Captain Bowie
Registered User
- Jan 18, 2012
- 27,139
- 4,414
Exactly, the numbers are very useful when given the proper context. Just because context is required, doesn't mean the numbers aren't extremely useful.
Top-100 Hockey Players of All-Time - Preliminary Discussion Thread (Revenge of Michael Myers)
- Thread starter quoipourquoi
- Start date
- Status
seventieslord
Student Of The Game
this seems to compartmentalize the issue way too much, into sample sets involving only certain goalies, and into arbitrarily created sets of shots against numbers. Isn't it more statistically appropriate to look at a giant sample of 2000+ games like I did, and compare the exact number of shots against to the exact save percentage game-by-game to determine correlation?
Do the numbers change if you include 55+ Minute games only? I would imagine low SA games are skewed downwards by awful performances where a goaltender was pulled before shot 20 or shot 30.
seventieslord
Student Of The Game
I would assume this only includes complete or mostly complete games... someone would have had to have thought of that, right?
VanIslander
20 years of All-Time Drafts on HfBoards
No stat crunching is gonna make Roy into a better regular season performer than his contemporary Hasek. If a stat does, that result signifies a weakness of the manipulated numbers.
Michael Farkas
Celebrate 68
It is 55+ minute games only.
Question isn't the individual number necessarily, that's less important than tying it to general compartments of volume. Doc does this in 5-shot intervals and it breaks the "right" way...which is, of course, the way that I'm suggesting...otherwise I would have just ignored the fact that those tables are gone...
Question isn't the individual number necessarily, that's less important than tying it to general compartments of volume. Doc does this in 5-shot intervals and it breaks the "right" way...which is, of course, the way that I'm suggesting...otherwise I would have just ignored the fact that those tables are gone...
Michael Farkas
Celebrate 68
Canadiens1958
Registered User
Biggest obstacle to judging performance is fatigue. NHL has recognized this in their scheduling the last generation. Hence the renewed focus on "days of rest" data and other workload stats.
Benchmarks are what they are. In the context of a game 30 is reasonable - shot against every two minutes.
The resulting save% should be viewed in terms of game flow and scheduling.
Game flow reflects how both goalies were tested. Close SOG totals withn 5, medium within 10 or one-sided,over 1o.
Scheduling. A rested team giving up x shots is different than one giving up x shots playing their 3rd game in 4 days.
Last edited:
seventieslord
Student Of The Game
But if it's true that "the more shots you face the better your save percentage gets" then that would show up in a more expansive and inclusive study and, unless I'm doing it wrong, it doesn't seem to.
quoipourquoi
Goaltender
Depends. Does stat crunching include docking goaltenders for long-term missed time the way we would for contemporary skaters like Forsberg or Lindros, or is that something we seemingly ignore because the most widely used goaltending statistic is an averaging one and there is no set expectation for GP in a given year? Started before Buffalo (1988 Olympics) and continued after he left (Detroit, Ottawa), so it wasn’t just some bad string of luck in 1993, 1996, 1997, 1999, and 2000; it was a trend.
Killion
Registered User
- Feb 19, 2010
- 36,763
- 3,225
I have absolutely no idea what you guys are talking about & likely not alone. Uh?
Also how many shots a goalie faces on powerplays. Even 1996 Patrick Roy was off when Vancouver had high PPO. When Florida had virtually none, he was a brick wall.
Michael Farkas
Celebrate 68
It does. I've seen it, that is to say, I saw it. Let's wait for Doc, Doc ran a big expansive, inclusive thing...
Michael Farkas
Celebrate 68
I have absolutely no idea what you guys are talking about & likely not alone. Uh?
No wonder...it's about stopping pucks, this seems a little above your skill set, killion...
Killion
Registered User
- Feb 19, 2010
- 36,763
- 3,225
No wonder...it's about stopping pucks, this seems a little above your skill set, killion...
Real funny guy Mr. Farkas..... and why I'm confused as apparently Im
one of the few here who actually has... What the Sam Hill is this Gibberish?
quoipourquoi
Goaltender
The Deathstroke thing is still the damndest thing I’ve seen in goaltending statistics. Don’t agree with the conclusions he draws from it (as there are goaltenders who recorded great numbers with the low-quantity workload like late-1980s Roy and early-2000s Turco), and even just logically it doesn’t make sense to me why 40 shots today is different than 20 today and 20 tomorrow, but... there’s enough smoke that I wouldn’t bet against the fire on rhythm goaltending.
Michael Farkas
Celebrate 68
I'm just pullin' your chain of course, k dog...The heart of the matter is tying save pct. to volume. If you face a lot of shots, you're going to stop a lot of shots. Brodeur gets bad marks because he didn't have to face as many shots per game as some other guys, for instance...that's not Marty's fault, he gave up 2 just like everyone else...it's about timeliness and quality of goals against, not about how many shots you face...saves don't win games, bad goals against lose games...
As we'll come to see when these goalies come up...Brodeur was a master of not surrendering 3rd period leads in the playoffs, can't say the same for some of his peers...more on that later though...the heart of the matter is save pct. isn't tied to wins or talent (otherwise, the best save pct. runs in playoff history would have a far greater success rate than "meh" and 10 of the top 15 goalies of all time certainly aren't playing in this era...or, conversely, the pre-forward pass era, where certainly save percentages were higher)...it's tied to shot volume...everyone gives up 2, because if you give up 1, the debate is over and if you give up 3, no one's talking about you...
It's actually the opposite. The high SA sample is biased by the fact that any goalie that played terribly was already pulled before they got that far.
A goalie who goes 12/16 probably stays in the net for the entire 60 minutes. A goalie who finishes the first period 12/16 is probably warming the bench when the second period starts. So that means that a 30+ shot sample has very few results like, say, a goalie going 24/32, because they are nearly always pulled before they get that far. And that's why the 30+ shot bin has better results, you have to be playing well even to get there in the first place. What that means is that it is an artifact of sampling bias.
To show another example of how this can happen, every goalie in the league puts up much better numbers in wins than they do in losses.
Brodeur: .939 in wins, .863 in losses
Hasek: .947 in wins, .879 in losses
Roy: .937 in wins, .870 in losses
This is a much stronger effect than the low-shot/high-shot samples, so using the same logic we should claim that is easier for goalies on the best teams to put up high save percentages and harder for goalies on the worst teams to do so. But those numbers above don't actually mean anything. There is some relationship between team strength and save percentage, depending on which era we're talking about, but nothing nearly as big as those numbers show, and if you want to see what it is you look at goalie's overall winning percentage and their overall save percentage and figure it out that way. When you divide a sample up in such a way to exclude specific events, you can end up biasing your sub-sample and getting weird results. If that relationship doesn't duplicate over the entire data set, then it doesn't mean anything.
I've never seen any good evidence to not treat all shots against as independent. If there was a real relationship between shots against and save percentage, it would show up in the aggregate. It simply does not, though, except for periods of time like the Original Six and the Expansion Era when the relationship was actually negative (i.e. lower shots against mean higher save percentages). That would be a particularly crazy result to observe if there was a universal law that facing fewer shots was significantly harder. Why else do you think Klein and Reif gave a bonus to goalies based on shots faced in their Perseverance Rating back in the 1980s? The idea that Ken Dryden had it tougher than Gilles Meloche because he was facing less rubber is completely absurd.
The heart of the matter is tying save pct. to volume. If you face a lot of shots, you're going to stop a lot of shots. Brodeur gets bad marks because he didn't have to face as many shots per game as some other guys, for instance...that's not Marty's fault, he gave up 2 just like everyone else...it's about timeliness and quality of goals against, not about how many shots you face...saves don't win games, bad goals against lose games...
As we'll come to see when these goalies come up...Brodeur was a master of not surrendering 3rd period leads in the playoffs, can't say the same for some of his peers...more on that later though...the heart of the matter is save pct. isn't tied to wins or talent (otherwise, the best save pct. runs in playoff history would have a far greater success rate than "meh" and 10 of the top 15 goalies of all time certainly aren't playing in this era...or, conversely, the pre-forward pass era, where certainly save percentages were higher)...it's tied to shot volume...everyone gives up 2, because if you give up 1, the debate is over and if you give up 3, no one's talking about you...
During his Avs run Roy was quite poor at defending 3rd period leads on a percentage basis.
I would. Deathstroke always defended Jonathan Quick as having a tough job because he routinely faced fewer shots against than average, but he never provided any evidence to substantiate that other than the aforementioned shots against bins.
The problem is that we now have an awful lot of data on Los Angeles Kings backup goalies during their run of consistently strong shot prevention, and since 2009-10 they have combined for a very impressive .918 save percentage, despite just 26.7 SA/60. That's not only well above league average, it's even better than Quick's .917. I haven't adjusted for every goalie per season, so it might be +/- a thousandth or two if you do that, although that's a similar save percentage context all throughout and it won't change the fact that the Kings clearly aren't suffering with anybody in net despite the relatively few shots they allow. Also, if you look at their numbers in other places, those goalies as a group are obviously doing better in L.A.
There's also the recent Andrew Berkshire piece at Sportsnet, you can quibble with the weightings he uses to generate his rankings (I certainly do), but I'm focusing here just on the underlying shot quality data which is from some fairly reliable sources. Here's the relevant quote:
"No starting goaltender in the NHL has more consistently faced easier shots than Quick, as the Kings play a choking style that clogs the passing lanes better than any other team. That advantage is lessening, but the shots he’s facing are still more simple than anyone else is seeing."
If there was a disadvantage to facing fewer shots, it sure seems weird that in the most extreme low shot team environment of the post-lockout period, which is also the only period in NHL history where there is even a slight positive correlation between save percentage and shots faced overall, Kings goalies still faced the easiest shots in the league as measured by shot quality and their backups were collectively excellent statistically over a large sample size.
ImporterExporter
"You're a boring old man"
You're right, it doesn't.
BUT, number crunching (which does have its merits), does get him much closer than the raw data which is flawed. See blow. And then you have a massive gap in postseason credentials. There is no way to get Hasek closer to Roy unless you just play the "he played for garbage" card, which isn't accurate. Yes, the bulk of his career was not played on a dynasty but Hasek was not manning the crease for expansion level teams.
For instance:
Roy led the entire league in save % in 87-88 at exactly .900!!! 90% would place you around 30 something or so in today's game.
Roy's career Save% is .910
Marc Andre Fleury's is .913
Last year a .900 save% would have placed you 39th in the entire league. 39th, alongside Carey Price btw. With 30 or more games played as the benchmark.
If THAT doesn't illustrate why you need to adjust for era I don't know what else to say.
Sentinel
Registered User
Yes. Leading the league in something >>>> coming second. It's calling "winning." Will to win, if you will.
More like "goals" >> "assists." I am of this viewpoint, even though I understand many aren't.
That's correct. There is nothing whatsoever special about the 32-46-82 statline. Leading the league in goals for the EIGHTH TIME is pretty darn special. Unique, even. It demonstrates phenomenal, inhuman consistency, the like of which was never demonstrated by anybody else, including Gretzky (5 times league leader) and Lemieux (3 times). Especially the current 31 teams, 600+ players league.
Add to that the total goal count a few years down the line that approaches 900, and voila. "Crosby who"?
*You* can say anything you want. While overall domination is a consideration, we can and should judge goalscorers exclusively on the number of goals they scored in all situations. Not "what they could do, if only they would concentrate exclusively on goalscoring." Who knows how many goals would Gordie Howe scored if his coach told him to forego his brilliant two-way play and concentrate exclusively on putting the puck in the net.
Most likely, he'd phone his buddy in Kremlin and that question would never come up again. Just as plausible.
MXD
Partying Hard
- Oct 27, 2005
- 51,735
- 17,635
...I'm pretty sure we're ranking players as, you know, hockey players.
Canadiens1958
Registered User
An attempt to understand the experienced and visual with the statistical.
Compounded by the statistical featuring fluid definitions.
- Status
