Patrick Roy vs Dominik Hasek 1994 Vezina

[archstanton]

I stumbled upon a link to a 1994 vezina trophy discussion that discusses Patrick Roy over Dominik Hasek.
Roy vs Hasek

Hi!
An interesting question that has yet to be resolved in this group is
wether Hasek or Roy had the better year last year.

At a glance it would seem that Hasek clearly had the better year.
He had a lower GAA, had .930 save percentage and won the Vezina.

Roy had a GAA which was .6 more, a .918 save percentage and was not
even an all star.

However, a closer examination reveals that
1) Roy faced more shots per game than Hasek 32 vs 25
2) Roy played in more games than Hasek

If you adjust for that fact then Roy had statistically an equivalent year
to Hasek. Considering Roy's importance to the Habs vs Hasek's I think it is
fair to say that once again the voters for the Vezina misinterpreted the
statistical evidence.

cheers,
kostadis

Sep 12, 1994, 12:15:15 PM

Strange discussion because Hasek has always been criticized for facing too many shots inflating his save %.

Click to read more...

 

[MadLuke]

Considering I never seen a year that team shots against and team shot against percentage was correlated, that a talking point you can spin as you want.

 

[Michael Farkas]

It's rate and averaging stats, you can make it say whatever you want...

Whether a team chooses to defend lines and front shots or chooses to yield ice and protect house/goalie will have a pretty significant impact on save pct. The amount of quality chances will be little different, if any.

Everyone gives up 2. Doesn't really matter if it's on 18 or 38.

 

[MadLuke]

Yet last season the correlation between shots against and save percentage was small but negative (opposite of what the surrounding discourse would make you expect),

save percentage of the 5 teams that allowed the less shots:

0.9

0.884

0.913

0.909

0.898

5 teams that allowed the most shots:

0.89

0.892

0.904

0.897

0.894

would anyone that is a strong believer that most shots = better save percentage, have pick which group was the 5 team facing the less shots (.908 save percentage) and which was the teams with the most shots (.895% average) among those 2, correctly ? Should it not give some pause to that theory, and last year was not the exception.

And goal allowed to seem to always been quite correlated to how many shots you allow, by a giant amount (and always more shots, more goals), last year team goal against average went from 2.41 up to 3.98, remove the atrocious sharks up to 3.63 or 50% more goals going in for the bad team versus the best one, not really a situation that every team allow 2 goals a game, what change is just how many shots they decide to allow.

Same goes for high danger chance, high danger chance a game went from 622 up to 901, they move a lot and very highly correlated with the shots against.

People can be right, but years after years, I could have looked more than 8 different seasons, a strong phenomenon between shots against boosting team shot percentages against does not seem to show up and seem often to be the other way around (small but the other way around), could be bad team allow more shots, bad team have bad goaltenders in general and hide the phenomenon and it could be true for that exact team, coach that exact season.

But it seem to be way too generalized as a rules by some or it would show up in league wide correlation between shots and S%, how could it not ?

 

[Michael Farkas]

Everyone allows 2 is a reference more relevant to the thread than 2024 hockey haha

Scoring chance isn't defined, so sources have different numbers. Generally you give up 9 to 12. Most teams 10 or 11.

Big key is types of chances against goalie strengths.

 

[MadLuke]

Generally you give up 9 to 12.

One of those is 33% more than the other, is that much different than how much shot against vary ?

How do you explain how little team save percentage and shots against correlate, league wide, usually (if not every year) ? While shots against and goals against correlate a lot.

 

[Michael Farkas]

Games aren't won on percentages these days and most teams aren't on the edge. 10 or 11, even if it's 10 or 12...

Teams don't make saves. Teams are (dis)credited with giving up empty net goals. Goalies make saves.

 

[MadLuke]

I am not sure the distinction being made (with team vs goaltender making save specially coming in the context of people saying that it is very much a team stats and how they play), the question is relatively simple, why we cannot simply see high S% among team that receive more shots if there is a supposed strong link between the 2 (and why we often see the other way around, the more shots team received a season, lower was their save percentage), I never heard a suggestion/answer to that question.

Last season, remove the Sharks, Montreal,etc... the 5th worst team had 1104 high danger chance against (subjective and not perfect but let roll with it), the 5th best had 878, that a bad team giving up 25% more than the good one.

That a huge gap, probably bigger than the different in raw amount of shots against (5th worst gave up 16% more than the 5th best), why say that it is a particularly low amount similar to each team, even if they give different amount of shots, in your own word you speak of a 20% difference, that really big, a team giving 90-100 more high danger chance than other during a season, in a sport were giving 0.2 goal a game is a big amount (players that add this level of offense over third line replacement that are paid lot of millions).

 

[Michael Farkas]

Because there are more variables than that in the game. Shot and save have no value on their own. We treat them as equal and important, but they're highly variable and unimportant.

High danger to whom? High danger to some goalies are not high danger to others. Look at what "prime" Tampa allows when they're just holding on, up by 1. I assume many models would consider it "high danger" because of the shot location. But it's not high danger to their all star goalie. It almost certainly is high danger to their unimpressive backups.

I just don't get the desire for the one-size, fits-all approach - like, what result do you want to conclude here? "Giving up chances = usually bad" ok, sold. But it'd be more interesting to find patterns in terms of shot distance, time between shot attempts, etc. and their correlation/effects.

There's only a handful of goalies in the league right now that are really consistently plus, there's only a handful of coaches that are super capable AND have a roster to have some control over the game...then there's also a lot of the opposite. So, the larger dataset isn't exactly more telling all the time...it's just introducing variables that we're not equipped to readily handle,

 

[MadLuke]

Because there are more variables than that in the game. Shot and save have no value on their own. We treat them as equal and important, but they're highly variable and unimportant.

This seem to go around the simple question, why is there no correlation between shots against team and team save percentage, but maybe you are not a proponent of the theory more shots make it easier to have higher shot percentage to start with, that was a general point remarked by the op (
Strange discussion because Hasek has always been criticized for facing too many shots inflating his save %.)

o, the larger dataset isn't exactly more telling all the time..

Obviously my argument here is not a Claude Julien-Chara or Hitchock-Pronger team cannot have some effect here, just people saying more shots = higher save percentage as a rule, that I wonder what is their explanation for what seem an impossibility when you look at large volume and do not try to start to look at tricks to find it, like individual game with more than X shots in a bucket vs less in another during a season.

I just don't get the desire for the one-size, fits-all approach - like, what result do you want to conclude here? "Giving up chances = usually bad" ok, sold. But it'd be more interesting to find patterns in terms of shot distance, time between shot attempts, etc. and their correlation/effects.

It obviously would be, but I would keep an open mind that the general rules would be that bad team allow more shots and the percentage of said shots that are good scoring chance with be 0.8+ correlated with how many shots they allow (and that why goal against is so highly correlated with how many shots they allow)

 

[Doctor No]

maybe you are not a proponent of the theory more shots make it easier to have higher shot percentage to start with,

I've still got too many confounding things in my analysis to share anything definitive, but I do believe that this phenomenon is functionally explained by two things:

(1) Score effects - teams trailing in the third period (or sooner) start trading quality for quantity under the auspices of "let's get something on net". So the leading goaltender sees more shots of lower quality.

(2) Scorekeeper differences - obviously, every goal is going to be counted consistently as a shot by all scorekeepers. So the difference is going to come from non-goals, which means that liberal scorekeepers will simultaneously (a) increase shot totals and (b) increase save percentages.

 

[MadLuke]

I think those are 2 good theory and would easily explain why they would balance out over the whole season, yet possible to find out looking at individual game.

 

[JackFr]

Because there are more variables than that in the game. Shot and save have no value on their own. We treat them as equal and important, but they're highly variable and unimportant.

High danger to whom? High danger to some goalies are not high danger to others. Look at what "prime" Tampa allows when they're just holding on, up by 1. I assume many models would consider it "high danger" because of the shot location. But it's not high danger to their all star goalie. It almost certainly is high danger to their unimpressive backups.

I just don't get the desire for the one-size, fits-all approach - like, what result do you want to conclude here? "Giving up chances = usually bad" ok, sold. But it'd be more interesting to find patterns in terms of shot distance, time between shot attempts, etc. and their correlation/effects.

There's only a handful of goalies in the league right now that are really consistently plus, there's only a handful of coaches that are super capable AND have a roster to have some control over the game...then there's also a lot of the opposite. So, the larger dataset isn't exactly more telling all the time...it's just introducing variables that we're not equipped to readily handle,

Well, considering the lack of detail in historical goalie stats I don't really know where any of this gets us.

The data suggests that the idea of more shots against = higher save percentage is a myth at the team level and at the goalie level, even when adjusting for score and venue. The vibes-based case that goalies benefit from facing more pucks can be just as convincingly argued against.

Do some teams concede lots of less dangerous shots? We've seen it. But with zero data to go off of that would tell us "Buffalo allowed fewer slot shots" or "Montreal conceded fewer shots but they were from closer to the net" or whatever, I don't think it's fair to brush aside save percentage. It's the best of a bad lot. In a vaccuum where we lack shot quality data, allowing 2 goals on 30 shots is better than allowing 2 goals on 20 shots.

 

[Michael Farkas]

Brush aside? Maybe, maybe not. But the data is available in some respects, it just needs to be mined properly. A couple things working against it: Time, the ubiquitous one. And then the other one is I don't think people want to challenge its merits because it's uncomfortable. It's like when Corsi came on and everyone thought, "THIS IS THE BEST!" and I'm sitting here going, "well, what's this actually telling us....oh I see, it's this. It's biased towards this player type and this player type. It's biased against this type. Ok...so this is........fine but it's not anything I'd want to evaluate a player on."

Save pct. certainly doesn't need to be on the pedestal that it is though, so I rail against its deity status whenever possible haha

 

[MadLuke]

The talk around save percentage being boosted by shots seem much more common on the Internet than pushing against it too, I am not sure that its valid, study about shoot quality by shots against being a significant impact or not to goaltender (outside extreme rare case) seem to be really common online, people try to do it a lot.

Hard to do seem more what is going on, would he have everyone stick position, helmet, skates, puck position at all time (speed-trajectory), data about goaltender shots quality and stop relative to shoot difficulty would pop-up everywhere right away, it is not some big taboo I do not think (if anything it is equally if not more a taboo the other way around, among vocal people).