Adjusted Save Percentage

[quoipourquoi]

What does the amount of time a goalie spends on the powerplay have to do with his save percentage?

Literally? Nothing.

What it does affect, however, is the type of shots a goalie faces, and by that, I mean that a team is more likely to score with a powerplay shot than an even-strength shot, because the man-advantage allows for better passing plays, and good powerplay teams will often pick-and-choose carefully enough to get a goal on just one or two shots, whereas if they had that much room in an even-strength situation, they'd probably just take a shot immediately rather than count on the two-minutes of offensive hockey to allow for an even better chance.

But by disregarding powerplays for this reason (or shorthanded situations for the reason of the increased likelihood of breakaways), we'd be opening up a can of worms, because not all shots are equal, and not all goaltenders are universally good on any type of shot. In the end, the most fair way to do it is the binary opposition of SAVE and GOAL, because if you go into it more deeply than that, the study will have unintentional bias.

 

[overpass]

I really don't know why you are doing this by year and not by season. Upon first look I can tell you that 1996-1997 is off. In 1995-1996 Brodeur played in 90% of the teams games. Any backup samples I find irrelevant. 1994-1995 was a shortened season. I'm not sure where you came up with these calculations but in '96-'97 the Devils allowed 25.5 shots per game while Brodeur was in net and 27 shots per game while his back ups were in net. As I pointed out earlier in '97-'98 it was something very similar. The following three seasons Brodeur was very close in SAPG with Terreri but then you have to look at quality of opponent and save %. Post-lockout doesn't really interest me. I think Brodeur prevented less shots because of the new rules.

I am doing it by season. I apologize for the ambiguity in the presentation. 1994 refers to the 1993-94 season, and so on.

As for the rest of it, I think if you look again you'll find the numbers are correct. I already said that I don't put a lot of weight on any single year, but if you take a weighted average of the difference you can get a good estimate of the effect. I ran the numbers to estimate the effect of quality of opponent. What exactly is your problem with these numbers? Did you even read the rest of my post after the table?

I have a little trouble believing in less than 2 rebounds a game, but there's another thing wrong with the conclusion drawn from the study. A rebound doesn't have to result in a quick shot to be a rebound. A shot wide off a rebound is still a rebound. A rebound that results in opponent possession is still a rebound even if they set up behind the net or on the boards, or pass back to the point, and get a shot more than 2 seconds later.

My main point in posting that link was to draw attention to Brodeur's quote that he prevents 8 to 10 shots per game with his puckhandling. If I'm skeptical of the ability of Devils fans and others to estimate the magnitude of Brodeur's shot prevention by watching the game, it's because Brodeur himself so obviously overrates his shot prevention abilities. I won't attack his motives for doing so, he's probably just overrating the impact of each individual play and implicitly comparing himself to a goalie who never handles the puck, rather than another NHL goalie.

 

[BM67]

My main point in posting that link was to draw attention to Brodeur's quote that he prevents 8 to 10 shots per game with his puckhandling. If I'm skeptical of the ability of Devils fans and others to estimate the magnitude of Brodeur's shot prevention by watching the game, it's because Brodeur himself so obviously overrates his shot prevention abilities. I won't attack his motives for doing so, he's probably just overrating the impact of each individual play and implicitly comparing himself to a goalie who never handles the puck, rather than another NHL goalie.

What is a rebound? If a goalie stops a shot that is going wide and gives up a rebound, is it really a rebound since the first shot wasn't a shot? I'm pretty sure the goalie is going to call it a rebound regardless of what it is called in the stats, and regardless of it resulting in another shot.

 

[overpass]

What is a rebound? If a goalie stops a shot that is going wide and gives up a rebound, is it really a rebound since the first shot wasn't a shot? I'm pretty sure the goalie is going to call it a rebound regardless of what it is called in the stats, and regardless of it resulting in another shot.

I don't disagree with your definition of a rebound. It's this quote that I disagree with.

"We give up 8-10 fewer shots a game here than other places because of the way I control rebounds."

Brodeur gave a clear estimate about the effect of his rebound control compared to other NHL goalies. I think his estimate is 100% incorrect. While I'm sure he knows every detail of his rebound control, either he isn't very good at estimating how many shots it prevents and/or he isn't very good at comparing his game to what they do in "other places".

 

[BM67]

I don't disagree with your definition of a rebound. It's this quote that I disagree with.

"We give up 8-10 fewer shots a game here than other places because of the way I control rebounds."

Brodeur gave a clear estimate about the effect of his rebound control compared to other NHL goalies. I think his estimate is 100% incorrect. While I'm sure he knows every detail of his rebound control, either he isn't very good at estimating how many shots it prevents and/or he isn't very good at comparing his game to what they do in "other places".

Overestimation? Possibly, maybe even probably, but by how much has as much to do with how you read his statement as anything.

"We always give up 8-10 fewer shots on goal a game here than all other places because of the way I control rebounds."

"We sometimes give up 8-10 fewer shots at net a game here than some other places because of the way I control rebounds."

 

[TheDevilMadeMe]

Florida has always been a pretty poor defensive team, though. is it so hard to believe that they'd always allow high shot totals?

THe difference between home and road shots does indicate a loose definition of "shot."

For me, it was Craig Anderson leading the league in save % for some time last year that made me really skeptical about Florida's shot counts. There is actually less of a difference between Florida's home and away than I thought there would be.

Nashville is the team that would appear to have the loosest definition of a "shot," it would appear, and by no small margin. This doesn't really surprise me either, as they always have a goalie near the top of the league in save %, yet they never stick for very long.

 

[seventieslord]

I have a little trouble believing in less than 2 rebounds a game, but there's another thing wrong with the conclusion drawn from the study. A rebound doesn't have to result in a quick shot to be a rebound. A shot wide off a rebound is still a rebound. A rebound that results in opponent possession is still a rebound even if they set up behind the net or on the boards, or pass back to the point, and get a shot more than 2 seconds later.

this is true; however, it could be said that if your rebound went to a point where the only shot the opponent could manage went wide, then you did pretty well with your rebound. That is shooter-dependent, and also dependent on your coverage of the angles. Game-by-game, these results (rebound shots allowed) wouldn't tell us anything, but over the course of a season or career, I think they would.

Just curious, BM, because you are always very logical and reasonable and have a very solid understanding of statistics, but you're also a Brodeur/Devils fan: Where do you currently rank Brodeur all-time among goalies?

 

[seventieslord]

Literally? Nothing.

What it does affect, however, is the type of shots a goalie faces, and by that, I mean that a team is more likely to score with a powerplay shot than an even-strength shot, because the man-advantage allows for better passing plays, and good powerplay teams will often pick-and-choose carefully enough to get a goal on just one or two shots, whereas if they had that much room in an even-strength situation, they'd probably just take a shot immediately rather than count on the two-minutes of offensive hockey to allow for an even better chance.

But by disregarding powerplays for this reason (or shorthanded situations for the reason of the increased likelihood of breakaways), we'd be opening up a can of worms, because not all shots are equal, and not all goaltenders are universally good on any type of shot. In the end, the most fair way to do it is the binary opposition of SAVE and GOAL, because if you go into it more deeply than that, the study will have unintentional bias.

I'm talking about time spent by a goalie with his team on the powerplay. This was brought up by haakon and he has not explained what this has to do with anything.

As for what you're saying, time spent facing the PP has a major effect on sv%. And yes, breaking it down does tell us a lot more than simply "binary" counting.

 

[quoipourquoi]

As for what you're saying, time spent facing the PP has a major effect on sv%. And yes, breaking it down does tell us a lot more than simply "binary" counting.

But if you remove or isolate powerplays from the equation because of the increased likelihood of goals, then how do you not open your analysis to the justification of removing goals caused by defensive miscues and breakdowns that- like powerplays- put the goalie in the position of having to make a stop on a high percentage shot?

 

[overpass]

Since I attempted to measure Brodeur's shot prevention effect, it's only fair that I do the same for Hasek and Roy.

Using the same method of looking at shots against, calculating the difference against backups season by season, and weighting for minutes played, I get the following estimates:

Hasek: 1.30 fewer shots faced per game
Roy: 0.65 more shots faced per game

This suggests that if Hasek and Roy affected their teams shots against, Hasek did so in a positive way and Roy did so in a negative way. Keep in mind that there may be quality of opposition effects as there were with Brodeur, but I don't want to take the time to calculate that. Most likely both faced slightly stronger teams on average than their backups.

I won't post the year by year results, but Roy consistently faced more shots than his backups in Montreal, less so in Colorado. Hasek's best years for shot prevention by this method came after he left Buffalo in Detroit and Ottawa, as well as 1993-94 vs Grant Fuhr.

 

[seventieslord]

But if you remove or isolate powerplays from the equation because of the increased likelihood of goals, then how do you not open your analysis to the justification of removing goals caused by defensive miscues and breakdowns that- like powerplays- put the goalie in the position of having to make a stop on a high percentage shot?

I have no interest in "removing" PPs from the equation. but they should be looked at separately where possible because of the obvious skewing effect they will have. If every goalie faced the same percentage of their shots on a PP, there would be no need.

What you are telling me in a roundabout way is that shot quality-adjusted sv% is more important than unadjusted. This is correct. With detailed shot data, a more detailed picture of shots can be formed, and a goalie who faced 24 shots could have had a tougher night than a goalie who faced 30. But:

- We don't have detailed shot data going back forever. We have PPOA data, though, and adjusting sv% for PPOA is over half the battle right there.

- And, it's not hard to tell by watching Brodeur and the Devils play, that any adjustments made to shot totals due to quality would only make his numbers look less impressive, not more. The Devils surrender low-quality chances more frequently than other teams.

 

[seventieslord]

Since I attempted to measure Brodeur's shot prevention effect, it's only fair that I do the same for Hasek and Roy.

Using the same method of looking at shots against, calculating the difference against backups season by season, and weighting for minutes played, I get the following estimates:

Hasek: 1.30 fewer shots faced per game
Roy: 0.65 more shots faced per game

This suggests that if Hasek and Roy affected their teams shots against, Hasek did so in a positive way and Roy did so in a negative way. Keep in mind that there may be quality of opposition effects as there were with Brodeur, but I don't want to take the time to calculate that. Most likely both faced slightly stronger teams on average than their backups.

I won't post the year by year results, but Roy consistently faced more shots than his backups in Montreal, less so in Colorado. Hasek's best years for shot prevention by this method came after he left Buffalo in Detroit and Ottawa, as well as 1993-94 vs Grant Fuhr.

Interesting, considering those who criticize Hasek's high sv% claim he wanted his team to let him face as many shots as possible.

 

[haakon84]

I think a more telling way to look at Brodeur's shot prevention is to look at his PP shots against pre-lockout then post-lockout. With the trapezoid in tact it made him a considerably less mobile goaltender and this effected him on the PK. His shot totals per power play jump significantly post-trapezoid yet he saw the just about the same amount of PP's. Defying all odds his PK save % did not seem to waver considering he faced more high quality shots. Crazy I know.

 

[overpass]

I think a more telling way to look at Brodeur's shot prevention is to look at his PP shots against pre-lockout then post-lockout. With the trapezoid in tact it made him a considerably less mobile goaltender and this effected him on the PK. His shot totals per power play jump significantly post-trapezoid yet he saw the just about the same amount of PP's. Defying all odds his PK save % did not seem to waver considering he faced more high quality shots. Crazy I know.

Do you have numbers to back that up?

 

[seventieslord]

OK, let's take it over here, where critical discussion of the all-time greats is the norm and not the exception:

from http://hfboards.com/showthread.php?t=717035&page=17

No there are plenty of holes. You are providing very one dimensional data and throwing a blanket statement out there applying it to all goalies. Again of course PP shots are harder and I would like to know each goalies save % on the Power Play.

So do I, but prior to 2000, we will never know. But they all have lower sv% versus the PP than they do at ES, and that is what matters. Face more PPs, you will face more high quality shots.

How many shots they face per Power Play.

Over time, relatively the same. If there was a way to determine this, which there isn't, I doubt you would see more than a 10% discrepancy. Whatever shtos they do face, though, are more difficult and detrimental to sv%.

How long the opposing teams are on the Power Play.

Again, over time, relatively the same. We will never have this data. Not all PPs are created equal, but if you contend that one team frequently killed off 5 minute PPs and another was frequently scored on in 10 seconds and it created a massive statistical edge, I disagree. These things wash out in the long term. Each PP, over time, will roughly be the same average length. Using PPOA is the best way to approximate the percentae of shots a goalie faced on the PP, and is more than reasonable.

What rank were the special teams for each goaltender.

This is useful to determine how good the team was at killing the penalties, but unlike most team stats, this is quite dependent on the goalie too, so it wouldn't tell you a lot. The goalie, after all, is the only one who plays all two minutes of every PP. As it relates to Brodeur, how good or bad he and New Jersey are at penalty killing is irrelevant compared to the fact that they are both much more vulnerable to have goals scored on them in this situation than otherwise.

This is all information pertinent to the discussion at hand. If you fail to see that then we are speaking different languages. You're just going to look and be concerned with save % and thats fine but I'm not happy with inconclusive and incomplete data. I don't see a correlation between Nabokov's save % and the amount of Power Plays he faced nor do I see one with Hasek. I am also sure there are other examples like this. Therefor I don't think necessarily facing more PP shots causes a goaltender to have a lower save %. If you take all the bad teams out there that have horrible penalty killing (such as the expansion teams in the 90's) the stats will certainly be skewed. Hasek never had a horrible penalty killing squad. Buffalo was a defensive minded team (not to the extent of NJ) but he was never on a lesser team like Ottawa or Tampa Bay.

You don't see a connection? Fine, then show me all the goalies who have had a better sv% versus the PP compared to at ES over a season. you can't.

Take a look at last season. I can't post a link that will work, but go to nhl.com, stats, players, 2008-09, regular season, all teams, goalies, all, all, special teams.

The only goalie who even comes close to his ES sv% on the PP is Backstrom (.918 vs. .923). Only two more goalies with 40 GP were within 10 points - Lundqvist (.913 vs. .920) and Kiprusoff (.898 vs. .907). Almost every other goalie had at least a 20 point difference, some as high as 40.

To look at Brodeur in particular: He had an excellent ES sv% of .933, but not so hot on the PK (.848) - naturally time spent (shots faced that is, but same thing) on the PK lowers his save% and it came out at .916. He faced 21% of those shots on the PP. If he maintained those same situational percentages and faced even 25% of those shots on the PP, his sv% would be .912. If New Jersey had been even more disciplined and he faced just 15% of those shots on the PP, his sv% would be .920. It has a major impact and none of your "questions" change that it does.

I can say that Brodeur is a more disciplined goalie than Hasek and Roy so he has very very very little to do with the Devils taking fewer penalties. But yes you can say he does have an effect. It does matter how hard or easy his power play shots are. This is the problem with save % in general. His intangibles are magnified when the Devils are on the Penalty Kill and will skew his save % accordingly. The fact is the Devils have had an easier time on the Penalty Kill with Brodeur in net than they would have with Hasek or Roy. Boy that was easy. I think if you dig deeper there will be telling differences with having Brodeur in net on the PP instead of a lesser puck-handling, rebound controlling goaltender. Why would this information not be important? Again we won't be able to quantify his effect without the factors I'm interested in.

Any impact a goalie has on his team not taking penalties would be statistically insignificant. The Devils helped Brodeur by not taking penalties. This is indisputable and you're wasting a lot of time disputing it.

 

[seventieslord]

Dead Puck

ES 800 * .920 = 64 GA PP 400 * .880 = 48 GA Total (1200/1088) = .907
ES 700 * .920 = 56 GA PP 500 * .880 = 60 GA Total (1200/1084) = .903 -.04

80's

ES 800 * .900 = 80 GA PP 400 * .835 = 66 GA Total (1200/1056) = .880
ES 700 * .900 = 70 GA PP 500 * .835 = 83 GA Total (1200/1047) = .872 -.08

This is why I think Roy benefitted more from a more disciplined Habs team. I think the Power Play was much more dangerous in the 80's (about 5%) as well as 5 on 5 (I don't think it the difference is as big as on the power play BUT I may be wrong). I would like to get more evidence on whether or not they are proportionate to each other (difference in ES and PP save % from the 80's to the dead puck era).

We will never, ever, have that data. The numbers you made up are completely arbitrary. Yes, under this example the goalie in the 80s saw his save% drop by further due to the same PP shot percentage rise but the example is completely imaginary.

If it interests you that much, take a look at the correlation between league average sv% at ES and on the PP over the last 10 years, where the data is available. (Don't use 1997-98, those results are obviously wrong). What's the % difference between the two? Does this correlate to league average PP efficiency at all?

10 years of data is enough to go by. My guess is they will be very proportionate to eachother.

As I keep reminding you, although Roy had a beneficially disciplined team, the result was that he was first in sv% four times. Brodeur had the same benefit, and it only got him into the top-10 a few times.

 

[haakon84]

Any impact a goalie has on his team not taking penalties would be statistically insignificant. The Devils helped Brodeur by not taking penalties. This is indisputable and you're wasting a lot of time disputing it.

I know it's insignificant I was interested in seeing if that was the only point you'd single out.

Ok you're thread crossing but hey I don't care take a look at this data. This is probably a better place to discuss these matters.

This is interesting...

Power play shots per season.

x>500 shots = .88 save% ES - 0.929 save% Total - 0.915 save%
500<x<400 = .87 save% ES - 0.921 save% Total - 0.910 save%
400<x<300 = .87 save% ES - 0.920 save% Total - 0.910 save%
300<x<200 = .87 save% ES - 0.917 save% Total - 0.906 save%
200<x<100 = .86 save% ES - 0.912 save% Total - 0.901 save%

Each goalie faced around 21 shots per game on even strength surprisingly.

It seems the more the power plays the better the goaltenders are at even strength. They see the same amount of shots yet in a shorter amount of time. Is the defensive thing? Discipline thing? Rebound control? Puck-handling? Either way more Power Play shots seem to benefit the goaltenders save %.

Now how can you say facing less Power Play shots a game HELPS a goaltenders save %?

To look at Brodeur in particular: He had an excellent ES sv% of .933, but not so hot on the PK (.848) - naturally time spent (shots faced that is, but same thing) on the PK lowers his save% and it came out at .916. He faced 21% of those shots on the PP. If he maintained those same situational percentages and faced even 25% of those shots on the PP, his sv% would be .912. If New Jersey had been even more disciplined and he faced just 15% of those shots on the PP, his sv% would be .920. It has a major impact and none of your "questions" change that it does.

You pick this season? The one injury riddled season of his career. You can do better than that.

 

[seventieslord]

I know it's insignificant I was interested in seeing if that was the only point you'd single out.

Ok you're thread crossing but hey I don't care take a look at this data. This is probably a better place to discuss these matters.




Now how can you say facing less Power Play shots a game HELPS a goaltenders save %?



You pick this season? The one injury riddled season of his career. You can do better than that.

- I'm thread crossing for a reason....

- I already showed you that facing less PP shots helps sv%. Every single goalie allows a higher percentage on the PP. The more they face, the lower their sv% will be. it's common sense. I don't know what those figures are trying to say, or where they are even from.

 

[BM67]

Just curious, BM, because you are always very logical and reasonable and have a very solid understanding of statistics, but you're also a Brodeur/Devils fan: Where do you currently rank Brodeur all-time among goalies?

I believe I had it Plante, Hasek, Roy, Hall, Brodeur, Sawchuk, Dryden on my top 120 list. All but Dryden in the top 30.

 

[haakon84]

- I'm thread crossing for a reason....

- I already showed you that facing less PP shots helps sv%. Every single goalie allows a higher percentage on the PP. The more they face, the lower their sv% will be. it's common sense. I don't know what those figures are trying to say, or where they are even from.

It's clearly not common sense. The data I posted shows that no matter how many powerplay shots a goalie faces a season the save % average is 88%. What does change is the ES save %. Not even shots per game on even strength changes it still averages out to 21 shots per game. Meaning that goalies who face more shots on the PP face the same amount of shots per game on ES in a shorter amount of time. What does it all mean? I'm not sure but it certainly debunks your theory. It's data I ran from timeonice.com.

It shows the more PP shots a goalie faces the BETTER his save % would be. The opposite of your hypothesis.

 

[Bear of Bad News]

The data I posted

You still haven't sourced your data, which makes it very difficult for others to know how you arrived at your conclusions.

 

[haakon84]

You still haven't sourced your data, which makes it very difficult for others to know how you arrived at your conclusions.

I thought I explained myself when I said it's data I ran from timeonice.com...

If you would like to double check my calculations please do so. I am not accustomed to spreadsheets. I did double check myself and it seems to be pretty accurate. I just ran the averages for save %'s for ES, PP, and overall. I also calculated shots per game by taken the total shots for each criteria and divided it by the average amount of games which came to 21 shots per game across the board.

http://www.timeonice.com/saves.html

 

[seventieslord]

It's clearly not common sense. The data I posted shows that no matter how many powerplay shots a goalie faces a season the save % average is 88%. What does change is the ES save %. Not even shots per game on even strength changes it still averages out to 21 shots per game. Meaning that goalies who face more shots on the PP face the same amount of shots per game on ES in a shorter amount of time. What does it all mean? I'm not sure but it certainly debunks your theory. It's data I ran from timeonice.com.

It shows the more PP shots a goalie faces the BETTER his save % would be. The opposite of your hypothesis.

It's not a hypothesis - it's true. It's proven.

Now that I can finally make sense of what you've posted - you're basing this on shots per season. how about shots per game, and only using good goalies? To me, it would make perfect sense that the goalies who face more shots (i.e. play more games) would be the ones with the better numbers at both even strength and vs. the PP.

 

[greatgazoo]

Does anyone bother to take into account the size of equipment that they wore and the amount of breakaways and 2 on 1's that goalies faced?

Like for example, Fuhr during his prime usually had a save pct around .880 but the guy played for a team in front of him that considered playing defense as an after thought.

In addition, Patrick Roy wore pads that were considerably larger than Fuhr's were! Many of his saves, playing for defensive minded teams, were due to his equipment more than it was reflexes or anticipation.

 

[Bear of Bad News]

I thought I explained myself when I said it's data I ran from timeonice.com...

It's hard to check what you did when you don't explain *what* you did. Your reference to "each goalie" implies that you totalled things up by goalie - is that what you did?

If seventieslord is right about what you did (post above mine), then are you really surprised that goalies who play more often would have higher save percentages?