Martin Brodeur

[haakon84]

Looking at only save % really hurts Martin Brodeur. In fact, I'm working on a study that shows that shots per game faced effects save % tremendously. Moreso than percentage of PP's faced. It makes sense when you think about it. Goalies save %'s tend to go up as they face above 30 shots in a game.

I have been accumulating game logs from http://hockeygoalies.org and categorizing each goaltenders shot into how many shots that goaltender faced in that game. I then divide the amount of shots faced in each category with the total amount of shots and get a percentage of the shots a goaltender faced. I also take the total saves in each category and divide it by total shots to get a save %.

For example if we take Martin Brodeur's mediocre '01-'02 season and look at the percentage of shots he faced in each category we get...

Shots per game | % | Career Save %
above 40 | 2.6% | .937
36-40 | 4.5% | .933
30-35 | 7.6% | .927
25-29 | 28.5% | .918
20-24 | 39.2% | .908
Under 20 | 17.3% | .873

According to his career save % his save % for that season should have been .908.

In fact it was .906.

Now let's apply it to Hasek's career save % numbers...

Shots per game | % | Career Save %
above 40 | 2.6% | .942
36-40 | 4.5% | .936
30-35 | 7.6% | .931
25-29 | 28.5% | .914
20-24 | 39.2% | .921
Under 20 | 17.3% | .876

Instead of getting Hasek's marvelous .922 career save %. He would have had a save% of .913.

So how about if we apply it to Brodeur's numbers with Hasek's share in '97-'98 where he posted an unbelievable .932.

Shots per game | % | Career Save %
above 40 | 10.1% | .937
36-40 | 26.5% | .933
30-35 | 32.2% | .927
25-29 | 17.2% | .918
20-24 | 8.65% | .908
Under 20 | 5.28% | .873

In this scenario we can expect Brodeur to have a save % of .924. Which by no means is .932 but it is an increase.

I plan on taking the game logs from the top 10 save % goalies post '93 and applying this formula to all goalies for a better comparison. Of the goalies I have done this for (Luongo, Roy, Brodeur, Belfour, and Hasek) EVERY goaltender has a much lower save % than their career save % when faced with Brodeur's shots per game percentage. Coincidentally every goaltender has a much HIGHER save % with Luongo's shots per game percentage.

One thing remains true, Hasek is the save % king and there is no disputing that but with this study Brodeur is not very far behind (a -.03 differential) on average. What is interesting is the ridiculous increase in Brodeur's save % when presented with a workload like Luongo.

 

[TheDevilMadeMe]

Looking at only save % really hurts Martin Brodeur. In fact, I'm working on a study that shows that shots per game faced effects save % tremendously. Moreso than percentage of PP's faced. It makes sense when you think about it. Goalies save %'s tend to go up as they face above 30 shots in a game.

I have been accumulating game logs from http://hockeygoalies.org and categorizing each goaltenders shot into how many shots that goaltender faced in that game. I then divide the amount of shots faced in each category with the total amount of shots and get a percentage of the shots a goaltender faced. I also take the total saves in each category and divide it by total shots to get a save %.

For example if we take Martin Brodeur's mediocre '01-'02 season and look at the percentage of shots he faced in each category we get...

Shots per game | % | Career Save %
above 40 | 2.6% | .937
36-40 | 4.5% | .933
30-35 | 7.6% | .927
25-29 | 28.5% | .918
20-24 | 39.2% | .908
Under 20 | 17.3% | .873

According to his career save % his save % for that season should have been .908.

In fact it was .906.

Now let's apply it to Hasek's career save % numbers...

Shots per game | % | Career Save %
above 40 | 2.6% | .942
36-40 | 4.5% | .936
30-35 | 7.6% | .931
25-29 | 28.5% | .914
20-24 | 39.2% | .921
Under 20 | 17.3% | .876

Instead of getting Hasek's marvelous .922 career save %. He would have had a save% of .913.

So how about if we apply it to Brodeur's numbers with Hasek's share in '97-'98 where he posted an unbelievable .932.

Shots per game | % | Career Save %
above 40 | 10.1% | .937
36-40 | 26.5% | .933
30-35 | 32.2% | .927
25-29 | 17.2% | .918
20-24 | 8.65% | .908
Under 20 | 5.28% | .873

In this scenario we can expect Brodeur to have a save % of .924. Which by no means is .932 but it is an increase.

I plan on taking the game logs from the top 10 save % goalies post '93 and applying this formula to all goalies for a better comparison. Of the goalies I have done this for (Luongo, Roy, Brodeur, Belfour, and Hasek) EVERY goaltender has a much lower save % than their career save % when faced with Brodeur's shots per game percentage. Coincidentally every goaltender has a much HIGHER save % with Luongo's shots per game percentage.

One thing remains true, Hasek is the save % king and there is no disputing that but with this study Brodeur is not very far behind (a -.03 differential) on average. What is interesting is the ridiculous increase in Brodeur's save % when presented with a workload like Luongo.

Interesting. I know there was a study that found no correlation between shots faced and save % across the league, but I still suspected that if you looked just at the top goalies in the league, you would find such a correlation. Kudos for taking the time to do this.

Two possible causes:

1) It's easier to stay focused when facing more shots. Ken Dryden, in particular talked about this.

2) It's largely a product of what the shot counter actually calls a shot (obviously all goals will be shots, so it's the times that it isn't a goal where there is sometimes a question).

If #1, I'd expect that most of the top goalies will show a similar effect.

If #2, I'd expect guys who played at home arenas that underreported shots (like Brodeur) and guys who played at home arenas that overreported shots to have the greatest difference. Actually, it would be interesting to compare Luongo's Florida days (notorious for inflating SOG) with his Vancouver days to see if he shows a similar effect.

 

[overpass]

Looking at only save % really hurts Martin Brodeur. In fact, I'm working on a study that shows that shots per game faced effects save % tremendously. Moreso than percentage of PP's faced. It makes sense when you think about it. Goalies save %'s tend to go up as they face above 30 shots in a game.

I have been accumulating game logs from http://hockeygoalies.org and categorizing each goaltenders shot into how many shots that goaltender faced in that game. I then divide the amount of shots faced in each category with the total amount of shots and get a percentage of the shots a goaltender faced. I also take the total saves in each category and divide it by total shots to get a save %.

For example if we take Martin Brodeur's mediocre '01-'02 season and look at the percentage of shots he faced in each category we get...

Shots per game | % | Career Save %
above 40 | 2.6% | .937
36-40 | 4.5% | .933
30-35 | 7.6% | .927
25-29 | 28.5% | .918
20-24 | 39.2% | .908
Under 20 | 17.3% | .873

According to his career save % his save % for that season should have been .908.

In fact it was .906.

Now let's apply it to Hasek's career save % numbers...

Shots per game | % | Career Save %
above 40 | 2.6% | .942
36-40 | 4.5% | .936
30-35 | 7.6% | .931
25-29 | 28.5% | .914
20-24 | 39.2% | .921
Under 20 | 17.3% | .876

Instead of getting Hasek's marvelous .922 career save %. He would have had a save% of .913.

So how about if we apply it to Brodeur's numbers with Hasek's share in '97-'98 where he posted an unbelievable .932.

Shots per game | % | Career Save %
above 40 | 10.1% | .937
36-40 | 26.5% | .933
30-35 | 32.2% | .927
25-29 | 17.2% | .918
20-24 | 8.65% | .908
Under 20 | 5.28% | .873

In this scenario we can expect Brodeur to have a save % of .924. Which by no means is .932 but it is an increase.

I plan on taking the game logs from the top 10 save % goalies post '93 and applying this formula to all goalies for a better comparison. Of the goalies I have done this for (Luongo, Roy, Brodeur, Belfour, and Hasek) EVERY goaltender has a much lower save % than their career save % when faced with Brodeur's shots per game percentage. Coincidentally every goaltender has a much HIGHER save % with Luongo's shots per game percentage.

One thing remains true, Hasek is the save % king and there is no disputing that but with this study Brodeur is not very far behind (a -.03 differential) on average. What is interesting is the ridiculous increase in Brodeur's save % when presented with a workload like Luongo.

Good work compiling the data and presenting it. However, I'm not sure these results are as conclusive as you might think.

First, did you remove games in which Brodeur/Hasek were pulled? These will bias the results, as shots will be artificially low in games that have a low save percentage.

I suspect at least some of the effect you have found is a result of playing to the score. Teams that are winning, especially in the 3rd period, tend to get outshot but have higher shooting percentages and save percentages, as the losing team fires shots from outside but gives up higher quality chances against as they press to score.

This has two implications for your results:

1. Games in which a team's goalie lets in goals early will tend to have fewer shots against that team.

2. Those fewer shots taken against the trailing team will tend to be higher quality chances.

Both of these effects cause lower-shot games to have lower save percentages, and higher shot games to have higher save percentages.

I actually noticed this effect some time ago, and recently compiled similar numbers to yours. In order to correct for the "playing to the score" effect, I compiled the first period numbers for all goalies over the last three years. (These numbers are by team, not by goalie, so it doesn't matter if the goalie was pulled.)

Shots/Per| TotalShots|TotalGA|SV%
0-6 | 7190 | 723 | 0.899
7-8 | 11698 | 1097 | 0.906
9-10 | 16233 | 1408 | 0.913
11-12 | 14838 | 1300 | 0.912
13-14 | 10285 | 859 | 0.916
15-16 | 5633 | 440 | 0.922
17+ | 4431 | 311 | 0.930

It appears that save percentages still tend to be higher when facing more shots, even after looking at the first period only.

What should we do with this information? TheDevilMadeMe suggested that poor shot counting may be a culprit. This is certainly part of the effect, as when you select a group of low-shot games you are actually selecting games with few shots against AND games where shots were under-counted. We would expect to see a lower save percentage in games in which shots were under-counted. The opposite is true for high-shot games - they will include a higher % of over-counted shot games, where we would expect to see a higher save percentage. However, we shouldn't correct for this effect. All goalies will have some games in which shots are over-counted and some games in which shots are under-counted. If there is a systemic bias by arena in shot counting, as recent research suggests, a simple correction for shot counting bias by arena is the appropriate remedy,

If it is inherently easier to have a high save percentage when facing more shots - due to Dryden's hypothesis or something else - then we should indeed consider this when evaluating goalies. However, if this is the case, why doesn't a correlation between shots against and save percentage show up in the seasonal numbers, as we would expect?*

*It's possible that it shows up in the 1st period only seasonal numbers, but I doubt it and I haven't checked it. It most certainly doesn't show up in the overall numbers, and never has in any era. If anything, it's been harder to have a high save percentage with high shots against in much of NHL history.

 

[seventieslord]

Dryden was a guy who faced very low shot totals but often led the league in sv%. If this data holds water (which I'm not sure it does) then Dryden was definitely better than Brodeur (by virtue of being able to have an elite sv% despite that oxmoronic disadvantage) but, of course, for a career half as long.

 

[Master_Of_Districts]

Interesting. I know there was a study that found no correlation between shots faced and save % across the league, but I still suspected that if you looked just at the top goalies in the league, you would find such a correlation. Kudos for taking the time to do this.

Two possible causes:

1) It's easier to stay focused when facing more shots. Ken Dryden, in particular talked about this.

2) It's largely a product of what the shot counter actually calls a shot (obviously all goals will be shots, so it's the times that it isn't a goal where there is sometimes a question).

If #1, I'd expect that most of the top goalies will show a similar effect.

If #2, I'd expect guys who played at home arenas that underreported shots (like Brodeur) and guys who played at home arenas that overreported shots to have the greatest difference. Actually, it would be interesting to compare Luongo's Florida days (notorious for inflating SOG) with his Vancouver days to see if he shows a similar effect.

You've made this assertion before, but have consistently failed to provide any supporting data.

I've ran the numbers, and there's absolutely no evidence that shots are being overcounted in Florida.

It's true that Panther home games have tended to feature more shots than Panther road games over the last several seasons.

However, this in and of itself does not demonstrate the existence of an overcounting bias. For example, it's conceivable that the Panthers play (and have played) a more entertaining style of hockey at home as compared to on the road.

One way to test your hypothesis is to compare the shooting percentage in Florida road games to the shooting percentage in Florida home games (for both teams, that is). If shots are indeed being overcounted in Sunrise, then one would expect to see a lower shooting percentage in Panther home games relative to Panther road games.

Looking at data from the last five seasons (03-04 to 08-09), there is no difference between the shooting percentage in Panther road games and the shooting percentage in Panther home games.

 

[haakon84]

Interesting. I know there was a study that found no correlation between shots faced and save % across the league, but I still suspected that if you looked just at the top goalies in the league, you would find such a correlation. Kudos for taking the time to do this.

See I find a flaw in that study because it is based on overall shots per game. When you categorize it like I have, it is extremely apparent that shots per game has an effect on save %.

What led me to do this work is say goaltender A faces 120 shots over 3 games and allows 3 goals per game.

111 saves on 120 shots = .925

Goaltender B faces 120 shots over 6 games and allows 2 goals per game

108 saves on 120 shots = .900

When we just look at save % goaltender A has a huge advantage although he let in more goals per game. Goaltender B has a huge advantage when it comes to GAA. My thinking is that a goalies save % is based on a game to game shot total not a yearly shots per game total and the data I'm presenting is starting to support that.

I also feel that elite goaltenders on average will allow 2-2.5 goals per game regardless of shots faced and my data supports this. When you have an elite defensive team like NJ they are still going to allow those 1-2 shots that will get through and are unstoppable and it may help a goaltenders GAA but hurt his save % and again my data supports this. The goaltender on my list whose save % with Brodeur's SPG share is closest to his career save % is Belfour. I think that has to do with him being on elite defensive teams as well.

Good work compiling the data and presenting it. However, I'm not sure these results are as conclusive as you might think.

Thanks of course you're not you are one of Hasek's biggest supporters . I'm not entirely sure either but the data will hold the truth.

First, did you remove games in which Brodeur/Hasek were pulled? These will bias the results, as shots will be artificially low in games that have a low save percentage.

I did not and that is the next step in my data. Also, I did not remove the games in which Hasek came in to relieve a goaltender and stopped say 14 out of 14 shots. I still believe that facing under 20 shots a goaltender will have a low save %. In fact I just took the games where Hasek was pulled out and his save % is .884 when facing below 20 shots. Broder's is actually .893 so taking out games where a goaltender is pulled helps Brodeur. I expect that to be true for every goalie but Luongo (who has a disproportionate amount of games where he was pulled under 20 shots to games finished under 20 shots.

I suspect at least some of the effect you have found is a result of playing to the score. Teams that are winning, especially in the 3rd period, tend to get outshot but have higher shooting percentages and save percentages, as the losing team fires shots from outside but gives up higher quality chances against as they press to score.

This may be true for every goaltender BUT Martin Brodeur. They had the opposite effect on the opposing team when they were ahead. So again this still supports Marty's case.

It appears that save percentages still tend to be higher when facing more shots, even after looking at the first period only.

What should we do with this information? TheDevilMadeMe suggested that poor shot counting may be a culprit. This is certainly part of the effect, as when you select a group of low-shot games you are actually selecting games with few shots against AND games where shots were under-counted. We would expect to see a lower save percentage in games in which shots were under-counted. The opposite is true for high-shot games - they will include a higher % of over-counted shot games, where we would expect to see a higher save percentage. However, we shouldn't correct for this effect. All goalies will have some games in which shots are over-counted and some games in which shots are under-counted. If there is a systemic bias by arena in shot counting, as recent research suggests, a simple correction for shot counting bias by arena is the appropriate remedy,

If it is inherently easier to have a high save percentage when facing more shots - due to Dryden's hypothesis or something else - then we should indeed consider this when evaluating goalies. However, if this is the case, why doesn't a correlation between shots against and save percentage show up in the seasonal numbers, as we would expect?*

*It's possible that it shows up in the 1st period only seasonal numbers, but I doubt it and I haven't checked it. It most certainly doesn't show up in the overall numbers, and never has in any era. If anything, it's been harder to have a high save percentage with high shots against in much of NHL history.

As I posted earlier in this message I don't think you'll find a correlation when you take overall shots per game with save % over every goaltender. But when you categorize it to shots per game with elite goaltenders there is no denying that they have a higher save % when they face more shots in a single game. The only goaltender who doesn't follow this trend is Hasek whose 20-25 shots save % is unusually high and his 25-30 save % is rather mediocre.

Just as a point of reference to show why I'm excited about this data. Let's take Luongo's '03-'04 season where he posted a ridiculous .931 save % and apply his Shots per Game Shares with Hasek and Brodeur:

Brodeur

Shots per game | % | Career Save %
above 40 | 34% | .937
36-40 | 25% | .933
30-35 | 18% | .927
25-29 | 16% | .918
20-24 | 6% | .908
Under 20 | 0 | .873
Adjusted save % | .929

Hasek

Shots per game | % | Career Save %
above 40 | 34% | .941
36-40 | 25% | .935
30-35 | 18% | .931
25-29 | 16% | .914
20-24 | 6% | .921
Under 20 | 0 | .876
Adjusted save % | .932

Notice the percentages. Almost a reverse of Brodeur's '01-'02 SPG shares. Now their MUST be something to this data.

 

[haakon84]

Dryden was a guy who faced very low shot totals but often led the league in sv%. If this data holds water (which I'm not sure it does) then Dryden was definitely better than Brodeur (by virtue of being able to have an elite sv% despite that oxmoronic disadvantage) but, of course, for a career half as long.

Going by this logic it would also mean Brodeur deserved the '96-'97 Vezina.

 

[seventieslord]

Also, I did not remove the games in which Hasek came in to relieve a goaltender and stopped say 14 out of 14 shots.

I believe TCG showed that goalies coming in after the starter being pulled tend to be rusty from half a game on the bench and/or facing a hot team, and still have low sv% stats. Can't recall for sure if it was him though.

Going by this logic it would also mean Brodeur deserved the '96-'97 Vezina.

Right, although I am not going by this data, just thinking out loud/playing Devil's Advocate, no pun intended.

 

[haakon84]

I believe TCG showed that goalies coming in after the starter being pulled tend to be rusty from half a game on the bench and/or facing a hot team, and still have low sv% stats. Can't recall for sure if it was him though.

Regardless I took the games under 40 minutes out and Hasek's save % is still well below his other save %'s at .884. Meanwhile Brodeur's is at .893.

 

[Dennis Bonvie]

Could it be that most of Hasek's low shot total games were when he was way over the hill in his Detroit days, thus making his SP much worse for that category?

 

[haakon84]

Could it be that most of Hasek's low shot total games were when he was way over the hill in his Detroit days, thus making his SP much worse for that category?

Hasek on Detroit = age 37.
Brodeur now = age 37.
Roy at a career year = age 37.

And he had just won a Vezina! Myth - busted.

 

[overpass]

Hasek on Detroit = age 37.
Brodeur now = age 37.
Roy at a career year = age 37.

And he had just won a Vezina! Myth - busted.

There's no doubt Hasek was past his prime by his two final years in Detroit - he was 42 and 43. Those Detroit teams did an excellent job of preventing shots, so you may indeed have some slight bias in there where a higher % of Hasek's low-shot games come when he's past his prime. I don't think it's the entire reason for the effect you found, or even close. But if you are going to use the career stats for Hasek and Brodeur, there is a bit of bias there.

One more thing - TCG addressed the issue of save percentages in low-shot games here, I think it's worth reading.

 

[Hawkey Town 18]

Hasek on Detroit = age 37.
Brodeur now = age 37.
Roy at a career year = age 37.

And he had just won a Vezina! Myth - busted.

Way to take things out of context, that was Hasek's first year with the Wings, he had 111 more games with them after that...97 of which were at 41+ years old.

 

[TheDevilMadeMe]

Could it be that most of Hasek's low shot total games were when he was way over the hill in his Detroit days, thus making his SP much worse for that category?

You're correct. And there could be a pro-Brodeur bias. as he hit his prime (or at least got more consistent) around 2002 or 2003, and then started facing far more shots after the lockout.

A season-by-season basis would be more revealing than a career-based analysis for these reasons.

 

[TheDevilMadeMe]

You've made this assertion before, but have consistently failed to provide any supporting data.

I've ran the numbers, and there's absolutely no evidence that shots are being overcounted in Florida.

It's true that Panther home games have tended to feature more shots than Panther road games over the last several seasons.

However, this in and of itself does not demonstrate the existence of an overcounting bias. For example, it's conceivable that the Panthers play (and have played) a more entertaining style of hockey at home as compared to on the road.

One way to test your hypothesis is to compare the shooting percentage in Florida road games to the shooting percentage in Florida home games (for both teams, that is). If shots are indeed being overcounted in Sunrise, then one would expect to see a lower shooting percentage in Panther home games relative to Panther road games.

Looking at data from the last five seasons (03-04 to 08-09), there is no difference between the shooting percentage in Panther road games and the shooting percentage in Panther home games.

It's very possible that the sunbelt teams play a "more entertaining" style at home to bring in the fans. Still though, I always found it fishy how every regular Florida goalie, including the backups (especially the backups), always seem to have a stellar save %, compared to goalies on other bad teams. And I'm obviously not the only one, as Florida was one of the main reasons puck prospectus did that study.

But what does that say about teams like NJ and St. Louis which face far fewer shots at home? Do they try extra hard to entertain the road fans? That doesn't make sense.

The puck prospectus study did take shooting % into account on a league-wide basis, but didn't isolate individual teams like you did with Florida.

 

[reckoning]

Yeah. This study might be better used by just comparing seasons rather than career totals, as any percentage or per game stat that is based on career totals will always hurt players who play into their 40s.

I'm also a little uncomfortable with the arbitrary shots per game categories. A 29 shot game is in the same category as a 25 shot game but not the same category as a 30 shot game, even though 29 is closer to 30 than 25.

Not trying to nitpick as this is an interesting subject to research.

 

[Master_Of_Districts]

It's very possible that the sunbelt teams play a "more entertaining" style at home to bring in the fans. Still though, I always found it fishy how every regular Florida goalie, including the backups (especially the backups), always seem to have a stellar save %, compared to goalies on other bad teams. And I'm obviously not the only one, as Florida was one of the main reasons puck prospectus did that study.

I'm not sure if that's entirely true.

I've pulled up a list on hockey reference of goaltenders that played at least one game for Florida from 2003-04 to present.

While Luongo, Vokoun and Anderson have all excelled, all three have also posted impressive save percentage numbers elsewhere (in Vancouver, Nashville and Colorado, respectively).

McLennan, Belfour, Auld, Clemmensen and Shields all have or had average or below average numbers in terms of save percentage.

But what does that say about teams like NJ and St. Louis which face far fewer shots at home? Do they try extra hard to entertain the road fans? That doesn't make sense.

The puck prospectus study did take shooting % into account on a league-wide basis, but didn't isolate individual teams like you did with Florida.

I've ran the numbers for the Devils as well. Unlike with Florida, it appears that there is genuine recording bias in New Jersey, albeit in the opposite direction (undercounting as opposed to overcounting). The shooting percentage in Devils home games has been consistently higher than the shooting percentage in Devils road games.

So when Devils fans claim that Brodeur's save percentage has historically been prejudiced by the Devils shot recorder, their complaint is a legitimate one.

 

[Hockey Outsider]

For example if we take Martin Brodeur's mediocre '01-'02 season and look at the percentage of shots he faced in each category we get...

Shots per game | % | Career Save %
above 40 | 2.6% | .937
36-40 | 4.5% | .933
30-35 | 7.6% | .927
25-29 | 28.5% | .918
20-24 | 39.2% | .908
Under 20 | 17.3% | .873

Sample size is a concern here. 2.6% of Brodeur's starts is 2 games, right? Later on you're extrapolating that to 34 games (to mirror Luongo's 2004 season). This work is interesting, but sample size is a big concern if you're only using single-season stats. Not trying to misquote you, so correct me if I'm wrong.

The other major issue (as some have already pointed out) is confusing correlation with causation. Some people think that facing fewer shots causes a lower save percentage. I think the relationship works the other way: a lower save percentage causes fewer shots. A goalie having a rough night (say 3-4 goals in the first period, which would almost result in a save percentage below around 85%) will get pulled, so he'll probably face less than 20 shots in the game. It was the goalie's poor play (low save percentage) that caused him to face fewer shots, not the other way around. I'm curious to see what impact this has on the data. It would be useful, I think, to re-run the numbers if you filter it so that all goalies playing less than, say, 50 minutes in a game are removed.

 

[Canadiens1958]

Suggestion

How many times a starting goalie gets pulled is a significant stand alone stat. Certainly would impact on how goalies are evaluated.

As for sv% it could be measured for a goalies complete game. For games that he started and was pulled or for games where he replaced the starting goalie.

Such segregation would reflect the game after game consistency, the range between his good and bad games while looking at his ability to step in cold.

SV% has the potential to explain these qualities in various goalies.

 

[Bear of Bad News]

How many times a starting goalie gets pulled is a significant stand alone stat. Certainly would impact on how goalies are evaluated.

As for sv% it could be measured for a goalies complete game. For games that he started and was pulled or for games where he replaced the starting goalie.

Both of these measures, while interesting, would be heavily influced by the goaltender's head coach. Ask any netminder who played for Mike Keenan.

 

[Canadiens1958]

Goalie / Coach Relationship

Both of these measures, while interesting, would be heavily influced by the goaltender's head coach. Ask any netminder who played for Mike Keenan.

So we get a measure of the goalie/coach relationship. Someone could measure how effective Mike Keenan's tactics were.

Coaches play goalies based on performance against certain teams and styles.Teams breakdown stats to suite their specific needs. Nothing wrong with applying the same approach globally.

 

[God Bless Canada]

There certainly can be a case made for Brodeur as the best goalie ever. I'm not a fan of using career stats as the primary basis for an argument. Too many holes. Too tilted towards those who played at a time when they play more games. (I think baseball is the only major team sport in North America where career stats work, because they've had close to the same length of schedule for most of the game's history. There's also a relatively level playing field in baseball for most of this century).

But a lot of people like career stats. Brodeur's No. 1 all-time in the two big ones: wins and shutouts. (He'll also likely be No. 1 for losses by the time he retires, but that's another discussion for another time).

It's a reflection of a lot of things. He's incredible. He's durable. He's consistent. When it comes to goaltenders, one of the first things I look for is reliability. From 1997-98 to 2007-08, he played at least 70 games every year. That's incredible. I don't know if that'll ever be topped. He won at least 38 games every year.

It's also reflective of playing at a time when they had an 82 game season, goalies are expected to play 65-70 games a season, and advances have allowed players to play at a higher level for long than in the past. That's why it's not a coincidence that the top four goaltenders in career wins started their careers in the last 25 years.

And he has three rings. As a goalie, you're paid to win. You're judged on wins. You judge yourself on wins. Not on goals against average or save percentage.

I look at other accomplishments. How do they compare with Jacques Plantes record six Stanley Cup wins? That's the one that means the most to me for a goalie. Or Glenn Hall's seven first-team all-star selections? The myriad of accomplishments for Terry Sawchuk? Hasek's back-to-back Hart Trophies? Or the other one that means the most to me: Patrick Roy's three Conn Smythe Trophies?

If anything, I'm probably more impressed that Brodeur had 10 straight seasons with 70 games and 38 wins. Or that he led the league in wins eight times in nine years. If you're a hockey person, that's incredible. When a team's brass can sit down at the start of the season and say "We can count on our goalie to play 70 games and get 38 wins, probably more than 40 wins," it's the greatest feeling in the world. You know that you're going to be in the playoffs, and likely going to have home ice advantage for at least the first round.

Has he benefitted from playing behind those great New Jersey teams? Sure. But those great New Jersey teams have benefitted more by having Brodeur behind them. They know that he has the right mental make-up to handle only 20-25 shots a game. They know that they only need two, some nights three goals, because Brodeur will allow only one or two goals. They know that if they're down a goal in the third period, that they can press for the equalizer, because they know that Brodeur will make the big save on an odd man rush to keep it a one-goal game. And they know that if they do have an off night, that Brodeur can steal a game.

I don't know if he's the greatest goalie ever. I don't think he is. I think Plante is. Roy's close. If you don't emphasize reliability like I do, you might have Hasek at No. 1.

But as far as consistency and reliability are concerned, I don't think anybody beats Brodeur. He is a freak. And he's probably the best puck-moving goalie of all-time, too, and while that doesn't mean a lot to most fans, to his defencemen, his coaches and his GM, it's a very big deal.

 

[haakon84]

This is my last post until I compile more data.

I have taken games where the goalies have played under 40 minutes. I feel playing 2/3rds of a game should round up to a game (there are very few of these cases). And surprisingly it makes Brodeur's case stronger. He has the biggest increase in save percentage and loses the least in percentage of shots faced under 20.

Sample size is a concern here. 2.6% of Brodeur's starts is 2 games, right? Later on you're extrapolating that to 34 games (to mirror Luongo's 2004 season). This work is interesting, but sample size is a big concern if you're only using single-season stats. Not trying to misquote you, so correct me if I'm wrong.

Yes that is a seasonal sample size and would be a the correct percentage of games. Luongo faced an absurd amount of games above 40 shots that year. I would be more concerned with sample size if the career numbers varied. On 10,000 shots faced on the 6 netminders I mentioned earlier the average save % is .939. Each goaltender is very close to this mark. Luongo has benefitted from this the most, he has already faced more games with above 40 shots than Brodeur, Roy, Hasek, and Belfour and that includes playoff games! Brodeur has faced the fewest by a good margin.

The other major issue (as some have already pointed out) is confusing correlation with causation. Some people think that facing fewer shots causes a lower save percentage. I think the relationship works the other way: a lower save percentage causes fewer shots. A goalie having a rough night (say 3-4 goals in the first period, which would almost result in a save percentage below around 85%) will get pulled, so he'll probably face less than 20 shots in the game. It was the goalie's poor play (low save percentage) that caused him to face fewer shots, not the other way around. I'm curious to see what impact this has on the data. It would be useful, I think, to re-run the numbers if you filter it so that all goalies playing less than, say, 50 minutes in a game are removed.

As I said earlier I set the mark at 40 games and it had the biggest impact on Luongo and that is probably because he has only had 15 games where he's faced fewer than 20 shots. The average save % under 20 shots over 8500 shots faced is .882.

I'll make all this data public once I clean it up.

There's no doubt Hasek was past his prime by his two final years in Detroit - he was 42 and 43. Those Detroit teams did an excellent job of preventing shots, so you may indeed have some slight bias in there where a higher % of Hasek's low-shot games come when he's past his prime. I don't think it's the entire reason for the effect you found, or even close. But if you are going to use the career stats for Hasek and Brodeur, there is a bit of bias there.

I challenge this notion with my theory that his save % was effected because he was in more games in the under 30 shots per game category on those defensive teams.I have taken Haseks post-Buffalo stats and Buffalo stats and applied the same formula above.

Post-Buffalo | |
Shots per game | SPG % | Save %
above 40 | 1% | .921
36-40 | 13% | .943
30-35 | 16% | .920
25-29 | 29% | .909
20-24 | 27% | .922
Under 20 | 14% | .905
Total | | .918

*Note that this is only 2 games (out of 257) and the sample here is very small if Hasek had one more game above 40 shots where he let in 3 goals his save % would jump to .929

Buffalo | |
Shots per game | SPG % | Save %
above 40 | 13% | .943
36-40 | 19% | .933
30-35 | 31% | .933
25-29 | 21% | .916
20-24 | 11% | .922
Under 20 | 4% | .856
Total | | .926

If we apply his Post-Buffalo SPG % with his career save % totals we can expect him to have a .917 save %. With his Buffalo days he would have a .926 save %.

If we apply his Buffalo SPG % to Brodeur's SPG save % the adjusted save % is .924.

Notice that in his Buffalo days 63% of his games had above 30 shots. The career totals that is closest to is Luongo where it was 67% of his games. Combine that with the dead-puck era and lower scoring and I think there is reason to believe that more shots faced provides an inflated save %.

If you look at his post-Buffalo days where he played on good teams. 70% of his games had under 30 shots faced. The closest career totals that comes to is Martin Brodeur who is at 65%.

One more thing - TCG addressed the issue of save percentages in low-shot games here, I think it's worth reading.

Thanks, I have read this study and I'm not sure he hasn't done more with it. If that is the case then he should really change his mind on Martin Brodeur.

He has been in more games with 20 shots or below (post '93) than Hasek, Roy, and Luongo COMBINED in HALF as many games. Nearly double Hasek, 4 times Roy's amount, and 13 times Luongo's amount.

 

[seventieslord]

He has been in more games with 20 shots or below (post '93) than Hasek, Roy, and Luongo COMBINED in HALF as many games. Nearly double Hasek, 4 times Roy's amount, and 13 times Luongo's amount.

What a backwards thread this has turned into, talking about low-shot games as though it's a disadvantage for a goalie and not an advantage.

This data is all for naught unless you can show that lower shots in a game causes lower sv% and not the other way around, which is much more compelling and logical to me at this point.

 

[haakon84]

What a backwards thread this has turned into, talking about low-shot games as though it's a disadvantage for a goalie and not an advantage.

When it comes to save % it is.

This data is all for naught unless you can show that lower shots in a game causes lower sv% and not the other way around, which is much more compelling and logical to me at this point.

That is what it shows. Where in my data does fewer shots cause a higher save %?

I'm baffled by these remarks. You're so easy to dismiss data but never really supply any of your own. Who are you to decide what is all for naught? How many people have said this is interesting?