Adjusted Save Percentage

[seventieslord]

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.

No. I am interested not in comparing Grant Fuhr's save percentage to Henrik Lundqvist's. I'm interested in knowing how Grant Fuhr dominated his peers compared to Lundqvist or anyone else. Raw comparisons of numbers clearly do not work and everyone on the HOH board knows this. Any advantages or disadvantages due to era (such as pads, lack of helmets, composite sticks, forward passing, expansion, coaching, etc) applied to all goaltenders equally. the best ones rose to the top in all eras. So when discussing Fuhr and his small pads, the short answer is: All other goalies he played against had small pads too.

 

[haakon84]

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?

I don't understand this... I explained what I did. I put the sorted out the total power play shots against and then ran the averages for each criteria. I did not total things up by goalie. No I'm not surprised goalies who play more have higher save %'s what I am surprised with is that as power play shots goes up so does the goalies PK save % along with his even strength save % and total save %. That is pretty much the opposite of what seventieslord is claiming to be true. I just ran the data with only good goalies, but this site is acting very bizarre and would not let me reply. I will tell you that it does not matter if it only consists of elite goaltenders - the results and correlations are the same. I will post it but I HATE redoing things.

 

[Bear of Bad News]

I don't understand this... I explained what I did. I put the sorted out the total power play shots against and then ran the averages for each criteria. I did not total things up by goalie. No I'm not surprised goalies who play more have higher save %'s what I am surprised with is that as power play shots goes up so does the goalies PK save % along with his even strength save % and total save %.

Without looking at it on a per-game basis, all you're getting is this:

Power Play Shots are correlated with Minutes Played, which are correlated with Goaltender Quality, which is correlated with ES SV% and SH SV%.

In other words, better goaltenders play more.

 

[haakon84]

Goaltenders: Vokoun, Thomas, Miller, Luongo, Dipietro, Roy, Kolzig, Khabibulin, Kiprusoff, Backstrom, Turco, Brodeur, Giguere, Lundqvists, Nabokov, Belfour, Roloson, Hasek, Joseph, Osgood

PP Shots | PK save% | ES save % | total save %
<500 | 88.75% | 93.08% | 91.73%
400-500 | 87.83% | 92.34% | 91.27%
300-400 | 87.58% | 92.22% | 91.25%
200-300 | 87.15% | 92.07% | 91.17%
100-200 | 86.44% | 91.59% | 90.62%

 

[Bear of Bad News]

Time period?

It's as if you are deliberately obfuscating your results so that no one else can check what you did. Be proud of your analysis!

 

[haakon84]

Time period?

It's as if you are deliberately obfuscating your results so that no one else can check what you did. Be proud of your analysis!

Seriously do you have to be so condesending?

It is from the only period we have PP shot data for after the '98 season.

 

[haakon84]

Ok so lets look at Luongo...

PP Shots | PK save% | ES save % | total save % | PP shots per game
<500 | 90.17% | 93.03% | 92.20% | 7.66
400-500 | 88.50% | 92.70% | 91.75% | 6.37
300-400 | 86.50% | 93.20% | 91.75% | 6.14

 

[seventieslord]

Goaltenders: Vokoun, Thomas, Miller, Luongo, Dipietro, Roy, Kolzig, Khabibulin, Kiprusoff, Backstrom, Turco, Brodeur, Giguere, Lundqvists, Nabokov, Belfour, Roloson, Hasek, Joseph, Osgood

PP Shots | PK save% | ES save % | total save %
<500 | 88.75% | 93.08% | 91.73%
400-500 | 87.83% | 92.34% | 91.27%
300-400 | 87.58% | 92.22% | 91.25%
200-300 | 87.15% | 92.07% | 91.17%
100-200 | 86.44% | 91.59% | 90.62%

I don't mean to be difficult (though sometimes I think you do), but I am still not seeing what you think this is showing. Goalies who play more often will see more shots. 12-25% of those will be PP shots. if a goalie plays enough games they will see 100+ PP shots and thus be included in the study. The better they are, the more likely that they will be counted on for more games, and the more likely will climb up higher in the "PP shots" bracket. These goalies all have a higher sv% at both ES and on the PP when they play more. They get to play more because they are better goalies!

This completely ignores the percentage of shots they faced on the PP. As you can see, all these goalies were about 5% higher at ES compared to the PP. Facing a higher percentage of PP shots would lead to a sv% decline. The raw number of PP shots they faced tells us nothing. A goalie who faced 500 PP shots also faced a high number of ES shots; the percentage of shots that he faced that resulted from PPs could actually be low.

You would need to run these figures using a percentage model, and instead of using raw sv% figures I think your best bet would be to compare the sv% to the league average. (the situational averages would be easy to calculate if you have copied that webpage into a spreadsheet; then just sum up the total ES shots and total ES goals, do the same on the PP)

 

[Bear of Bad News]

Seriously do you have to be so condesending?

It is from the only period we have PP shot data for after the '98 season.

People on the History board are used to a high level of discourse, and are also used to rigorous questioning of techniques and methodology. If you find this condescending, I cannot help you with that.

 

[seventieslord]

By the way, Haakon, what is your position? Are you arguing Brodeur is the best of all-time? Better than just Hasek? Better than just Roy? Not just a product of the system? Top-10? What?

my position is that he's 4th-9th all-time and I find it incredible that so many people have taken to attacking that position lately. Any "undercutting" of Brodeur is done with the intent to show his deficiencies relative to Roys and Haseks, not relative to Vokouns and Nabokovs.

 

[haakon84]

People on the History board are used to a high level of discourse, and are also used to rigorous questioning of techniques and methodology. If you find this condescending, I cannot help you with that.

Thats not what I find condescending, if you can't understand this, I can not help you with that.

 

[Bear of Bad News]

You are still only exhibiting that goaltenders who have higher save percentages are correlated with goaltenders who play more.

Do this on a per-game basis and you'll have something to argue about.

 

[haakon84]

By the way, Haakon, what is your position? Are you arguing Brodeur is the best of all-time? Better than just Hasek? Better than just Roy? Not just a product of the system? Top-10? What?

my position is that he's 4th-9th all-time and I find it incredible that so many people have taken to attacking that position lately. Any "undercutting" of Brodeur is done with the intent to show his deficiencies relative to Roys and Haseks, not relative to Vokouns and Nabokovs.

I'd say he is around 3rd-7th best all-time but I really can't explain for goalies I haven't seen just what I have read about their legends and style of play. He does have a slight chance to move up if he wins a Cup, Vezina, Smythe. I also think that he is going to be performing at a very top level for another handful of years. He is very interesting because people have seemed to sum up his career before it has ended since about 2007.

I think Brodeur is like most goaltenders on this board that he is not as good as the people who overrate him but not as bad as the people who underrate him. I think he gets an unfair shake with things when it comes to save %. I understand you hold Hasek and Roy in high regard and they are the 3 goalies who I have seen the most of throughout my life (Lundqvist and Luongo is quickly gaining on them). I have obviously watched Brodeur the most in my lifetime and I really think he brings much more to the table in intangibles than people realize.

These advanced statistics are way over my head but the little research I have done shows that the thing that changes when a goalie faces more PP shots is the even strength save % not the PK save %.

 

[seventieslord]

I'd say he is around 3rd-7th best all-time but I really can't explain for goalies I haven't seen just what I have read about their legends and style of play. He does have a slight chance to move up if he wins a Cup, Vezina, Smythe. I also think that he is going to be performing at a very top level for another handful of years. He is very interesting because people have seemed to sum up his career before it has ended since about 2007.

I think Brodeur is like most goaltenders on this board that he is not as good as the people who overrate him but not as bad as the people who underrate him. I think he gets an unfair shake with things when it comes to save %. I understand you hold Hasek and Roy in high regard and they are the 3 goalies who I have seen the most of throughout my life (Lundqvist and Luongo is quickly gaining on them). I have obviously watched Brodeur the most in my lifetime and I really think he brings much more to the table in intangibles than people realize.

These advanced statistics are way over my head but the little research I have done shows that the thing that changes when a goalie faces more PP shots is the even strength save % not the PK save %.

But they don't show that. I am sure if they were done properly they would show exactly what I am saying. I wish I had the time but I will be on a trip for a week. If you haven't figured it out by then, I'll whip something up now that I know about that webpage

 

[haakon84]

But they don't show that. I am sure if they were done properly they would show exactly what I am saying. I wish I had the time but I will be on a trip for a week. If you haven't figured it out by then, I'll whip something up now that I know about that webpage

Sounds good, look forward to seeing what you come up with. I'll try and play around but I'm no statician (quite the opposite actually) and see if I can come up with something that illustrates my point.

 

[haakon84]

Ok so I think I figured a few things out. I took out any goaltender who played in less than 55 games and then I got the percentage of shots on the penalty kill. I added up the total shots for penalty kill and the total saves and got the save %. Same for even strength and overall. I still do not see a significant advantage in the data which says that a goaltender that faces more shots on the powerplay necessarily has a lower save %.

I set the over and under to 20%. Here are the results.

% on PK | PK save % | ES save % | total save %
>20% | 87.40% | 92.20% | 91.08% |
<20% | 87.04% | 92.07% | 91.18%

Of course there is a .001 advantage in save % to those who faced less than 20% of their shots on the penalty kill. Nothing revolutionary but still noteworthy.

 

[BM67]

Ok so I think I figured a few things out. I took out any goaltender who played in less than 55 games and then I got the percentage of shots on the penalty kill. I added up the total shots for penalty kill and the total saves and got the save %. Same for even strength and overall. I still do not see a significant advantage in the data which says that a goaltender that faces more shots on the powerplay necessarily has a lower save %.

I set the over and under to 20%. Here are the results.

% on PK | PK save % | ES save % | total save %
>20% | 87.40% | 92.20% | 91.08% |
<20% | 87.04% | 92.07% | 91.18%

Of course there is a .001 advantage in save % to those who faced less than 20% of their shots on the penalty kill. Nothing revolutionary but still noteworthy.

You are missing the point that seventieslord has been trying to make that the goalies facing >20% on the PP have both a higher PP SV% and a higher ES SV%, yet their overall SV% is lower.

 

[seventieslord]

Ok so I think I figured a few things out. I took out any goaltender who played in less than 55 games and then I got the percentage of shots on the penalty kill. I added up the total shots for penalty kill and the total saves and got the save %. Same for even strength and overall. I still do not see a significant advantage in the data which says that a goaltender that faces more shots on the powerplay necessarily has a lower save %.

I set the over and under to 20%. Here are the results.

% on PK | PK save % | ES save % | total save %
>20% | 87.40% | 92.20% | 91.08% |
<20% | 87.04% | 92.07% | 91.18%

Of course there is a .001 advantage in save % to those who faced less than 20% of their shots on the penalty kill. Nothing revolutionary but still noteworthy.

A much better effort, but... yeah, what BM67 said. Did you notice that the goalies on top had better sv% in all situations yet came out below in total? That's exactly what I'm talking about.

There still could be more controls done on this. For one thing, you may want to "normalize" each goalie's situational sv% to account for the differences in scoring levels. (I know these seasons were mostly similar, but they are not the same) - So for example, you may want to go with a baseline average of .925 and .885 (or whatever is close to the average over this period) and for seasons where the average is actually .920 at ES, multiply all results by 1.0054. I'm sure you get what I mean. Also, normalizing the PP% might be a good idea too. I think it will be relatively constant, but in case it's not, you may want to use a baseline (perhaps 20%) as a normalizing point so if you find that in one year for example, you have only 17% of shots coming from the PP, you multiply all goalies' PP% by 1.176.

This way all results will be based on situational save percentages relative to the average and on percentage of shots faces on the PP relative to the average.

Also, for greater depth of data and to show a correllation or lack thereof, you should break this down into 5 categories or even more. Instead of just <20 and >20.

 

[haakon84]

A much better effort, but... yeah, what BM67 said. Did you notice that the goalies on top had better sv% in all situations yet came out below in total? That's exactly what I'm talking about.

There still could be more controls done on this. For one thing, you may want to "normalize" each goalie's situational sv% to account for the differences in scoring levels. (I know these seasons were mostly similar, but they are not the same) - So for example, you may want to go with a baseline average of .925 and .885 (or whatever is close to the average over this period) and for seasons where the average is actually .920 at ES, multiply all results by 1.0054. I'm sure you get what I mean. Also, normalizing the PP% might be a good idea too. I think it will be relatively constant, but in case it's not, you may want to use a baseline (perhaps 20%) as a normalizing point so if you find that in one year for example, you have only 17% of shots coming from the PP, you multiply all goalies' PP% by 1.176.

This way all results will be based on situational save percentages relative to the average and on percentage of shots faces on the PP relative to the average.

Also, for greater depth of data and to show a correllation or lack thereof, you should break this down into 5 categories or even more. Instead of just <20 and >20.

I have a better idea of what you are talking about and starting to realize the errors in my ways. I'll try to normalize the data when I get some more time and post my findings. I can tell you that 2006 was an anomaly an unusually high amount of PP shots.

I did try to sectionalize it by under 15%, 15%-20%, 20%-25%, and over 25%. The categories were too weighted in the 15%-25% category and one goaltenders performance (Theodore) greatly skewed the under 15% and over 25%.

Interesting stuff none the less.

 

[seventieslord]

I have a better idea of what you are talking about and starting to realize the errors in my ways. I'll try to normalize the data when I get some more time and post my findings. I can tell you that 2006 was an anomaly an unusually high amount of PP shots.

I did try to sectionalize it by under 15%, 15%-20%, 20%-25%, and over 25%. The categories were too weighted in the 15%-25% category and one goaltenders performance (Theodore) greatly skewed the under 15% and over 25%.

Interesting stuff none the less.

I did suspect 2006 would have a higher PP% than usual, since that was when they cracked down on obstruction and (mostly) stuck to it. Can't remember where I heard this, but I think almost all of the scoring rise was due to powerplays; even strength play barely changed.

Even 15-20 and 20-25 are pretty large margins. 20 is 33% larger than 15, and 25 is 25% larger than 20. I would personally narrow it down further. Maybe find the margin that roughly 80% of results fall into (16-24 for example) and break that up into four sections, then also show "below 16" and "above 24" for a total of 6 sections. These numbers are off the top of my head of course. You should go with what the numbers indicate.

 

[haakon84]

I did suspect 2006 would have a higher PP% than usual, since that was when they cracked down on obstruction and (mostly) stuck to it. Can't remember where I heard this, but I think almost all of the scoring rise was due to powerplays; even strength play barely changed.

Even 15-20 and 20-25 are pretty large margins. 20 is 33% larger than 15, and 25 is 25% larger than 20. I would personally narrow it down further. Maybe find the margin that roughly 80% of results fall into (16-24 for example) and break that up into four sections, then also show "below 16" and "above 24" for a total of 6 sections. These numbers are off the top of my head of course. You should go with what the numbers indicate.

Yes that was my plan of attack. If you'd like to fiddle around with the numbers here is the data I'm using...
'99-'09 Goaltenders who played +55 games PP Data

 

[seventieslord]

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

You're trying to convince me that Brodeur doesn't belong in the category of Roy or Hasek. I'm not the biggest "fanboy" in the world, in fact I would love to see a goalie come along and break Marty's records. Having witnessed 552 (give it whatever credence you want, it's a major accomplishment) in person and owning a piece of that history, it would be a privilege to witness again, even from the TV. Any evidence that points to Brodeur potentially being better than either Roy or Hasek is irrelevant to you.

Save percentage wise, he's nowhere near Hasek, yet it's better than Roy. So you invoke "pre-1993" and "context." When Brodeur's numbers show improvement after the lockout, comparison gets restricted to when "both were in the league." When this occurs, you take numbers out that make Roy's save % look worse and numbers out that make Brodeur's save % look better! That's using statistics in your favor, nothing more.

Are we comparing careers or just a sample of their career? The 2001 Playoffs made Marty look human, his sv% was below .900! Roy just happened to be victorious in the final (which is usually used as an indication that Roy is automatically better, a 1-time event). What does he do 2 years later? Faces more shots than he did en route to any of his previous Finals, sv% of .934, NHL playoff record for SO. Well that makes Brodeur look pretty awful and pretty ****ing good at the same time, doesn't it?

You have to take the good and the bad when comparing these two (or more) goalies or else you're not truly doing it objectively. Considering what I have done in my life, using 1 piece of evidence as "proof" isn't really proof, especially when you do everything in your power to make that statistic fit your opinion (which is exactly what it is, remember).

Accept the fact these goalies are going to be compared for what they've done in their careers, not in small chunks. Using chunks can make Goalie A look awesome while B is the worst ever, switch some parameters and Goalie B is the greatest of all time while A looks like Jim Carey circa 1998. Until then, we are completely wasting our time and not having a useful discussion as the same crap gets said in different ways every post.

I've invited this user to come to this thread and discuss this if he would like to.

Dear Mr. Devils Fan:

If some goalie came and broke Brodeur's records it wouldn't mean he was better than Brodeur any more than it means Brodeur is better than Roy and Hasek. you're getting too hung up on career totals.

Read this thread, it is an excellent study done by a very well-respected and educated member of the HOH forum. If you read it and make sense of it, it will show you what you have failed to acknowledge while chatting with me in the NHL section - that scoring levels changing make comparison of stats across eras impossible without context.

Context is exactly what this study does. As you can see, Roy was much further above the league average much more often than Brodeur was.

I am showing you data that spans their entire career, and you're accusing me of cherrypicking! Classic.

"Same crap gets said in different ways every post"? You mean like you saying "Brodeur has a better sv% than Roy" and me saying "no, actually, with era taken into consideration, Brodeur's sv% is not close to Roy's"? Yeah, I know, that gets tiring, doesn't it?

As for evidence that Brodeur is better than Roy and Hasek, there is precious little of it, and most of it revolves around career totals of team-based accomplishments. i want to know who the better goalie was, not whose team was better.

 

[Hockey Outsider]

Your position on wins should be revisited. Allowing the first goal or two in a game is a major factor and changes the way the game is played. Having to play catch-up provides more high quality offensive opportunities for the opposition.

The impact of allowing two goals on thirty shots is identical in terms of SV% whether the two goals are in the first or last period but the impact on the game is significant.

One of the strengths of Ken Dryden was not allowing the early goal.
First/second goals have to be considered.

You bring up a good point -- save percentage is, by definition, an average. An average doesn't take into account whether a goalie is consistent, or plays better in certain situations. For example:

Goalie A: 91% in all game situations
Goalie B: 91% in "normal" situations, 89% in blowout games, 93% in close situations
Goalie C: 91% in "normal" situations, 91% in blowout games, 89% in close situations

Let's assume that there are roughly the same number of blowouts and close situations (OT and tied third periods?) so all three goalies have the same save percentage: 91%.

Save percentage would tell you that all three goalies are equally good, but common sense tells you that Goalie B is the most valuable. Who cares that his level of performance drops a little bit when the outcome of the game is essentially decided -- he's unbeatable when his team really needs him. Goalie C is the opposite, shutting down opponents in the middle of blowouts while faltering in critical situations.

The interesting questions are:
- Are there any goalies who exhibit this type of clutch ability in the regular season?
- How much clutch ability do they demonstrate?
- Is it consistent & repeatable over many years, or is it just a product of small samples sizes?
- Even if we think we've found a goalie with clutch ability, how much of that is due to the goalie, and how much is due to the team?

 

[arrbez]

The interesting questions are:
- Are there any goalies who exhibit this type of clutch ability in the regular season?
- How much clutch ability do they demonstrate?
- Is it consistent & repeatable over many years, or is it just a product of small samples sizes?
- Even if we think we've found a goalie with clutch ability, how much of that is due to the goalie, and how much is due to the team?

Gerry Cheevers is a guy who apparently never cared for his stats in a game that had already been decided, to the point where he even left the net to avoid a Bobby Hull slapshot in a game that was already out of hand (if the story is true). But in a close game, he was money. It would be an interesting study to see if this holds true.

Somewhat along the same lines, the likelihood of a goalie being replaced during a blowout has surely changed over time, and even between coaches. A guy who was left in more often to get shelled will probably suffer stat-wise.

 

[seventieslord]

Cheevers being money in big or close games doesn't seem to jive with the results of his biggest and most important games. I wonder if this was a reputation he earned rather mysteriously. Sometimes I wonder if Cheevers was just the Mike Vernon of his time.