Overall, Patrick Roy was .949 in his 10 bounce-back games after a particularly rough night. So while Roy couldn't exactly expunge 5+ GA games from his hockey card statistics, his team always seized back momentum off the strength of his play in the followup game. I would argue that those ten 5+ GA games (once every 13 or so games) made less of an impact on their series than they do on Roy's per-game numbers.
So your argument is that Roy's strong play after getting shelled won back momentum for his team? Here are the scores of the games after giving up 5+ GA in a loss:
4-0, 4-2, 4-1, 7-0, 6-0, 6-2, 8-2, 4-1, 2-1, 4-3
I'm not sure the rest of the team needed Roy to give them momentum, their offensive performance suggests they could have won nearly all of them with much worse than a .949. In the games where Roy allowed 5+ GA, his team scored 3 or more in 8 out of 10. I'd say the goal distribution suggests that, if anything, Roy cost his team more by his bad games than he gained them by bouncing back strong.
In Patrick Roy's best four-series run (1993), he started out with a .936 when only .861 was expected against the opponent.
I appreciate the effort to adjust for opposing performance, but there are a few issues with team shooting percentage. It includes empty net goals, it is notoriously variable from season to season and it can be skewed by performances against non-playoff teams especially when parity is low. The 1992-93 Nordiques are one of the best examples of why team shooting percentage can sometimes be an imperfect measure.
Quebec Shooting Percentage:
1991-92: 11.7%
1992-93: 13.9%
1993-94: 11.5%
'93 Quebec goals per game against playoff teams: 3.63
'93 Quebec goals per game against non-playoff teams: 5.03
Not only did Quebec almost certainly shoot well above its true talent level in 1992-93, the Nords also majorly padded their stats against the weakest teams in the league. I think the best guess for Quebec's team shooting talent against decent opposition would be in the neighbourhood of 12% rather than 13.9%.
That said, I don't actually have what I think is a good method for adjusting team shooting percentage given these challenges. Playoffs have both better shooters and better goalies and a stronger defensive effort, and it seems like those factors wash out to some extent, although I'd suggest the more teams have to play hard to make the playoffs, the closer regular season and playoff numbers seem to sync up.
Clearly higher, but I don't know if I'd say different level entirely of the save percentage era goalies. If that's what you call a different level, then Roy's in the apartment below directly below him, banging on the ceiling with a broom, and the three floors below Patrick Roy's apartment are probably completely vacant.
Well, I'm arguing that Hasek was in the ballpark of .010 above Roy comparing peak-to-peak, so yes, I'd say that 's a different level. Look at an extended multi-season sample and compare guys .010 apart, and there's usually a pretty solid talent gap. If you have a different estimate of Montreal team effects, then you might claim that they are closer than that. But it looks like most people apparently make zero adjustment for playing in late '80s/early '90s Montreal, and that's a huge advantage for Patrick Roy in these discussions.
When a statistical edge disappears when the games matter more, then the value of that statistical edge comes into serious question.
This is a fair point. But again, I think it comes down to how that statistical edge is calculated and what assumptions you make about team effects. Team effects are usually magnified in the playoffs. And the majority of Roy's playoff outperformance comes from his time in Montreal, when he was most likely to have benefited from his teammates.
You think the difference in regular season peak is enough to overcome Roy's better playoffs and greater reliability. I don't. And surprisingly, neither do the majority of the people who gave Hasek most of his awards.
I'd phrase it more like I think Hasek has a significantly better regular season peak, a better overall career and better per-game team-adjusted playoff performances than Roy. And as for sportswriter opinion, the majority of people who gave Glenn Hall his 7 First Team All-Stars would have rated him below Terry Sawchuk, but that didn't stop both of us from ranking Hall ahead in the top goalies project. People like wins and Cups, doesn't mean they're right.