Interesting. You are starting to persuade me....
I've seen some analysis that puts Pavelec's adjusted save% last season well into the top 10 (adjusted for shot quality). Is it reliable data, or does that matter? Not sure.
Over 19.5% of Pavelec's shots against were while the Jets were short-handed. For a number of the other top goalies that figure was more like 11-12%. Is it harder to post a QS when a higher proportion of the shots are while short-handed? A while back I compared several goalies last year based on their save% "adjusted" for their ES vs. Short-handed shots against. Because Pavs faced a lot more of his shots while killing penalties (when most goalies post a lower save%), this enhanced his stats vs. others. I assume that you adjust your analyses for that, though.
Don't get me wrong, I still think he's a really crappy goalie, but I think it helps to try to look objectively at all of the data when assessing performance for a given season.
What we're discussing is Pavelec's performance on whether or not the Jets made the playoffs despite him. Not on whether or not Pavelec is a good goalie. I'm not trying to convince you on that, although we seemingly agree on that discussion point if we were to make it.
Descriptive vs predictive.
My point is that while Pavelec stopped 92% of shots rather than his norm 90.6%, it was grouped in a manner that was least productive towards helping the Jets win%... especially given the Jets were an above average outshooting team with average shooting percentage.
Pavelec did stop a higher percentage of shots than the average goalie did.
This fact does not change if you adjust for shot location.
If he were someone who maintained this for long periods of time (which he failed to do), I would also conclude he is likely to continue stopping pucks.
He did however perform below average for most of his season, and only performed well near the end.
This is why I say despite Pavelec, even though Pavelec did perform exceptionally well for a period.
This is pretty much the only reason/way I'd use QS%.