goalkeeper. It means that Chelsea's opponents finished at a high rate, but extrapolating that to imply that the defense was at fault is unwarranted - the whole point of post-shot xG is that the defending isn't included in the calculation.
Also I just fundamentally disagree re: corners. Kepa has made fewer punches than any goalkeeper in the league. If the defence is being forced into 1v1s unnecessarily because the goalkeeper is stuck on his line, where does the fault really lie?
Opposition finishing does not have an impact, from where the opposition took their shots, the scenario and the body part used are the determining factors here. It quantifies the quality of shot the GK faced and a higher number means that on an average the shots Kepa faced were from positions where its more likely to score from as compared to other teams.
Again, I do find his goalkeeping comical at times and do believe that replacing him will mean that you'll concede atleast 5-10 goals less, but you're in for a shock if you believe that you believe that Azpi, Rudiger, Alonso and Zouma can make you title contenders (and I like Azpi and Rudiger, but they aren't that good).
For your second point, Kepa has stopped 7.2% of the crosses faced. This is better percentage than Ederson, Dave, Lloris, Leno and Schmeichel. I'm sorry I couldn't find the punches data available anywhere. And also, I'm sorry if the GK is facing a 1v1 situation, it's hardly ever the GKs fault unless his distributtion was that poor
And