Goaltending Stats

[roy_133]

It's often been debated around here and I'm sure in many hockey circles but what do you guys think is the best way to assess goalies statistically? Is it the old school GAA, SV%, Wins route? What about shot quality based stats? Something else?

All goaltending stats that I've come across are flawed IMO, some way more than others but there's not really a great metric because there are so many variables, obviously. I'm just very interested to hear everyone's thoughts.

[tmash]

It's often been debated around here and I'm sure in many hockey circles but what do you guys think is the best way to assess goalies statistically? Is it the old school GAA, SV%, Wins route? What about shot quality based stats? Something else?

All goaltending stats that I've come across are flawed IMO, some way more than others but there's not really a great metric because there are so many variables, obviously. I'm just very interested to hear everyone's thoughts.

All hockey stats are somewhat flawed because it's impossible to take into account all factors.

For goalies, I find that over the span of a season, save percentage tends to average out to be a pretty good indicator. It can vary wildly game to game as it doesn't take shot quality into account, but over time it usually is able to correct itself.

[roy_133]

Well SV% obviously can favor (or hurt) guys playing behind different teams with different systems and margin for error is very high. Like I understand it's purpose and in some cases it doesn't bother me as much, like if you said over his last 40 games this goalie has a .920 SV% I wouldn't say anything but if you use it to compare 2 goals I have a problem with it because it paints an inaccurate picture.

I just don't see how we can fairly measure a goalie without knowing what kind of saves he makes and what kind of goals he allows and SV% doesn't tell us that with any kind of accuracy.

[tmash]

Well SV% obviously can favor (or hurt) guys playing behind different teams with different systems and margin for error is very high. Like I understand it's purpose and in some cases it doesn't bother me as much, like if you said over his last 40 games this goalie has a .920 SV% I wouldn't say anything but if you use it to compare 2 goals I have a problem with it because it paints an inaccurate picture.

I just don't see how we can fairly measure a goalie without knowing what kind of saves he makes and what kind of goals he allows and SV% doesn't tell us that with any kind of accuracy.

Sure, but how can you truly judge a forward without knowing the matchups they get, their linemates, their coach, their system, etc. In a way, all stats are meaningless because you don't know the context they occurred in (and it's why no matter what stat you bring up, someone will find a way to either say "well it's only this low because ... " or "this is overinflated because .... ")

I mean, it's tough to watch all 30 teams goaltenders for a significant period of time, so you have to generalize with stats. Save percentage obviously depends on the defense in front of the goalie and the teams you play against (do they just throw anything at the net or save it all for good chances?) However, normally if I look at the end of the season at save percentages, it roughly seems to rank goalies about where I'd expect them to be.

[roy_133]

Graeme, of course there are no perfect stats in hockey (although it's easier to find out what kind of players a forward is playing with and against) but what I'm asking is should we settle for a stat because we're used to it or look for a better way? I think there are better ways to measure a goalie's play out there, no perfect stat but I guess one that comes closer to perfect.

My problem is really with people comparing goalie's SV% over a small sample to make an argument and the people who use wins frustrate me even more. Like I said, I don't even hate SV% it's far better than wins or GAA and I'll use SV% in isolated situations in lieu of a better stat but never to say "Well Osgood has a .925 SV% in his last 5 games and Halak's is only .915 so Osgood has been playing better."

[mortwinkler]

I go by the old school stats, GAA and SV%, also, I do look at wins too.

I also check out saves and shots faced.

[roy_133]

I go by the old school stats, GAA and SV%, also, I do look at wins too.

I also check out saves and shots faced.

Well saves and shots faced is basically SV% although I suppose you could look at it to see what kind of sample you're dealing with.

Do you see how GAA and wins are basically team oriented and fails to isolate the goalie at all? Like I just don't see what that tells us about the goalie himself.

[Phizetheole]

I never liked judging a goalie by the amount of wins they have because you could have the best goaltender in the world but if hes on a horrible team, then he's not going to get wins. The other stuff like GAA and SV% is good though, however you can completely judge them by those either due to the fact that some many shots they don't have a chance of saving, but some are shot right into them. The best way to determine who you think is the best is bassically to just watch the way they play.

[mortwinkler]

Well saves and shots faced is basically SV% although I suppose you could look at it to see what kind of sample you're dealing with.

Do you see how GAA and wins are basically team oriented and fails to isolate the goalie at all? Like I just don't see what that tells us about the goalie himself.

Saves as in, did it trickle in from centre ice, or was it a sweet goal.

The level of difficulty I suppose.

[roy_133]

Saves as in, did it trickle in from centre ice, or was it a sweet goal.

The level of difficulty I suppose.

Well that's where a shot quality related stat would come in. IMO GAA and wins are irrelevant to a goalie. A goalie can give up 4 goals and play great and give up 2-3 goals and play poorly also wins are completely team oriented. SV% is by far the best of the 3 "old school" stats but still IMO can be improved on and has a lot of room for error when comparing 2 goaltenders head to head.

[kovalev47]

I think that Conn Smyths, Stanley Cups, and save percentage are the most important criterion respectively. If a goalie can win the cup, it proves that the goalie can handle the pressure regardless of the strength of the team. That's one reason I feel Osgood is very underrated. Not only does he win the cup, but he has a great save percentage in the playoffs too.

[SteenIsThaFuture]

The thing about SV% is it's very volatile, so its hard at tims to doubt its reliability in single seasons as an indicator of true ability.

Here are a few great goalies and their SV% over the years:

Patrick Roy 1990-1997

.912.

.906

.914

.894

.918

.906

.908

.923

(Interesting to note Roy never had a SV% above .915 until his 9th season and none above .920 until his 12th season).

Martin Brodeur 94-00

.915

.902

.911

.927

.917

.906

.910

Roberto Luongo entire career 00-10

.906

.920

.915

.918

.931

.914

.921

.917

.920

.906

Marty Turco entire career 01-10

.925

.921

.932

.913

.898

.910

.909

.898

.920

The only one who seemed to have a real model of consistency at any point, was Hasek in his prime in Buffalo, I'd hypothesize its because he faced such a vast amount of shots that the sample size evened out more.

Dominik Hasek 94-01

.930

.930

.920

.930

.932

.937

.919

.921

Even that has some fluctuation but obviously not as much as others, but before 94 he put up:

.893

.896

And after 01:

.915

.907

.925

.913

.902

So that was still volatile too, which describes goalies as a whole. Goalies are wildly inconsistent, game to game, week to week, season to season. So we don't even know if the SV% leader was actually THAT good or just a victim of sample size and luck.

[js2]

Over-analyzed.

[roy_133]

Over-analyzed.

That's so simplistic. I actually find it to be one of the most under analyzed positions in sports given the nature of it's importance to a team.

[js2]

That's so simplistic. I actually find it to be one of the most under analyzed positions in sports given the nature of it's importance to a team.

Around here.

[roy_133]

Around here.

By a couple members? People generally still used stats like wins and GAA when assessing goalies and until that stops the position is being under analyzed.

It's really a culture thing, it was the same in baseball for a lot of years. These stats are there so they must be right, otherwise why would the hockey news print them or nhl.com have them listed. A lot of people thought Bill James just had too much time on his hands. If you look deep enough and for the right stats a lot of evens in sports are far more predictable over the long run than people realize.

[roy_133]

And PS - I'm not trying to say I have the answers because I don't, that's why I brought it up I wanted to see what others think and if there are any new/interesting ideas out there.

[Guest]

I know this would probably go into the shot quality ( im not 100% certain how they currently rate shot quality)category but rating shots according to where they were taken on the ice in a more "official" manner could be interesting. For example,,,,if the shot came from the crease that could be called a zone 1 shot,,,,from the slot, a zone 2,,,face off circle, zone 3 and ect...Add to that if the goal came off a deflection or rebound you could then rate it as a class 1 deflection,class 2 rebound or class 3 direct goal. So then you get a description such as a zone 1 class 2 goal, ect...

Is this currently a similar way that shot quality is determined ???? I would be interested on that breakdown ( assuming an official one exists) if its widely different.

[SteenIsThaFuture]

I know this would probably go into the shot quality ( im not 100% certain how they currently rate shot quality)category but rating shots according to where they were taken on the ice in a more "official" manner could be interesting. For example,,,,if the shot came from the crease that could be called a zone 1 shot,,,,from the slot, a zone 2,,,face off circle, zone 3 and ect...Add to that if the goal came off a deflection or rebound you could then rate it as a class 1 deflection,class 2 rebound or class 3 direct goal. So then you get a description such as a zone 1 class 2 goal, ect...

Is this currently a similar way that shot quality is determined ???? I would be interested on that breakdown ( assuming an official one exists) if its widely different.

Well if you head to this page here:

http://www.puckprospectus.com/article.php?articleid=247

You'll see a shooting % chart (excluding rebounds) and the % of it going in from various locations.

That's basically how SQ is determined, not by class 1, class 2 etc. but by taking the location of the total shots and given a projected SV% based on that (and a few other factors).

[Guest]

Well if you head to this page here:

http://www.puckprospectus.com/article.php?articleid=247

You'll see a shooting % chart (excluding rebounds) and the % of it going in from various locations.

That's basically how SQ is determined, not by class 1, class 2 etc. but by taking the location of the total shots and given a projected SV% based on that (and a few other factors).

Interesting read MP. Thanks for that.

Has the NHL adopted any of this at all in its determination of the awarding of the Vezina??

Maybe its time to give a breakdown or rating of every goal scored on a goaltender. Seems like a lot of work, however it might go a long way in seperating a skilled goalie from a goalie with a skilled defence. I'm sure GM's would appreciate that type of knowledge before handing out these long term multy million dollar contracts.

[roy_133]

Interesting read MP. Thanks for that.

Has the NHL adopted any of this at all in its determination of the awarding of the Vezina??

Maybe its time to give a breakdown or rating of every goal scored on a goaltender. Seems like a lot of work, however it might go a long way in seperating a skilled goalie from a goalie with a skilled defence. I'm sure GM's would appreciate that type of knowledge before handing out these long term multy million dollar contracts.

I believe the voting is done by the NHL GMs for the Vezina so they used whatever they want to determine who they vote for, I`m sure some are aware of indepth stats like this and use them and other`s are more old school and traditional.

As for GMs liking to know this information, team have all this information and then some at their disposal and they can obviously get the info if they desire and I`m sure a fair but of teams atleast have it.

[SteenIsThaFuture]

Interesting read MP. Thanks for that.

Has the NHL adopted any of this at all in its determination of the awarding of the Vezina??

Maybe its time to give a breakdown or rating of every goal scored on a goaltender. Seems like a lot of work, however it might go a long way in seperating a skilled goalie from a goalie with a skilled defence. I'm sure GM's would appreciate that type of knowledge before handing out these long term multy million dollar contracts.

As roy said, some teams use advanced stats and some don't. The knowledge is out there though, like maybe on these forums ideas like Shot Quality are new and weird but most teams are "aware" of the stat, whether they choose to use it is a whole other issue but it would be bad business if they never looked at a competetive edge option.

Teams I KNOW use it are Minnesota (under Risenbrough though, but I'm like 95% sure though even under Fletcher they haven't changed much internally) and Vancouver and the ones I'm 50/50 on are the Kings, Sharks, Sens and Canes.

Some criticize it immediately because of its flaws, while some give it a glance and some put emphasis on it, not enough people I think do to have a serious impact on the Vezina or league-wide goalie decisions, they probably rely more on video and pro scouting.

In terms of rating every goal in terms of difficulty, I'll just say this:

[Johnny_rudeboy]

Over-analyzed.

I Agree.

Around here.

[SteenIsThaFuture]

I Agree.

I suppose this means by myself and I guess Roy at times since I know of nearly no other member that you'd be referring to. Please don't throw the rest of my fellow forum members under the bus with your ignorance. If you want to target people for stating blatant truths even though it goes against your old-fashioned, simple-minded beliefs, please state the names and don't generalize.

Thanks.

[blingbling_mtl]

I suppose this means by myself and I guess Roy at times since I know of nearly no other member that you'd be referring to. Please don't throw the rest of my fellow forum members under the bus with your ignorance. If you want to target people for stating blatant truths even though it goes against your old-fashioned, simple-minded beliefs, please state the names and don't generalize.

Thanks.

just 3 points:

1) an opinion is not blatant thruth;

2) new-fashioned belief is nothing more than belief;

3) insults like simple-minded or ignorance didn't change point 1 and 2.

(And I'm not sure I have such a great interest to remain on a forum with this kind of messages)