Top-100 Hockey Players of All-Time - Round 2, Vote 6
- Thread starter quoipourquoi
- Start date
MXD
Partying Hard
- Oct 27, 2005
- 51,736
- 17,641
You can still edit this post.
This is the problem comparing raw numbers in different era with different scoring rates and opponents.
Dennis Bonvie
Registered User
You mean its a problem for your scenario?
Sentinel
Registered User
Beat Detroit. And then Detroit won in 02 without Fetisov.
No it's a general problem overall or else we would see a lot more players from the 80s right?
Sentinel
Registered User
Clarke's defensive advantage over Yzerman is smaller than Yzerman's offensive advantage over Clarke. Their leadership is about even (Yzeman slightly ahead), and their classiness is incomparable.
I tend to agree with the first sentence.
The second sentence has little to no bearing on my opinion of the 2 layers in this project.
MXD
Partying Hard
- Oct 27, 2005
- 51,736
- 17,641
ImporterExporter
"You're a boring old man"
Clarke also was the best player on his team during his entire career.
Yzerman was not. Hell when Fedorov was around and actually gave a shit Y wasn't even the best defensive forward on Detroit, circa early through mid 90's. I think Yzerman was a fine defensive player but it took many, many years for him to gain a strong reputation (Bowman years) for his defensive work.
Clarke's adjusted +/- dwarfs Yzerman and you can see the returns on and off the ice that strongly favor Clarke here:
Clarke never enjoyed playing for a dynasty either. Won 3 Harts in the same time frame as Orr, Lafleur and Espo. Yzerman was a Hart finalist, once. Clarke took his team to 3 consecutive finals (winning 2 in 74-75) with the insanely strong Habs dynasty putting an end to the broad street bullies in 76. No lost points for finally being crushed by one of the greatest hockey dynasties ever. I mean Ken Dryden vs Wayne Stephenson. The ridiculous disparity of the defensive corps?
I respected and loved Yzerman as a young man. One of my favorites, but Clarke was simply better. Yzerman had the class, no doubt about it but when it came to complete and total package, Clarke was a superior player. And I loathe Philly haha.
Yzerman was not. Hell when Fedorov was around and actually gave a shit Y wasn't even the best defensive forward on Detroit, circa early through mid 90's. I think Yzerman was a fine defensive player but it took many, many years for him to gain a strong reputation (Bowman years) for his defensive work.
Clarke's adjusted +/- dwarfs Yzerman and you can see the returns on and off the ice that strongly favor Clarke here:
Clarke never enjoyed playing for a dynasty either. Won 3 Harts in the same time frame as Orr, Lafleur and Espo. Yzerman was a Hart finalist, once. Clarke took his team to 3 consecutive finals (winning 2 in 74-75) with the insanely strong Habs dynasty putting an end to the broad street bullies in 76. No lost points for finally being crushed by one of the greatest hockey dynasties ever. I mean Ken Dryden vs Wayne Stephenson. The ridiculous disparity of the defensive corps?
I respected and loved Yzerman as a young man. One of my favorites, but Clarke was simply better. Yzerman had the class, no doubt about it but when it came to complete and total package, Clarke was a superior player. And I loathe Philly haha.
MXD
Partying Hard
- Oct 27, 2005
- 51,736
- 17,641
I was about to write a long post when I realize that Yzerman isn't eligible, so I won't bother.
ImporterExporter
"You're a boring old man"
DitchMarner
TheGlitchintheSwitch
Interestingly, Jagr leads that entire table and you don't seem to think too highly of him as an overall player...not an insult or anything; merely an observation.
seventieslord
Student Of The Game
Manipulated Save Percentage - and how it applies to Sawchuk, Hall and especially Brodeur - Part I
Hockey Outsider's Adjusted Save Percentage is currently the best simple era-adjusted save percentage that anyone has done. You can find it here. Adjusted save percentage (regular season, 1956-2018)
I used it as a springboard for the following post. Particularly, post #5, which is a table of the leaders for "peak adjusted save percentage" - based on their best 7 non-consecutive seasons. This is as close to a "7-year VsX" as we have for goalies, disregarding for a moment that save percentage is a rate stat and points are a counting stat (though HO mitigates that by requiring 2400 adjusted minutes for a season to count as a season).
First thing I had to do, was add one more decimal place to these numbers. We are talking about massive sample sizes, so these are significant figures here. Plus, I'm talking about using other factors to manipulate save percentage, so it would be statistically irresponsible to be subtracting .002 from someone's numbers when that really means "anywhere from .0015 to .0024". Here are the top-22 all-time, plus one other relevant goalie:
[TBODY]
[/TBODY]
Sawchuk lags way, way, way behind the rest. But there are reasons for that, which I will get to.
Right off the bat, let's agree that these are really excellent goalies. In the long run, the best goalies of a generation will post the best save percentage numbers - clearly.
But Martin Brodeur, the not-quite-consensus third best goalie of the last 30 years, is barely ahead of Barrasso and Joseph.... and behind Luongo and Belfour! What gives?
Save percentage is a good starting point stat to discuss a goalie's value, but if we have more information we should use it. There are three factors that need to be looked at:
1. Brodeur faced far fewer powerplays, on a per-game basis, than any other goalie in the eras his career spanned. (adjusting for this will hurt his save percentage).
2. Brodeur has had a noticeable effect on the number of shots the Devils surrender, and a less-noticeable but measurable effect on the number of shots that the Devils take (adjusting for this will help his save percentage).
3. Brodeur has had his shots against count supressed by stingy shot counters in New Jersey, at least since 1998-99. (adjusting for this will help his save percentage).
Let's start with the powerplays. I checked every season between 1993-94 and 2009-10 (except the incomplete 2008-09) and compared Brodeur's percentage of shots that were PP shots, to the average of the other top-40 goalies (I'd have checked all goalies, but it was manual data entry and too time-consuming, so I took the 40 that accounted for most of the shots taken in each season). I took note of what Brodeur's rate of PP shots were compared to the average (for example, in 1993-94, his rate was .924, meaning his ratio of PP shots was 92.4% as high as the average). I then calculated how many more PP shots Brodeur would had to have faced in order for his ratio to match the average. Using his PP save percentage I then calculated how many goals he'd have surrendered on those shots.
But, the NHL only had situational play tracked from 1997-98 onwards, so some projections had to be done for the 93-94 to 96-97 seasons. It turned out to be pretty straight forward. I noticed, unsurprisingly, that Brodeur's PP shot rate compared to the average of other goalies, almost perfectly tracked the Devils' PPOA rate compared to the average of the league. So using that number, and the actual calculated results from 97-98 to 09-10 on a data plot with a trendline, I was able to estimate the save percentage drop he should have, and build the "missing" shots and saves from there.
Here's what happened to his numbers:
[TBODY]
[/TBODY]
As you can see, lack of powerplays against have really helped Brodeur. If you normalize this to a league average, he has seasons where he loses as much as 0.0059 from his save percentage. But on average it's a loss of .0024. His best seven seasons are still the same as they were before: 1994, 1996, 1997, 1998, 2006, 2007, 2008. His 7-year adjusted SP becomes .9172. Refer back to the list up above, and he actually drops five spots on the list, below Hall, Barrasso, Joseph, Vokoun and Lundqvist.
Wait, wait, put the pitchforks away! This was just the start. We have to get to the good stuff. It's important to note, however, that even if we stopped right here, Brodeur would have the peak numbers of a top-20 goalie of the save percentage era, which now goes back 65 years.
The next thing to do is use his PP-adjusted numbers as a the starting point for a shot under-counting adjustment. As BM67 has shown us:
(I added one more table into the quote to demonstrate the difference more easily)
I don't see any other way to interpret this as, it was harder to get credited with a shot in New Jersey. This applied to both teams equally. Some scorekeepers have different standards for what constitutes a shot. In the long run this should wash out for most players, but for Brodeur, it happened for almost his whole career and to be fair, we must account for it.
There does not appear to be an issue with shot counting until 1998-99, at which points the scorers seemed to count an average of 3.75 fewer shots per game in New Jersey, than the Devils and their opponents were credited with in other buildings. Divide this number by two (because Brodeur and the opposing goalie were equally affected) and then by two again (because this only happened in home games, 50% of Brodeur's games played) and you get 0.94. This is the number of shots against per game (and saves per game) that must be added to Brodeur's totals to fairly account for this - in 1998-99 and beyond. If we do that, we get:
[TBODY]
[/TBODY]
1999-00 actually replaces 1997-98 as one of his seven best seasons, and Brodeur ends up with a peak adjusted sv% of .9194. Refer back up to the table at the top, and you can see he's actually right back where he started, and 3 points higher now, in a tie with Dan Bouchard.
Lastly, we need to account for Brodeur's puckhandling which affected shots for and against for the Devils. Again, BM67's data:
This is a very large data sample - it may feel right to say that Devils' backups played about 40 games from 1997 to 2014, but they actually logged 311 in that time - more than enough to confidently say that, compared to a field of middling puckhandlers, Brodeur's impact was felt in the form of 0.58 shots per game, and 1.04 shots against per game. This is a net of 1.62 shots (and saves) per game that Brodeur deserves to be credited with.
Doing so, we get this:
[TBODY]
[/TBODY]
The "AdjSv%4" column reflects Brodeur's adjusted save percentage after all manipulations are made. The "diff" column shows how many points his save percentage changes by, positively, following these adjustments. A few seasons are affected by 7 or 8 sv% points!
Brodeur's 7-year peak adjusted save percentage now becomes .9231. Refer back to the top list, and he goes all the way up to 6th, right between Bernie Parent and Jacques Plante!
Now, don't get too used to the orders of the other goalies on this list. That requires more manipulation, which I'll do in part 2.
Also note that Brodeur's string of save percentage top-10 finishes would look like this now: 1, 2, 2, 2, 2, 5, 5, 6, 9. Compare to Patrick Roy (adjusted to account for Brodeur's adjusted placements): 1, 1, 1, 1, 2, 2, 4, 4, 5, 6, 7, 8, 8, 10. Roy still has a much better peak of delivering high-end puck stopping, and great longevity doing so, where Brodeur, even after adjustments, fails to make the top-10 in save percentage for half his career. Still, he acquits himself much better after adjustments, and his non-top-10 years (through 2010) are not bad, or even below average.
I did my best to account for all factors affecting Martin Brodeur, but I did not change anything for any other goalie. A true statistical analysis of all goalies of the save percentage era would need to include factors like this to be as accurate as possible, but I wonder whether it's necessary. If you could identify a goalie who always faced way more or less than his share of powerplays, played in an arena that overcounted or undercounted shots for a long period of time, or was exceptionally good or bad atpuckhandling, then I suppose they could be handled as a one-off, but for the most part, most long-career goalies played in a number of different situations. Not Brodeur. His statistical impediments and advantages were true for nearly his entire career and simply had to be accounted for.
Voila:
Partially manipulated all-time peak adjusted save percentage leaders:
[TBODY]
[/TBODY]
Hockey Outsider's Adjusted Save Percentage is currently the best simple era-adjusted save percentage that anyone has done. You can find it here. Adjusted save percentage (regular season, 1956-2018)
I used it as a springboard for the following post. Particularly, post #5, which is a table of the leaders for "peak adjusted save percentage" - based on their best 7 non-consecutive seasons. This is as close to a "7-year VsX" as we have for goalies, disregarding for a moment that save percentage is a rate stat and points are a counting stat (though HO mitigates that by requiring 2400 adjusted minutes for a season to count as a season).
First thing I had to do, was add one more decimal place to these numbers. We are talking about massive sample sizes, so these are significant figures here. Plus, I'm talking about using other factors to manipulate save percentage, so it would be statistically irresponsible to be subtracting .002 from someone's numbers when that really means "anywhere from .0015 to .0024". Here are the top-22 all-time, plus one other relevant goalie:
| Ken Dryden | 0.9338 |
| Dominik Hasek | 0.9319 |
| Tony Esposito | 0.9292 |
| Patrick Roy | 0.9291 |
| Bernie Parent | 0.9264 |
| Glenn Resch | 0.9229 |
| John Vanbiesbrouck | 0.9220 |
| Jacques Plante | 0.9209 |
| Johnny Bower | 0.9200 |
| Ed Belfour | 0.9199 |
| Roberto Luongo | 0.9196 |
| Dan Bouchard | 0.9194 |
| Martin Brodeur | 0.9191 |
| Glenn Hall | 0.9188 |
| Tom Barrasso | 0.9186 |
| Curtis Joseph | 0.9184 |
| Tomas Vokoun | 0.9178 |
| Henrik Lundqvist | 0.9175 |
| Rogie Vachon | 0.9168 |
| Sean Burke | 0.9167 |
| Kelly Hrudey | 0.9167 |
| Andy Moog | 0.9166 |
| Terry Sawchuk | 0.9033 |
Sawchuk lags way, way, way behind the rest. But there are reasons for that, which I will get to.
Right off the bat, let's agree that these are really excellent goalies. In the long run, the best goalies of a generation will post the best save percentage numbers - clearly.
But Martin Brodeur, the not-quite-consensus third best goalie of the last 30 years, is barely ahead of Barrasso and Joseph.... and behind Luongo and Belfour! What gives?
Save percentage is a good starting point stat to discuss a goalie's value, but if we have more information we should use it. There are three factors that need to be looked at:
1. Brodeur faced far fewer powerplays, on a per-game basis, than any other goalie in the eras his career spanned. (adjusting for this will hurt his save percentage).
2. Brodeur has had a noticeable effect on the number of shots the Devils surrender, and a less-noticeable but measurable effect on the number of shots that the Devils take (adjusting for this will help his save percentage).
3. Brodeur has had his shots against count supressed by stingy shot counters in New Jersey, at least since 1998-99. (adjusting for this will help his save percentage).
Let's start with the powerplays. I checked every season between 1993-94 and 2009-10 (except the incomplete 2008-09) and compared Brodeur's percentage of shots that were PP shots, to the average of the other top-40 goalies (I'd have checked all goalies, but it was manual data entry and too time-consuming, so I took the 40 that accounted for most of the shots taken in each season). I took note of what Brodeur's rate of PP shots were compared to the average (for example, in 1993-94, his rate was .924, meaning his ratio of PP shots was 92.4% as high as the average). I then calculated how many more PP shots Brodeur would had to have faced in order for his ratio to match the average. Using his PP save percentage I then calculated how many goals he'd have surrendered on those shots.
But, the NHL only had situational play tracked from 1997-98 onwards, so some projections had to be done for the 93-94 to 96-97 seasons. It turned out to be pretty straight forward. I noticed, unsurprisingly, that Brodeur's PP shot rate compared to the average of other goalies, almost perfectly tracked the Devils' PPOA rate compared to the average of the league. So using that number, and the actual calculated results from 97-98 to 09-10 on a data plot with a trendline, I was able to estimate the save percentage drop he should have, and build the "missing" shots and saves from there.
Here's what happened to his numbers:
| Year | PPOA% | PPSA% | Extra SOG | Extra G | Sv% | PPAdj | Diff | AdjSv% | AdjSv%2 | Diff |
| 1993-94 | 0.924 | 0.924 | 96 | 10 | 0.9152 | 0.9138 | 0.0014 | 0.9253 | 0.9239 | 0.0014 |
| 1994-95 | 0.713 | 0.713 | 128 | 16 | 0.9020 | 0.8986 | 0.0033 | 0.9064 | 0.9031 | 0.0034 |
| 1995-96 | 0.772 | 0.772 | 119 | 16 | 0.9115 | 0.9088 | 0.0026 | 0.9181 | 0.9155 | 0.0027 |
| 1996-97 | 0.699 | 0.699 | 162 | 18 | 0.9265 | 0.9231 | 0.0034 | 0.9268 | 0.9234 | 0.0034 |
| 1997-98 | 0.813 | 0.737 | 110 | 15 | 0.9171 | 0.9136 | 0.0035 | 0.9159 | 0.9124 | 0.0035 |
| 1998-99 | 0.905 | 0.872 | 55 | 8 | 0.9063 | 0.9047 | 0.0016 | 0.9035 | 0.9019 | 0.0016 |
| 1999-00 | 0.946 | 0.936 | 25 | 2 | 0.9104 | 0.9105 | -0.0001 | 0.9110 | 0.9112 | -0.0001 |
| 2000-01 | 0.851 | 0.688 | 167 | 27 | 0.9058 | 0.8999 | 0.0058 | 0.9075 | 0.9017 | 0.0059 |
| 2001-02 | 0.784 | 0.821 | 68 | 10 | 0.9057 | 0.9037 | 0.0021 | 0.9030 | 0.9009 | 0.0021 |
| 2002-03 | 0.727 | 0.687 | 133 | 18 | 0.9138 | 0.9103 | 0.0036 | 0.9100 | 0.9064 | 0.0035 |
| 2003-04 | 0.764 | 0.861 | 95 | 12 | 0.9165 | 0.9144 | 0.0021 | 0.9106 | 0.9085 | 0.0021 |
| 2005-06 | 0.727 | 0.798 | 143 | 19 | 0.9112 | 0.9084 | 0.0028 | 0.9149 | 0.9121 | 0.0028 |
| 2006-07 | 0.681 | 0.736 | 159 | 15 | 0.9216 | 0.9205 | 0.0011 | 0.9214 | 0.9203 | 0.0011 |
| 2007-08 | 0.895 | 1.057 | 30 | -3 | 0.9196 | 0.9199 | -0.0003 | 0.9154 | 0.9157 | -0.0003 |
| 2009-10 | 0.789 | 0.638 | 115 | 16 | 0.9162 | 0.9132 | 0.0030 | 0.9099 | 0.9069 | 0.0030 |
As you can see, lack of powerplays against have really helped Brodeur. If you normalize this to a league average, he has seasons where he loses as much as 0.0059 from his save percentage. But on average it's a loss of .0024. His best seven seasons are still the same as they were before: 1994, 1996, 1997, 1998, 2006, 2007, 2008. His 7-year adjusted SP becomes .9172. Refer back to the list up above, and he actually drops five spots on the list, below Hall, Barrasso, Joseph, Vokoun and Lundqvist.
Wait, wait, put the pitchforks away! This was just the start. We have to get to the good stuff. It's important to note, however, that even if we stopped right here, Brodeur would have the peak numbers of a top-20 goalie of the save percentage era, which now goes back 65 years.
The next thing to do is use his PP-adjusted numbers as a the starting point for a shot under-counting adjustment. As BM67 has shown us:
(I added one more table into the quote to demonstrate the difference more easily)
I don't see any other way to interpret this as, it was harder to get credited with a shot in New Jersey. This applied to both teams equally. Some scorekeepers have different standards for what constitutes a shot. In the long run this should wash out for most players, but for Brodeur, it happened for almost his whole career and to be fair, we must account for it.
There does not appear to be an issue with shot counting until 1998-99, at which points the scorers seemed to count an average of 3.75 fewer shots per game in New Jersey, than the Devils and their opponents were credited with in other buildings. Divide this number by two (because Brodeur and the opposing goalie were equally affected) and then by two again (because this only happened in home games, 50% of Brodeur's games played) and you get 0.94. This is the number of shots against per game (and saves per game) that must be added to Brodeur's totals to fairly account for this - in 1998-99 and beyond. If we do that, we get:
| Year | GA | Shots | Saves | Sv% | AdjSv%3 |
| 1993-94 | 115 | 1334 | 1219 | 0.9138 | 0.9239 |
| 1994-95 | 105 | 1070 | 965 | 0.9019 | 0.9063 |
| 1995-96 | 189 | 2073 | 1884 | 0.9088 | 0.9155 |
| 1996-97 | 138 | 1795 | 1657 | 0.9231 | 0.9234 |
| 1997-98 | 142 | 1679 | 1534 | 0.9136 | 0.9124 |
| 1998-99 | 170 | 1849 | 1679 | 0.9081 | 0.9053 |
| 1999-00 | 163 | 1890 | 1727 | 0.9138 | 0.9144 |
| 2000-01 | 193 | 1996 | 1803 | 0.9033 | 0.9050 |
| 2001-02 | 166 | 1791 | 1625 | 0.9073 | 0.9046 |
| 2002-03 | 165 | 1908 | 1743 | 0.9135 | 0.9097 |
| 2003-04 | 166 | 2011 | 1845 | 0.9175 | 0.9115 |
| 2005-06 | 206 | 2316 | 2110 | 0.9111 | 0.9148 |
| 2006-07 | 186 | 2405 | 2229 | 0.9268 | 0.9266 |
| 2007-08 | 165 | 2132 | 1967 | 0.9226 | 0.9184 |
| 2009-10 | 184 | 2189 | 2005 | 0.9159 | 0.9096 |
1999-00 actually replaces 1997-98 as one of his seven best seasons, and Brodeur ends up with a peak adjusted sv% of .9194. Refer back up to the table at the top, and you can see he's actually right back where he started, and 3 points higher now, in a tie with Dan Bouchard.
Lastly, we need to account for Brodeur's puckhandling which affected shots for and against for the Devils. Again, BM67's data:
This is a very large data sample - it may feel right to say that Devils' backups played about 40 games from 1997 to 2014, but they actually logged 311 in that time - more than enough to confidently say that, compared to a field of middling puckhandlers, Brodeur's impact was felt in the form of 0.58 shots per game, and 1.04 shots against per game. This is a net of 1.62 shots (and saves) per game that Brodeur deserves to be credited with.
Doing so, we get this:
| Year | GA | Shots | Saves | Sv% | AdjSv%4 | OrigSv% | Diff |
| 1993-94 | 115 | 1405 | 1290 | 0.9181 | 0.9283 | 0.9152 | 0.0030 |
| 1994-95 | 105 | 1129 | 1024 | 0.9070 | 0.9114 | 0.9020 | 0.0050 |
| 1995-96 | 189 | 2193 | 2004 | 0.9138 | 0.9205 | 0.9115 | 0.0023 |
| 1996-97 | 138 | 1899 | 1761 | 0.9273 | 0.9276 | 0.9265 | 0.0008 |
| 1997-98 | 142 | 1790 | 1648 | 0.9207 | 0.9194 | 0.9171 | 0.0035 |
| 1998-99 | 170 | 1963 | 1793 | 0.9134 | 0.9106 | 0.9063 | 0.0072 |
| 1999-00 | 163 | 2006 | 1843 | 0.9188 | 0.9194 | 0.9104 | 0.0084 |
| 2000-01 | 193 | 2112 | 1919 | 0.9086 | 0.9104 | 0.9058 | 0.0028 |
| 2001-02 | 166 | 1908 | 1742 | 0.9130 | 0.9102 | 0.9057 | 0.0073 |
| 2002-03 | 165 | 2026 | 1861 | 0.9186 | 0.9147 | 0.9138 | 0.0047 |
| 2003-04 | 166 | 2134 | 1968 | 0.9222 | 0.9162 | 0.9165 | 0.0057 |
| 2005-06 | 206 | 2434 | 2228 | 0.9154 | 0.9192 | 0.9112 | 0.0042 |
| 2006-07 | 186 | 2532 | 2346 | 0.9265 | 0.9263 | 0.9216 | 0.0049 |
| 2007-08 | 165 | 2257 | 2092 | 0.9269 | 0.9227 | 0.9196 | 0.0073 |
| 2009-10 | 184 | 2310 | 2126 | 0.9204 | 0.9140 | 0.9162 | 0.0042 |
The "AdjSv%4" column reflects Brodeur's adjusted save percentage after all manipulations are made. The "diff" column shows how many points his save percentage changes by, positively, following these adjustments. A few seasons are affected by 7 or 8 sv% points!
Brodeur's 7-year peak adjusted save percentage now becomes .9231. Refer back to the top list, and he goes all the way up to 6th, right between Bernie Parent and Jacques Plante!
Now, don't get too used to the orders of the other goalies on this list. That requires more manipulation, which I'll do in part 2.
Also note that Brodeur's string of save percentage top-10 finishes would look like this now: 1, 2, 2, 2, 2, 5, 5, 6, 9. Compare to Patrick Roy (adjusted to account for Brodeur's adjusted placements): 1, 1, 1, 1, 2, 2, 4, 4, 5, 6, 7, 8, 8, 10. Roy still has a much better peak of delivering high-end puck stopping, and great longevity doing so, where Brodeur, even after adjustments, fails to make the top-10 in save percentage for half his career. Still, he acquits himself much better after adjustments, and his non-top-10 years (through 2010) are not bad, or even below average.
I did my best to account for all factors affecting Martin Brodeur, but I did not change anything for any other goalie. A true statistical analysis of all goalies of the save percentage era would need to include factors like this to be as accurate as possible, but I wonder whether it's necessary. If you could identify a goalie who always faced way more or less than his share of powerplays, played in an arena that overcounted or undercounted shots for a long period of time, or was exceptionally good or bad atpuckhandling, then I suppose they could be handled as a one-off, but for the most part, most long-career goalies played in a number of different situations. Not Brodeur. His statistical impediments and advantages were true for nearly his entire career and simply had to be accounted for.
Voila:
Partially manipulated all-time peak adjusted save percentage leaders:
| Ken Dryden | 0.9338 |
| Dominik Hasek | 0.9319 |
| Tony Esposito | 0.9292 |
| Patrick Roy | 0.9291 |
| Bernie Parent | 0.9264 |
| Martin Brodeur after adjustments | 0.9231 |
| Glenn Resch | 0.9229 |
| John Vanbiesbrouck | 0.922 |
| Jacques Plante | 0.9209 |
| Johnny Bower | 0.92 |
| Ed Belfour | 0.9199 |
| Roberto Luongo | 0.9196 |
| Dan Bouchard | 0.9194 |
| Real life Martin Brodeur | 0.9191 |
| Glenn Hall | 0.9188 |
| Tom Barrasso | 0.9186 |
| Curtis Joseph | 0.9184 |
| Tomas Vokoun | 0.9178 |
| Henrik Lundqvist | 0.9175 |
| Rogie Vachon | 0.9168 |
| Sean Burke | 0.9167 |
| Kelly Hrudey | 0.9167 |
| Andy Moog | 0.9166 |
| Terry Sawchuk | 0.9033 |
ImporterExporter
"You're a boring old man"
Since you quoted me and mentioned double J....
I always gave respect to Jagr for his even strength scoring. It's elite. One of the selling points for him.
My biggest problem again, was his scoring dipped, often in the postseason, sometimes by a good deal and the Pens never played D, so they chased empty points often in the regular season.
Some people have him over Lidstrom and I'll never understand why. Watched both hundreds of times. One guy was an artist. The other was Rembrandt.
ImporterExporter
"You're a boring old man"
DitchMarner
TheGlitchintheSwitch
It's really a question of how much adjusted plus/minus matters and how much stock should be put in it. I think it has relevance, but I personally can't say how much. Some of the scores on that table support the obvious (Orr was a beast, Lindros titled the ice heavily etc), but then you have guys like Kunitz and Khristich in that table.
In any event, I don't know if anyone would argue against the notion that Clarke was better defensively than Yzerman. Clarke may be the best defensive forward ever.
I agree to no such thing about Kelly Hrudey.
And I imagine I'm jumping the gun, because there is a part 2 (I presume), but I'm guessing Sawchuk is probably playing with his 6th best to 12th best seasons here.
Hockey Outsider
Registered User
- Jan 16, 2005
- 9,485
- 15,776
This is a great post - the kind of post that we should be referring to years from now as a starting point for any serious discussion about Brodeur. (You did pretty much what I was hoping you'd do when examining the numbers).
It's late and I'll dig through the numbers in more detail tomorrow, but at a high level everything seems to make sense.
Obviously since you're only looking at save percentage (a rate statistic), Brodeur is still being underrated because his ability to play a huge number of games, year after year, is being ignored. In my original thread, I also looked at a couple of stats - goals versus average and goals versus threshold - which, in different ways, give him credit for his durability.
Even before these adjustments, he ranked 14th in career GVA 4th in career GVT - I could easily see him climbing to 10th and 3rd based on his adjusted numbers. He was also just outside the top ten for peak in both metrics, but since there's not a lot of separation, he could climb 3-5 spots pretty quickly.
I had Hall ahead of Brodeur last round, but I'm going to seriously reconsider if that's the right order.
Phil Esposito
An offensive superstar. Big for Boston during their mini-dynasty. Perhaps my math is off but my ES totals for the Bruins 1970-72 mini-dynasty has Esposito’s playoff R-on at 2.12 (53/25) and Orr’s at 2.09 (67/32). Espo had 37 ES points which seems a tad low compared to other superstar centres who are usually closer to 80% on the eyeball test, so perhaps there were a few free pluses, but I don’t know if it’s red flag low.
Also big for Team Canada where he was without Bobby Hull or Bobby Orr, the two guys who seem to be the source of the big knock against Phil (in that he got to play with them.)
Orr was surprisingly healthy during his prime when it came to not missing games. Still, even in the small sample sizes here: Phil Esposito Without the Orr Factor Phil Esposito is scoring at an Art Ross pace without Bobby Orr in 1968 (which he didn’t actually win), and 1969 and 1973 (which he did actually win). Even in 1974 the non-Orr pace is enough to win a Ross against non-Bruins.
[TBODY]
[/TBODY]
It is also possible that Orr carrying the puck had a push/pull effect on Esposito's scoring. The obvious interpretation has been that Orr scoring is a chance to collect points, but Orr outshooting Esposito (Shot share is just shots between the two of them) doesn't really seem to benefit Phil. In 1970 at least, it's possible that Phil's scoring chances were gobbled up in the opportunity cost of Orr having the puck all the time. Kind of like Lebron James on last year's Cavs.
[TBODY]
[/TBODY]
An offensive superstar. Big for Boston during their mini-dynasty. Perhaps my math is off but my ES totals for the Bruins 1970-72 mini-dynasty has Esposito’s playoff R-on at 2.12 (53/25) and Orr’s at 2.09 (67/32). Espo had 37 ES points which seems a tad low compared to other superstar centres who are usually closer to 80% on the eyeball test, so perhaps there were a few free pluses, but I don’t know if it’s red flag low.
Also big for Team Canada where he was without Bobby Hull or Bobby Orr, the two guys who seem to be the source of the big knock against Phil (in that he got to play with them.)
Orr was surprisingly healthy during his prime when it came to not missing games. Still, even in the small sample sizes here: Phil Esposito Without the Orr Factor Phil Esposito is scoring at an Art Ross pace without Bobby Orr in 1968 (which he didn’t actually win), and 1969 and 1973 (which he did actually win). Even in 1974 the non-Orr pace is enough to win a Ross against non-Bruins.
| Year | Non-Orr GP | Points | W/ Orr GP | Pts | Non Orr Pace | Actual |
| 1968 | 27 | 32 | 47 | 52 | 87 | 84 |
| 1969 | 9 | 14 | 65 | 112 | 115 | 126 |
| 1972 | 2 | 0 | 74 | 133 | 0 | 133 |
| 1973 | 15 | 24 | 63 | 106 | 125 | 130 |
| 1974 | 4 | 5 | 74 | 140 | 98 | 145 |
| Total | 57 | 75 | 323 | 542 | 100 | 128 |
| 1960s | 36 | 46 | 112 | 164 | 94 | X |
| 1970s | 21 | 29 | 211 | 379 | 108 | X |
It is also possible that Orr carrying the puck had a push/pull effect on Esposito's scoring. The obvious interpretation has been that Orr scoring is a chance to collect points, but Orr outshooting Esposito (Shot share is just shots between the two of them) doesn't really seem to benefit Phil. In 1970 at least, it's possible that Phil's scoring chances were gobbled up in the opportunity cost of Orr having the puck all the time. Kind of like Lebron James on last year's Cavs.
| Espo Pts | Orr Pts | Espo Shots | Orr Shots | Shot Share |
| 84 | 31 | 284 | 173 | 62.1% |
| 126 | 64 | 351 | 285 | 55.2% |
| 99 | 120 | 405 | 413 | 49.5% |
| 152 | 139 | 550 | 392 | 58.4% |
| 133 | 117 | 426 | 353 | 54.7% |
| 130 | 101 | 411 | 282 | 59.3% |
| 145 | 122 | 393 | 384 | 50.6% |
| 127 | 135 | 346 | 384 | 47.4% |
[/TBODY]
Sergei Makarov
One caveat off the bat, is that I don’t see the Soviet League stats translating very well to the NHL context. The Red Army was usually stacked.
1988-89 with Green Unit: 30-8-6, 224 GF / 118 GA
1989-90 with 0.5 of a Kasatonov: 33-9-6, 200 GF / 106 GA
That being said, in the 1981 and 1984 Canada Cups, Makarov and Bossy seemed to be of comparable quality. Of course Makarov outlasted Bossy, had a strong 1987 Canada Cup and when the chance to play in the NHL arrived he added a few solid seasons of being Hakan Loob.
I think Bossy did things that impress me a little more often than Makarov. (See: the playoffs 1980-83). But Makarov is in an odd place where you could liken him to Mike Tyson. Tyson had fewer good opponents in his heyday, but he dominated them with early knockouts. You can only punch the guys they put in front of you right? While I’m currently leaning towards this logic that puts Makarov high, that logic also works for Frank McGee. And in this case he was outscored by Fetisov at the Olympics 33-28 over 22 games, and wasn’t perennially the team’s top scorer at the IIHF World Championships. That’s not too dissimilar from Bossy in the playoffs though, and her certainly passed the eye/stats tests in best-on-best settings.
One caveat off the bat, is that I don’t see the Soviet League stats translating very well to the NHL context. The Red Army was usually stacked.
1988-89 with Green Unit: 30-8-6, 224 GF / 118 GA
1989-90 with 0.5 of a Kasatonov: 33-9-6, 200 GF / 106 GA
That being said, in the 1981 and 1984 Canada Cups, Makarov and Bossy seemed to be of comparable quality. Of course Makarov outlasted Bossy, had a strong 1987 Canada Cup and when the chance to play in the NHL arrived he added a few solid seasons of being Hakan Loob.
I think Bossy did things that impress me a little more often than Makarov. (See: the playoffs 1980-83). But Makarov is in an odd place where you could liken him to Mike Tyson. Tyson had fewer good opponents in his heyday, but he dominated them with early knockouts. You can only punch the guys they put in front of you right? While I’m currently leaning towards this logic that puts Makarov high, that logic also works for Frank McGee. And in this case he was outscored by Fetisov at the Olympics 33-28 over 22 games, and wasn’t perennially the team’s top scorer at the IIHF World Championships. That’s not too dissimilar from Bossy in the playoffs though, and her certainly passed the eye/stats tests in best-on-best settings.
seventieslord
Student Of The Game
Manipulated Save Percentage - and how it applies to Sawchuk, Hall and especially Brodeur - Part II
Let me start by posting the table from the end of part 1:
[TBODY]
[/TBODY]
This table contains the top-23 all-time in peak adjusted save percentage, plus Terry Sawchuk, and a fairly adjusted Martin Brodeur. But I'm just getting started!
Take a look at numbers 1, 3, 5, and 7 on this list - all 1970s goalies. Watered down talent pool thanks to the WHA and fast expansion. Sure, they were great goalies, but comparing them to the league average save percentage really overstates their greatness.
Then, take a look at 9, 10, 15, and... I don't know, 155? Plante, Bower, Hall and Sawchuk were the best statistical goalies of their time but none of them can top the adjusted results of the 1970s goalies. They had to compete against a league average that was usually a HHOF goalie like Worsley, Sawchuk, or Lumley - that's not fair or right. Now I know HO had a system and it's completely sound mathematically and I respect things like that, but anyone who has known me long enough knows I am a fan of fudging and manipulating numbers subjectively to help them "look" right, and this is a case where that is necessary.
First of all, poor Terry Sawchuk. He's taking a kicking in this discussion, and maybe rightfully so, but his peak was hella dominant, and the save percentage era doesn't cover any of that. His first five seasons are probably his best five, and none of them are counted here because they predate regular season save percentage - also, one Jacques Plante season is missing. If I could just find a way to simulate save percentage for five seasons, then the entire careers of two top-6 goalies of all-time would be covered....
...so I did that. This is by no means an exact science, but there is enough information to use to make educated guesses. The NHL of 1950-55 was very similar to the NHL of 1956-60. We had a situation where one team was the best, and they also had the consensus best goalie. We can be certain that he had a very dominant save percentage, but also certain that his dominant team was surrendering fewer shots than the league average. And the total number of league goals was quite consistent from 1950 right through 1957, so you can expect that the total shots taken would have been as well. As it turned out, while I was manipulating the figures, I had to walk a fine line between making the league's SOG/60 averages abnormally low, and pumping up the league's average sv% to abnormally high figures. I ended up with league average save percentages of 91.01, 91.38, 91.84, 91.76, and 91.44.
I then assigned Sawchuk a workload in each season similar to what Jacques Plante, starter on the most dominant team from 1956-1960 had, and from there gave him save percentages of .9266, .9293, .9300, .9295, and .9298. I believe this is consistent with a goalie who would win the 1st or 2nd all-star team spot in each of these seasons. This was also quite conservative, because out of all pre-expansion 50-game seasons, these were just the 11th, 14th, 17th, 25th and 26th best seasons posted by any goalie. Then again, Plante had 5 in that range, Hall had 7, Bower had 4 and Worsley had 3 (and Giacomin and Hodge one apiece), so that seems fair. Still, I'm revisiting this, because intuitively, Sawchuk was thought to have the best peak of the era... doesn't mean he needs to put up all of the best five seasons ever, but his best placing higher than 11th would be a start.
(ten minutes later)... took a look at those seasons and noticed that scoring was 10% higher in 56-60 than it was in 51-55, so since I have no idea if it was because goalies were stopping more shots, or there were fewer shots, I manipulated shot totals to the point where they were 50% responsible, and save percentages accounted for the rest.
As you can imagine, this did wonders for Sawchuk's adjusted save percentage. I'd show you, but that was an intermediate step and I didn't save that step like I did for Brodeur.
The next major manipulation I did, was to subjectively change the "league average" save percentages every season to better reflect the competition level among goalies. I didn't go by any special formula, I simply went case-by-case.
In 1951-1955, I bumped the average down similar to how I did in 1956-1967 (see below)
In 1956-1967, I bumped the league average down to be approximately in-between the lowest starting goalie and the 2nd-lowest. This is done so that adjusted SP now shows how a goalie compared to the 5th-6th best goalie, instead of the 3rd-4th (which makes a world of difference)
In 1968-1972, I did nothing. There were 12 teams and a platoon system, so plenty of decent goalies with only one major league to play in. The league was not watered down. (Competition level by division is an issue, but adjusting for that would likely be a nightmare)
In 1973-1979, I bumped the league average up to be approximately 1/3 of the way up the list instead of right in the middle. The league expanded too quickly, and the best WHA goalies could have played in the NHL, keeping the really bad ones (who hurt the league average) out of games. Goalies compared to this horrible league average look too good.
In 1980-2018, I did nothing. There have been 21-31 teams, the NHL was again the only top pro league, and if you want to say it was watered down in the first half of the 80's, I wouldn't disagree, but it was nothing compared to the mid-late 70s. In 1985, in theory the worst goalie in the league was the 42nd best in north america. In 1977, the 60th best north american goalie could get playing time in the NHL.
This had the effect of making Plante, Hall, Sawchuk and Bower look better - and it should. They were not that dominant compared to the league average, only because the league average was an impossibly high standard. It also had the effect of making Dryden, Espo, Resch, Parent and Vachon look worse - and again, it should. They tore apart a severely weakened NHL with a league average baseline influenced by some truly terrible goalies.
(in the meantime I managed to make it so that Sawchuk had the 5th, 9th, 20th, and 25th best seasons posted by any goalie before the lockout, better appreciating his early peak while still being conservative).
After all that, here is what I got:
[TBODY]
[/TBODY]
Note that the modern goalies show up at 1, 4, 10, 11. The pre-expansion guys are at 2, 3, 6, and 8. The 70s guys are 5, 7, 9, 12. I think that after these adustments that reflect how difficult their benchmarks were to hit, the fact that they are sprinked among the top-12 with a fairly even distribution tells us that this manipulated adjusted save percentage does treat all goalies equally, by not treating their eras as equal.
This is completely subjective. But you must agree that when Gump Worsley and Terry Sawchuk are league average goalies, that's not the same as when Wayne Stephenson and Doug Favell are league average.
Except in the case of Johnny Bower (for whom something funny might have been going on), I think the above list does a fair job of ranking these goalies by their peaks. Considering longevity and consistency, I have no problem with putting Brodeur ahead of Parent, Sawchuk and Dryden (and of course, Bower), but for where they all look as peak performers, I like it.
Let me start by posting the table from the end of part 1:
| Ken Dryden | 0.9338 |
| Dominik Hasek | 0.9319 |
| Tony Esposito | 0.9292 |
| Patrick Roy | 0.9291 |
| Bernie Parent | 0.9264 |
| Martin Brodeur after adjustments | 0.9231 |
| Glenn Resch | 0.9229 |
| John Vanbiesbrouck | 0.922 |
| Jacques Plante | 0.9209 |
| Johnny Bower | 0.92 |
| Ed Belfour | 0.9199 |
| Roberto Luongo | 0.9196 |
| Dan Bouchard | 0.9194 |
| Real life Martin Brodeur | 0.9191 |
| Glenn Hall | 0.9188 |
| Tom Barrasso | 0.9186 |
| Curtis Joseph | 0.9184 |
| Tomas Vokoun | 0.9178 |
| Henrik Lundqvist | 0.9175 |
| Rogie Vachon | 0.9168 |
| Sean Burke | 0.9167 |
| Kelly Hrudey | 0.9167 |
| Andy Moog | 0.9166 |
| Terry Sawchuk | 0.9033 |
This table contains the top-23 all-time in peak adjusted save percentage, plus Terry Sawchuk, and a fairly adjusted Martin Brodeur. But I'm just getting started!
Take a look at numbers 1, 3, 5, and 7 on this list - all 1970s goalies. Watered down talent pool thanks to the WHA and fast expansion. Sure, they were great goalies, but comparing them to the league average save percentage really overstates their greatness.
Then, take a look at 9, 10, 15, and... I don't know, 155? Plante, Bower, Hall and Sawchuk were the best statistical goalies of their time but none of them can top the adjusted results of the 1970s goalies. They had to compete against a league average that was usually a HHOF goalie like Worsley, Sawchuk, or Lumley - that's not fair or right. Now I know HO had a system and it's completely sound mathematically and I respect things like that, but anyone who has known me long enough knows I am a fan of fudging and manipulating numbers subjectively to help them "look" right, and this is a case where that is necessary.
First of all, poor Terry Sawchuk. He's taking a kicking in this discussion, and maybe rightfully so, but his peak was hella dominant, and the save percentage era doesn't cover any of that. His first five seasons are probably his best five, and none of them are counted here because they predate regular season save percentage - also, one Jacques Plante season is missing. If I could just find a way to simulate save percentage for five seasons, then the entire careers of two top-6 goalies of all-time would be covered....
...so I did that. This is by no means an exact science, but there is enough information to use to make educated guesses. The NHL of 1950-55 was very similar to the NHL of 1956-60. We had a situation where one team was the best, and they also had the consensus best goalie. We can be certain that he had a very dominant save percentage, but also certain that his dominant team was surrendering fewer shots than the league average. And the total number of league goals was quite consistent from 1950 right through 1957, so you can expect that the total shots taken would have been as well. As it turned out, while I was manipulating the figures, I had to walk a fine line between making the league's SOG/60 averages abnormally low, and pumping up the league's average sv% to abnormally high figures. I ended up with league average save percentages of 91.01, 91.38, 91.84, 91.76, and 91.44.
I then assigned Sawchuk a workload in each season similar to what Jacques Plante, starter on the most dominant team from 1956-1960 had, and from there gave him save percentages of .9266, .9293, .9300, .9295, and .9298. I believe this is consistent with a goalie who would win the 1st or 2nd all-star team spot in each of these seasons. This was also quite conservative, because out of all pre-expansion 50-game seasons, these were just the 11th, 14th, 17th, 25th and 26th best seasons posted by any goalie. Then again, Plante had 5 in that range, Hall had 7, Bower had 4 and Worsley had 3 (and Giacomin and Hodge one apiece), so that seems fair. Still, I'm revisiting this, because intuitively, Sawchuk was thought to have the best peak of the era... doesn't mean he needs to put up all of the best five seasons ever, but his best placing higher than 11th would be a start.
(ten minutes later)... took a look at those seasons and noticed that scoring was 10% higher in 56-60 than it was in 51-55, so since I have no idea if it was because goalies were stopping more shots, or there were fewer shots, I manipulated shot totals to the point where they were 50% responsible, and save percentages accounted for the rest.
As you can imagine, this did wonders for Sawchuk's adjusted save percentage. I'd show you, but that was an intermediate step and I didn't save that step like I did for Brodeur.
The next major manipulation I did, was to subjectively change the "league average" save percentages every season to better reflect the competition level among goalies. I didn't go by any special formula, I simply went case-by-case.
In 1951-1955, I bumped the average down similar to how I did in 1956-1967 (see below)
In 1956-1967, I bumped the league average down to be approximately in-between the lowest starting goalie and the 2nd-lowest. This is done so that adjusted SP now shows how a goalie compared to the 5th-6th best goalie, instead of the 3rd-4th (which makes a world of difference)
In 1968-1972, I did nothing. There were 12 teams and a platoon system, so plenty of decent goalies with only one major league to play in. The league was not watered down. (Competition level by division is an issue, but adjusting for that would likely be a nightmare)
In 1973-1979, I bumped the league average up to be approximately 1/3 of the way up the list instead of right in the middle. The league expanded too quickly, and the best WHA goalies could have played in the NHL, keeping the really bad ones (who hurt the league average) out of games. Goalies compared to this horrible league average look too good.
In 1980-2018, I did nothing. There have been 21-31 teams, the NHL was again the only top pro league, and if you want to say it was watered down in the first half of the 80's, I wouldn't disagree, but it was nothing compared to the mid-late 70s. In 1985, in theory the worst goalie in the league was the 42nd best in north america. In 1977, the 60th best north american goalie could get playing time in the NHL.
This had the effect of making Plante, Hall, Sawchuk and Bower look better - and it should. They were not that dominant compared to the league average, only because the league average was an impossibly high standard. It also had the effect of making Dryden, Espo, Resch, Parent and Vachon look worse - and again, it should. They tore apart a severely weakened NHL with a league average baseline influenced by some truly terrible goalies.
(in the meantime I managed to make it so that Sawchuk had the 5th, 9th, 20th, and 25th best seasons posted by any goalie before the lockout, better appreciating his early peak while still being conservative).
After all that, here is what I got:
| Dominik Hasek | 0.9319 |
| Jacques Plante | 0.9303 |
| Johnny Bower | 0.9294 |
| Patrick Roy | 0.9291 |
| Ken Dryden | 0.9284 |
| Glenn Hall | 0.9283 |
| Tony Esposito | 0.9262 |
| Terry Sawchuk after Adjustments | 0.9257 |
| Bernie Parent | 0.9234 |
| Martin Brodeur after adjustments | 0.9231 |
| John Vanbiesbrouck | 0.922 |
| Glenn Resch | 0.9203 |
| Ed Belfour | 0.9199 |
| Roberto Luongo | 0.9196 |
| Dan Bouchard | 0.9194 |
| Real life Martin Brodeur | 0.9191 |
| Tom Barrasso | 0.9186 |
| Curtis Joseph | 0.9184 |
| Tomas Vokoun | 0.9178 |
| Henrik Lundqvist | 0.9175 |
| Sean Burke | 0.9167 |
| Kelly Hrudey | 0.9167 |
| Andy Moog | 0.9166 |
| Rogie Vachon | 0.9128 |
| Real life Terry Sawchuk | 0.9033 |
Note that the modern goalies show up at 1, 4, 10, 11. The pre-expansion guys are at 2, 3, 6, and 8. The 70s guys are 5, 7, 9, 12. I think that after these adustments that reflect how difficult their benchmarks were to hit, the fact that they are sprinked among the top-12 with a fairly even distribution tells us that this manipulated adjusted save percentage does treat all goalies equally, by not treating their eras as equal.
This is completely subjective. But you must agree that when Gump Worsley and Terry Sawchuk are league average goalies, that's not the same as when Wayne Stephenson and Doug Favell are league average.
Except in the case of Johnny Bower (for whom something funny might have been going on), I think the above list does a fair job of ranking these goalies by their peaks. Considering longevity and consistency, I have no problem with putting Brodeur ahead of Parent, Sawchuk and Dryden (and of course, Bower), but for where they all look as peak performers, I like it.
seventieslord
Student Of The Game
well you're wrong about Kelly hrudey.
But yes that's obviously correct about Sawchuk.
Last edited:
seventieslord
Student Of The Game
I'm not sure if the durability thing is a big deal. Other goalies getting injured is a factor, I agree, but for the most part, the elite goalies all stayed healthy, and who's to say they couldn't have played 70-75 games if their teams just figured out, like New Jersey figured out early on, that if you have a wagon you ride it because they can handle it?
