HFBoards

Go Back   HFBoards > NHL Western Conference > Pacific Division > Edmonton Oilers
Notices

Defense influence on SV%

Closed Thread
 
Thread Tools
Old
03-23-2004, 10:10 PM
  #1
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Defense influence on SV%

A while ago igor posted The Profound Importance of Goaltending and used the San Jose Sharks as an example to illustrate the impact that goalie ability has on outcomes in games.

While I agreed that goalies were profoundly important, I questioned whether SV% alone could allow accurate determination of a goalie's impact. Consequently, one of the ensuing discussions was whether or not the goalie's save percentage reflected primarily how well the goalie was playing. igor's contention was that SV%, for the most part, measures goalie performance (which might involve luck) and is not influenced very heavily by the abilities of the team in front of the goalie, or the style of play.

I am not convinced by this, although I'm not really convinced against it either. Although I have some opinions based on my own viewing of a number of games, I'm willing to challenge them. I believe some hard data would help out.

One way I suggested to get data to help out is to look at goalies who have changed teams. I still think that would be a worthwhile pursuit.

Another possibility is to look at teams that change substantially in front of the goaltender. If a good defense is weakened by injury or trade, then either the goalie's SV% will change or remain about the same. A large change would suggest that goalie SV% are substantially dependent on the abilities of the defense in front of them. The implication is a causal relationship, although obviously no hard conclusions would be possible unless (a) the statistical power was high enough to draw firm conclusions, and (b) one could observe the cause-effect relationship in some way. Still, a large swing in SV% due to changes in D would be enough to raise doubts about the belief that goalie SV% is almost solely a result of a goalie's ability/achievements.

This year presents two cases that I could think of. In the following table are the pertinent goalie stats for two goalies, Martin Brodeur and Dan Cloutier. Brodeur played almost half the season without Scott Stevens in front of him (injury). Cloutier played a smaller number, but still a substantial number without Jovanovski. Its relatively uncontentious to say that Stevens and Jovo are anchors of their respective defenses.
Code:
GOALIE           DEFENSE  GP  GA   SA  SV%   SA/GP
--------------------------------------------------
BRODEUR  with    STEVENS  37  68  885  0.923  23.9
         without STEVENS  32  74  817  0.909  25.5
CLOUTIER with    JOVO     35  79  967  0.918  27.6
         without JOVO     19  48  448  0.893  23.5
Both Cloutier and Brodeur's SV% stat dropped significantly when their top Ds were not in front of them. At least Cloutier's shots against actually was higher when his top defender was in front of them. So its not clear that SV% is related to shots, although if it is, this suggests it may be at least sometimes negatively related. One possible explanation that fits the data is that good defenders force the opposition into lower-percentage shots. This would predict both SV% and SA/GP changes.

Comments?

oilswell is offline  
Old
03-24-2004, 01:58 AM
  #2
jadeddog
Registered User
 
jadeddog's Avatar
 
Join Date: Sep 2003
Location: Regina, Saskatchewan
Posts: 11,811
vCash: 500
good post.... i would be interested in the winning percentage and GF average of the teams jersey and vancouver played during these times as well though...... because a weaker schedule will most certainly increase a goalies SV% as will playing teams that are more prone to score

the thing is, that all these things obviously affect a goalies SV%..... if the oilers lost brewer and smith to injury and played ottawa, detroit, vancouver and tampa bay, i would be willing to bet that whatever goalie was playing would not have great SV% numbers..... but generally speaking these things tend to average themselves out over the season, in that everybodys strength of schedule is relatively the same and most teams have to deal with some form of injury (albeit some are affected much moreso than others)...... when looking at SV%, you have to make sure that you are taking a good sample size, or your error is going to be fairly high, but if your looking at 30-50 games over a season, usually that will be a large enough sample to take into account things that affect this statistic

jadeddog is offline  
Old
03-24-2004, 02:17 AM
  #3
dawgbone
Registered User
 
Join Date: Jun 2002
Country: Canada
Posts: 21,104
vCash: 500
Send a message via ICQ to dawgbone Send a message via MSN to dawgbone
Quote:
Originally Posted by jadeddog
but generally speaking these things tend to average themselves out over the season, in that everybodys strength of schedule is relatively the same and most teams have to deal with some form of injury (albeit some are affected much moreso than others)......
I don't know about that part.

Look who the Oilers had to face:

24 games vs teams in the Northwest (18 against .500 or better). 8 games against teams in the Northeast (all 8 against .500 or better)...

The Oilers have had a pretty rough schedule. Don't forget, the upper echelon teams don't play against themselves, which tends to lighten their schedule.

dawgbone is offline  
Old
03-24-2004, 03:27 AM
  #4
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Quote:
Originally Posted by jadeddog
good post.... i would be interested in the winning percentage and GF average of the teams jersey and vancouver played during these times as well though...... because a weaker schedule will most certainly increase a goalies SV% as will playing teams that are more prone to score
This is of course a valid concern. Take NJ for example. There's a few more games with Stevens than without. 14 of the games were with common teams. Of the reamaining, here are the differences:
Code:
WITHOUT                WITH
------------------------------------------------------------
ATL CAR CAR CBJ MIN    PIT PIT PIT WSH WSH 
NYR NYR NYR CGY        FLA PHO BUF BUF BOS
MTL MTL MTL STL        BOS EDM NSH NYI NYI
TOR VAN DET TB         NYI NYI NYI DAL S.J
                       COL COL
By eye, the schedule difficulties look comparable between the with and without conditions. Anyone willing to calculate a degree-of-difficulty is encouraged to do so. With Stevens in the lineup, the Devils played several fairly crummy teams, however they clearly played a non-trivial number of hard teams. Yet with Stevens the SV% is higher. Not sure if schedule difficulty accounts well for it. Fluke might though. I haven't calculated what the expected deviation would be between two random samples from that data. It might be close to .140.

Quote:
the thing is, that all these things obviously affect a goalies SV%..... if the oilers lost brewer and smith to injury and played ottawa, detroit, vancouver and tampa bay, i would be willing to bet that whatever goalie was playing would not have great SV% numbers
I guess this goes back to the original argument. One point of view is that the GAA might rise but the SV% might not. The whole point was to compare the arguments against some numbers to see which ones are supported.

Good food for thought jadedog.

oilswell is offline  
Old
03-24-2004, 09:19 AM
  #5
kyle
Registered User
 
kyle's Avatar
 
Join Date: Aug 2003
Location: Toronto
Country: Canada
Posts: 636
vCash: 500
Quote:
Originally Posted by dawgbone
I don't know about that part.

Look who the Oilers had to face:

24 games vs teams in the Northwest (18 against .500 or better). 8 games against teams in the Northeast (all 8 against .500 or better)...

The Oilers have had a pretty rough schedule. Don't forget, the upper echelon teams don't play against themselves, which tends to lighten their schedule.
Buffalo has the hardest schedule in the league, Tampa the easiest.

kyle is offline  
Old
03-24-2004, 12:25 PM
  #6
jadeddog
Registered User
 
jadeddog's Avatar
 
Join Date: Sep 2003
Location: Regina, Saskatchewan
Posts: 11,811
vCash: 500
Quote:
Originally Posted by dawgbone
I don't know about that part.

Look who the Oilers had to face:

24 games vs teams in the Northwest (18 against .500 or better). 8 games against teams in the Northeast (all 8 against .500 or better)...

The Oilers have had a pretty rough schedule. Don't forget, the upper echelon teams don't play against themselves, which tends to lighten their schedule.
i thought this as well earlier in the year (that edmonton and other teams that play in a hard division like the NW would have a harder schedule).... but then i happened to look at ESPNs power index thingie, and i saw that the strength of schedule for all teams in the league is almost exactly the same .... the lowest is 524 (detroit and jersey) and the highest is 539 (buffalo) ... so basically, over the entire season, it all averages out pretty good

jadeddog is offline  
Old
03-24-2004, 02:42 PM
  #7
Matts
Registered User
 
Join Date: Nov 2002
Posts: 2,911
vCash: 500
Igor talks about this all the time

how the Oilers don't do themselves any favours by asking the league for those extra home games vs the Leaves and Habs. I know the season ticket holders want good matchups but maybe they'd be happier if the Oilers had a better sked.

It wasn't too long ago that the Oilers were doing this stupid thing of playing weekend home games on back to back nights. I could never understand it and I read last year that Lowe said they were asking for this but they no longer would

Matts is offline  
Old
03-24-2004, 03:37 PM
  #8
dawgbone
Registered User
 
Join Date: Jun 2002
Country: Canada
Posts: 21,104
vCash: 500
Send a message via ICQ to dawgbone Send a message via MSN to dawgbone
Quote:
Originally Posted by Matts
how the Oilers don't do themselves any favours by asking the league for those extra home games vs the Leaves and Habs. I know the season ticket holders want good matchups but maybe they'd be happier if the Oilers had a better sked.

It wasn't too long ago that the Oilers were doing this stupid thing of playing weekend home games on back to back nights. I could never understand it and I read last year that Lowe said they were asking for this but they no longer would
It was to save on travel and operating costs.

dawgbone is offline  
Old
03-24-2004, 04:00 PM
  #9
Matts
Registered User
 
Join Date: Nov 2002
Posts: 2,911
vCash: 500
Yeah

but it arguably cost them wins and points and fortunately they realized it

Matts is offline  
Old
03-27-2004, 06:05 PM
  #10
igor*
 
Join Date: Mar 2002
Posts: 3,276
vCash: 500
Quote:
Originally Posted by Matts
how the Oilers don't do themselves any favours by asking the league for those extra home games vs the Leaves and Habs. I know the season ticket holders want good matchups but maybe they'd be happier if the Oilers had a better sked.

It wasn't too long ago that the Oilers were doing this stupid thing of playing weekend home games on back to back nights. I could never understand it and I read last year that Lowe said they were asking for this but they no longer would
Yeah, the Oilers market the weekend packages pretty hard. As well as the 7 game Mini-Packs that are made up of weekend games. It helps them sell tix to some of the visiting teams that are poor draws. (i.e. when you buy your 7 game weekend mini-pack you get a few good games and three iffy ones).

And a lot of fans come in from out of town to catch a couple Oiler games in one weekend. It is an important group of customers for the Oilers.

This isn't Lowe's focus though, he's on the hockey operations side. He did blurt out his frustration with this scheduling issue last season however (I think we probably both heard the same interview) ... Lowe was getting pretty pished about the affect that this marketing strategy was having on the team's results, and vowed that they would change it for next season. And they have (Thank Christmas!).

You seem to see a lot of Friday and Sunday games now ... which is good I think. Makes it a bit tougher for the out-of-towners ... but still doable. They tend to start the SUN games early (6-ish) unless there are TV commitments. And I think they should be starting the Friday games later. 8:00 or 8:30 methinks. Partly to help out the fans that are travelling into E-Town to see the game ... but mostly because late-starting games on the weekend always seem to have a fun, loud crowd.

igor* is offline  
Old
03-27-2004, 06:28 PM
  #11
igor*
 
Join Date: Mar 2002
Posts: 3,276
vCash: 500
Quote:
Originally Posted by oilswell
...
Another possibility is to look at teams that change substantially in front of the goaltender. If a good defense is weakened by injury or trade, then either the goalie's SV% will change or remain about the same. A large change would suggest that goalie SV% are substantially dependent on the abilities of the defense in front of them. The implication is a causal relationship, although obviously no hard conclusions would be possible unless (a) the statistical power was high enough to draw firm conclusions, and (b) one could observe the cause-effect relationship in some way. Still, a large swing in SV% due to changes in D would be enough to raise doubts about the belief that goalie SV% is almost solely a result of a goalie's ability/achievements.

This year presents two cases that I could think of. In the following table are the pertinent goalie stats for two goalies, Martin Brodeur and Dan Cloutier. Brodeur played almost half the season without Scott Stevens in front of him (injury). Cloutier played a smaller number, but still a substantial number without Jovanovski. Its relatively uncontentious to say that Stevens and Jovo are anchors of their respective defenses.
Code:
GOALIE           DEFENSE  GP  GA   SA  SV%   SA/GP
--------------------------------------------------
BRODEUR  with    STEVENS  37  68  885  0.923  23.9
         without STEVENS  32  74  817  0.909  25.5
CLOUTIER with    JOVO     35  79  967  0.918  27.6
         without JOVO     19  48  448  0.893  23.5
...
Good stuff, oilswell

Though I think we can agree that 2 players is a remarkably small sample size. Just looking at Jovo ... does this mean that when he is in the lineup that more shots are allowed? But that these are of lower quality? I dunno, but that doesn't make sense to me intuitively, I think the opposite ... when he is on his game the puck doesn't spend much time in his end of the ice. But when the high-risk plays go wrong ... a glorious chance the other way.

If you are keen ... you could write a script that automatically would run this calc for every player in the league that missed a significant chunk of games. I suppose you could run their EV+/- vs the affect on SV% when they were injured ... unfortunately the fact that special teams and even strength numbers aren't divorced will muddy things a bit.

Still, it shouldn't be too difficult to do if you are keen. I expect that you would get a distribution curve remarkable similar to a random distribution. And it would be skewed to the side less than .005 points I'd also guess that centres have the biggest impact on goalie SV% ... though still not a bunch.

igor* is offline  
Old
03-27-2004, 07:34 PM
  #12
Oi'll say!
Registered User
 
Oi'll say!'s Avatar
 
Join Date: Nov 2002
Location: Oil in 9
Country: Canada
Posts: 12,017
vCash: 500
Quote:
Originally Posted by igor
Good stuff, oilswell

Though I think we can agree that 2 players is a remarkably small sample size. Just looking at Jovo ... does this mean that when he is in the lineup that more shots are allowed?
[edit: it's a small sample size, but the numbers are drastic. A small change could be mostly ignored] I don't understand why you would assume that a better defenseman would allow more shots on net than a poor one.The ultimate goal of any defenseman is to limit the number of shots on net, if you have better defensemen your goalie will face less shots by definition.

Quote:
But that these are of lower quality?
If a defenseman wants to get paid more money he will be probably want to allow fewer quality chances.
Quote:
I dunno, but that doesn't make sense to me intuitively, I think the opposite ... when he is on his game the puck doesn't spend much time in his end of the ice. But when the high-risk plays go wrong ... a glorious chance the other way.
That's exactly what did happen to Jovo in the playoffs imo, he played a very high risk/reward game and got burned bigtime. He scored 7 goals but was a minus 5 himself.


I wonder what Salo's sv% was before and after Ulanov, there were a lot of games where Fergie/Bergie or Fergie/Semenov or even Semi/Bergie were both in the lineup at the same time before Iggy came. That hasn't happened since.

Oi'll say! is offline  
Old
03-27-2004, 08:54 PM
  #13
igor*
 
Join Date: Mar 2002
Posts: 3,276
vCash: 500
Quote:
Originally Posted by Oi'll say!
... I don't understand why you would assume that a better defenseman would allow more shots on net than a poor one.
I don't, you've misread my post.

Quote:
... it's a small sample size, but the numbers are drastic
Exactly. Which is why you have to look at the bigger picture. There will always be outlying numbers ... if there weren't, I'd be completely shocked.

Oilswell is a great poster, I always enjoy reading his stuff. And quite frankly, I'm pretty sure that he's smarter than me ... and he should have figured this out by now. And he hasn't. And if he wanted to ... he could have figured out how to separate the ES SV% numbers on a game-by-game basis. (props to mudcrutch for realizing just how significant this is ... some spooky stuff comes from this)

In the thread that oilswell refers to I gave him a link to the financial statements of an online bookmaker (as opposed to oddsmaker). Some things should have jumped out at him, but I don't think he's interested. And, as an aside, I would have at least expected a comment like "wtf is goodwill? and why is this number so huge ?!?" Some things never change. Human nature is human nature.

OR: "How in gawd's name can they run a 2% margin on a hockey line when they clearly need to average 7% to be viable?"

When oilswell talks about 'Poisson's equation' what he's really saying is that goals-scored is darn near random ... and that chances created are talent-driven. And when he says "that looks like a Poisson distribution" what he's really saying is "it's completely *** random". Of course it isn't ... momentum swings smudge it a bit, as much as 6% from completely random, and that's depending on what you're looking at. And there are reasons for that ... and theoil, Digger and a couple others have inadvertantly (I think) lead me to two of them. Remarkable intuit on a few folks here . Quite frankly ... that is the reason that a conversation like this is still worth having on a public forum (I'm not intentionally trying to bore most of you to tears here with this dry stuff )

People are people. A game is a game. And the same shyte happens in the same way, over and over, year after year after year after year. Believe it or not ... it really is completely inevitable. And the best part ... it's intuitively unlikely.

Anyhoo ... we're going to have to agree to disagree. Either of us may be right, wrong or indifferent ... who cares?

igor* is offline  
Old
03-28-2004, 01:10 AM
  #14
Oi'll say!
Registered User
 
Oi'll say!'s Avatar
 
Join Date: Nov 2002
Location: Oil in 9
Country: Canada
Posts: 12,017
vCash: 500
Yeah we disagree, no harm done Igor. I don't mean to sound confrontational in my responses but I try to keep them short and to the point, a lot of meaning can get lost in a verbose obscure reply.

As far as misreading your post is concerned, while you didn't say that a top d-man like Jovo would allow more shots on net, you steered the line of questioning in that direction which would seem like a red herring to me. The chances of that being the case are slim in the extreme and would be highly illogical.

As far as your argument for outlying numbers, of course there will be some. They would be easily predictable given enough data by using the standard of deviation formula from high school math but I doubt I have to tell you that. What you and I don't know yet is whether or not these fall under the category of "outlying numbers" at all, they could be the norm or just a fraction off of it. Furthermore, noone has said "case closed" yet, so all Oilswell is doing is starting up a little data pool using the two most relevant cases and we're commenting on it.

Hard pressed, yeah I'll admit that I'm a little peeved by this topic. Not by the intellectual diversion it provides, but by the fact that save % was brought up in the first place as a means of "proving" Tommy Salo was the goat here. Nothing of the sort has ever been done, and I have no doubt whatsoever that the fact the Oilers have broken in 7 brand-spanking new defensemen and 3 or 4 centers in front of him in the past two years has had it's toll on him in more ways than one. Not just that I'm seeing it in retrospect either, when the Oilers were poised to start the 2002/03 season with only 5 defensemen that had ever played 1 or more nhl games I called it a mistake in advance. 1 Starting goalie later I may have a case.

Oi'll say! is offline  
Old
03-28-2004, 02:25 AM
  #15
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Quote:
Originally Posted by igor
Though I think we can agree that 2 players is a remarkably small sample size.
Its small. But maybe not in the way you're thinking.

Sample sizes are of interest when one wants to be sure one can generalize across a population. With only 2 players...obviously non-randomly sampled (key defenders)...I wasn't aiming for solid generalization over all goalies and all defensemen. While that would be nice, its a bigger step.

Instead, the analysis was primarily set up to assess the impact of Stevens on Brodeur's SV% and Jovo on Cloutier's. I wanted to do both because it hints at whether it might be a generalizable statement about #1 defenders, but also because having the two might help in exploring the issues. Anyway, in such a study, samples are games played (or minutes played, etc.). If the sample size is too small, the impact of the defender on SV% is lost to the vagries of the games, opposition, home/away, etc. In the case of Brodeur, there are 37 and 32 games in the two conditions, respectively. About half a season.

Is that too small a sample? Yes. A single-factor ANOVA test measures F=2.01 and p=.16 between the two conditions (Stevens, no Stevens). Not terrific. Statistically significant at the p<.16 level isn't great. If I can figure the stats implications correctly it means this data says...err...its a 4 in 25 chance that you'd be wrong if you conclude Stevens affects Brodeur's SV%. Hardly conclusive, but not the same as picking up an outlier. I didn't bother posting this as the more interesting issue to me is whether the excellent hockey people here can support it or kill it with argument. (BTW the conclusions might possibly be firmed up by meta analysis of Stevens and Jovanovski's data combined: statistics gurus enjoined to try!)

Does it mean all top defense have an important impact on a goalie's SV%? Impossible to say from this, as there isn't enough data.

Quote:
Just looking at Jovo ... does this mean that when he is in the lineup that more shots are allowed? But that these are of lower quality? I dunno, but that doesn't make sense to me intuitively, I think the opposite ... when he is on his game the puck doesn't spend much time in his end of the ice. But when the high-risk plays go wrong ... a glorious chance the other way.
I dunno either. Your argument about Jovo does make sense because of the type of defender he is, however it would tend to lower the SV% and (if we accept the data in the table for a moment), that doesn't appear to happen.

Quote:
you could write a script that automatically would run this calc for every player in the league that missed a significant chunk of games. ... I expect that you would get a distribution curve remarkable similar to a random distribution.
No doubt. You'd expect great players and bad players. Sampling every player in the league will tell you exactly nothing (in the naive case) about the impact on goalie SV% just because they're expected to even out. The issue at hand is whether a top-10 (say) defense makes a significant difference on SV%. There can't be more than a couple of handfuls of defenders that, by themselves, have this level of impact.


Last edited by oilswell: 03-28-2004 at 02:29 AM.
oilswell is offline  
Old
03-28-2004, 02:38 AM
  #16
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Quote:
Originally Posted by Oi'll say!
I'm a little peeved...by the fact that save % was brought up in the first place as a means of "proving" Tommy Salo was the goat here. ...I have no doubt whatsoever that the fact the Oilers have broken in 7 brand-spanking new defensemen and 3 or 4 centers in front of him in the past two years has had it's toll on him in more ways than one.
Yes, this was one of the motivations for looking into the allegation that Salo's numbers were influenced by the team in front of him. To illustrate the issue with a question: if the mighty Brodeur punches in a .909 behind a strong defensive team, but minus Stevens, what would he do on the Oilers? (Frankencross, Ferguson, Brewtal, overworked Staios, etc., etc.)

I have serious doubts whether these issues can be illuminated with a statistical study. But I like the idea of at least trying.

oilswell is offline  
Old
03-28-2004, 12:31 PM
  #17
Oi'll say!
Registered User
 
Oi'll say!'s Avatar
 
Join Date: Nov 2002
Location: Oil in 9
Country: Canada
Posts: 12,017
vCash: 500
Quote:
Originally Posted by oilswell
Yes, this was one of the motivations for looking into the allegation that Salo's numbers were influenced by the team in front of him. To illustrate the issue with a question: if the mighty Brodeur punches in a .909 behind a strong defensive team, but minus Stevens, what would he do on the Oilers? (Frankencross, Ferguson, Brewtal, overworked Staios, etc., etc.)

I have serious doubts whether these issues can be illuminated with a statistical study. But I like the idea of at least trying.
Good job, keep fighting the good fight oilswell.

Proving "Defense influence on SV%" could be as hard as you want to make it I guess, some idiot filled a whole book with a proof that one + one = 1 . I just don't think that our side is the one that has to be proven to the greatest extent seeing as it appears to be the common sense side of the argument, and stats that fall into the realm of the expected don't require quite as much investigation as the ones that go against the grain. Basically, where we find two strong cases that support our theory, I'd like to see three very solid numbers going the other way to keep pace. It's till 2-0.

Like they say on Vcr Isl, Oilswell that ends well.

Oi'll say! is offline  
Old
03-28-2004, 12:52 PM
  #18
igor*
 
Join Date: Mar 2002
Posts: 3,276
vCash: 500
Quote:
Originally Posted by oilswell
No doubt. You'd expect great players and bad players. Sampling every player in the league will tell you exactly nothing (in the naive case) about the impact on goalie SV% just because they're expected to even out. The issue at hand is whether a top-10 (say) defense makes a significant difference on SV%. There can't be more than a couple of handfuls of defenders that, by themselves, have this level of impact.
Au contraire, oilswell. Sampling every player in the league will maximize the validity of your result.

I think the point you are trying to prove or disprove is that good defensive players help a goalie's save percentage, no?

Simpler yet (and I'm presuming this is what you are trying to arrive at):
Does a good defensive skater get better results mostly because fewer shots are allowed when he is on the ice?
Or mostly because the shots allowed are of lower quality?

And conversely for a player with a high goals-against rate. And for the average players as well.

So you would be looking at a skaters:
ES ice-time and EV- are what you need to know, as well as which games they played in by the NHL game number.
And the SA and SV in those games and which goalie was in net. Apply common sense and stir.

If you just watch a few games with an unbiased mind this should become obvious anyways.

igor* is offline  
Old
03-28-2004, 01:00 PM
  #19
Oi'll say!
Registered User
 
Oi'll say!'s Avatar
 
Join Date: Nov 2002
Location: Oil in 9
Country: Canada
Posts: 12,017
vCash: 500
I think it's time to find some valid numbers that support your case and bring them forth igor, the ball is in your court.

Oi'll say! is offline  
Old
03-28-2004, 02:07 PM
  #20
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Quote:
Originally Posted by Oi'll say!
Good job, keep fighting the good fight oilswell.
Truthfully I'm only fighting against our collective ignorance. I will cheerfully accept SV% is 90% goalie if it turns out to be like that, but I just haven't yet seen anything that is convincing (I've already laid out what weakness I see in igor's otherwise outstanding previous post). I'd like to believe I'm not chasing phantoms here but its not clear to me that it is true. More later as I have time.

oilswell is offline  
Old
03-28-2004, 02:27 PM
  #21
oilswell
Registered User
 
Join Date: Mar 2002
Location: Wandering the globe
Posts: 1,760
vCash: 500
Quote:
Originally Posted by igor
Au contraire, oilswell. Sampling every player in the league will maximize the validity of your result.
Will consider it carefully. Not convinced yet.

Quote:
I think the point you are trying to prove or disprove is that good defensive players help a goalie's save percentage, no?
Sort of. I'm trying to disprove or limit the conjecture that a goalie is the lone primary contributor to SV%. While I accept that SV% is a useful yardstick - especially within-team - I am not convinced it is comparable between-teams. If not, I'd want to know what non-goalie factors affect it and to what extent. It might have to be on a case-by-case basis. In this case I was wanting to figure out how important the Devel's top notch D is to Brodeur's SV%. Stevens was being used as an indicator of this because we don't have data on Brodeur with, say, Tampa Bay or Oilers' defense corps in front of him instead. So a poorer defense corps is being "simulated" by taking Stevens out and subbing in someone else with little else changing on the team.

Quote:
Simpler yet...
Does a good defensive skater get better results mostly because fewer shots are allowed when he is on the ice? Or mostly because the shots allowed are of lower quality?
While I definitely care what the answer is to this (and I was thinking aloud about it so maybe I shouldn't have done that), I think it is immaterial to the question I am asking, which is a factor strength question about goaltending SV%. One explanation for the seasons the Sharks and Flames have had is that their goalies have been incredible. Another explanation is that their defense/systems/opponents/etc reduce the quality of shots against, making their goalies look better than they otherwise would be. A third explanation is a combination of both. Your question, I think you'll see, is immaterial to that.

Quote:
If you just watch a few games with an unbiased mind this should become obvious anyways.
You keep saying that things become obvious with watching...and I'm trying, honestly. I've even rewound on some goals but you're an ex-coach and I'm not. You may have to hold me up as an example of a thick non-Leafs fan that can't see the obvious without help from raw data.

oilswell is offline  
Old
03-28-2004, 02:55 PM
  #22
mudcrutch79
Registered User
 
Join Date: Jul 2003
Location: The Big Smoke
Posts: 3,903
vCash: 500
I'd just have one comment to make about this idea. While I can certainly see the value in what you're trying to do here, it seems to me that by focusing solely on the question of what impact there is on the sv% of the goalie, we're missing the three dimensional picture. Every player on the ice, in a rational world, will make his decisions based on the probability of a positive outcome. In hockey, that positive outcome is scoring goals, and you need to weight that against the risk of giving one up. Ed Jovanovski and Scott Stevens are different players and Jovo probably has a higher risk tolerance-more importantly, so do his coaches. Thus evaluating what happens when he is out can't really be compared to what happens when other guys out.

The other problem here is that this is all relative-there is the rest of the team to consider. I don't know when Jovo went out, but, and correct me if I'm wrong, but didn't the whole Bertuzzi fiasco go down during that span? That had a serious impact on the Canucks as a team, and I'd suspect you can explain away some of the difference with that.

Additionally, there is the factor of who remains after the player is gone. Taking St. Louis as an example, they lost Pronger last year, but the huge letdown was probably somewhat mitigated by the emergence of Jackman. I don't know how this can be accounted for.

I believe that I'm on oilswell's side on this one, I tend to think that the players on the ice have an impact on save percentage, but I also think that the volume of shots will generally produce a higher save percentage, although I have yet to prove this or look into it. Maybe once I'm done these damn exams...

mudcrutch79 is offline  
Old
03-28-2004, 10:52 PM
  #23
Oi'll say!
Registered User
 
Oi'll say!'s Avatar
 
Join Date: Nov 2002
Location: Oil in 9
Country: Canada
Posts: 12,017
vCash: 500
Quote:
Originally Posted by mudcrutch79
Ed Jovanovski and Scott Stevens are different players and Jovo probably has a higher risk tolerance-more importantly, so do his coaches. Thus evaluating what happens when he is out can't really be compared to what happens when other guys out.
Look at it this way. When a $5M a year nhl'er is injured you use your best possible ahl'er or a guy who is a pb regular in the lineup. Basically you're making a short term trade, Jovo for Bryan Allen etc. While Bryan isn't taking all of Jovo's minutes the fact remains that less skilled blueliners are getting mins that Jovo would have had, and in the regular season Jovo is a reliable d-man. The goalie is now behind a less experienced, less skilled d-corps.

Quote:
The other problem here is that this is all relative-there is the rest of the team to consider. I don't know when Jovo went out, but, and correct me if I'm wrong, but didn't the whole Bertuzzi fiasco go down during that span? That had a serious impact on the Canucks as a team, and I'd suspect you can explain away some of the difference with that.
Jovo's injury happened long before the Bert incident, he missed 54 games and only about the last 8 were post-BertMoore.

Quote:
Additionally, there is the factor of who remains after the player is gone. Taking St. Louis as an example, they lost Pronger last year, but the huge letdown was probably somewhat mitigated by the emergence of Jackman. I don't know how this can be accounted for.
This type of thing would lessesn the blow to the goalie, but I'd bet that the stats were still effected quite a bit. While Jackman was good, he wasn't as good as Chris Pronger, and his mins at the start wouldn't have been great either. He would have had to have earned those through steady play. My guess is that in the early stages the #2 through 6 guys ate up the lion's share of icetime.

Quote:
I believe that I'm on oilswell's side on this one, I tend to think that the players on the ice have an impact on save percentage, but I also think that the volume of shots will generally produce a higher save percentage, although I have yet to prove this or look into it. Maybe once I'm done these damn exams...
There are times when a goalie stops 41 of 42 and it bloats his save %, but the goalie who faces 19 shots a night and lets in one or two will eventually prevail in the stat department.

Eg - in NJ Stevens and White are keeping guys to the outside, clearing rebounds, tying up guys in the crease area, most of the time all Brodeur has to do is focus on the puck, make the save and control the rebound.

A goalie playing behind a d-corps with one or two raw, 1st game nhl'ers isn't facing anywhere near that same situation. The rooks will get beaten more, cough up the puck more, miss more checks, clear less rebounds, make poorer reads, the goalie will have to play the shooter with half a mind on the guy who's open beside the net, the shot will come from closer, at a better angle, the guy will have more time to fire it, etc, on a certain percentage of shots.

At least that's my guess.

I think that Stevens played better in his 885th nhl game than Bergeron did in his 2nd, or Henry in his 4th, or Semenov, or Luoma, or Allen, or Haakana, or Pisa, or even Lynch. I think there were a lot of missed asignments by those guys that Stevens wouldn't have made. I think that those missed assignments lend themselves to a bevy of opportunities that will go into the net more than .090 of the time. But that's just a guess mind you. Somewhere there exists proof that I'm wrong, someone just doesn't want to reveal it until it's proper time.

Oi'll say! is offline  
Closed Thread

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 05:40 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. 2014 All Rights Reserved.