Round 2, Vote 1 (HOH Top Goaltenders)

[Bear of Bad News]

No, it doesn't magically even out.

The H/R split for both teams over the 13 years I worked the RS numbers dropped from a 9.6% higher shooting% at home to only 5.4% higher at home in the playoffs.

The SOG/G difference drops from 3.85/G to 3.68/G, but OT probably skews that a bit. (The Devils played 28 OT games, 12 at home, 16 on the road, but 53 more OT minutes at home.)

Brodeur's career H/R split difference for SOG/60 is actually slightly higher in the playoffs, but his PO SV% is higher at home, which it rarely was during the RS. His road SV% is still slightly higher over the last 13 playoffs though, .915 at home and .916 on the road.

This seems to suggest that the effect is more of a function of where the scorer's vantage point is in New Jersey (compared with other arenas).

Having never been to either venue, is there something to this?

 

[quoipourquoi]

Obviously it was a little more out of his own control than you're presenting it, given that he actually posted better numbers that his "competition" and was the goalie shipped out, regardless (small sample marginally statistically better than Belfour in his Calder/Vezina year, and similar to Belfour but miles better than Waite the next year, then better numbers than Puppa, Fuhr, and Draper his first year in Buffalo).

He posted better numbers after he played himself out of contention for being a starter at the beginning of both Chicago seasons, and his save percentage compared to Puppa and Fuhr was almost negligible given the fact that they barely recorded enough games to be on the official record (Hasek was literally one fewer relief performance away from not counting).

You're arguing with me for the sake of arguing. Hasek's 1992-93 - an injured season of 28 games - is not part of his prime. It's barely an official rookie season. And yes, Roy held a similar edge in GAA over Craig Billington (0.58 in the DPE) seven years after his prime had ended.

And if you keep saying that he was better than Ed Belfour in 1990-91 because he had chunks of three games against three teams that averaged a .500 record with a higher save percentage (.914 to .910) than a 74 GP league leader, then I'll probably stop responding to you.

 

[Canadiens1958]

Quebec Leagues

This seems to suggest that the effect is more of a function of where the scorer's vantage point is in New Jersey (compared with other arenas).

Having never been to either venue, is there something to this?

Various developmental leagues in Quebec - midget/junior and tournaments found that vantage point is the major factor.

Very evident when comparing SOG results from a 1,000 - 1,200 seat arena with low vantage points to a NHL venue like the Colisee and points in between. Other factors included distance to each net, end or side vantage points, lighting, experience of the counter and who gets the benefit of the doubt on close shots. Some favour the shooter,some the goalie some split the calls.

 

[Theokritos]

I honestly don't know where those games are coming from but I'm guessing some of those are the Ivestia tourny

Some are, yes.

which from what I have read, is the equivalent of the Spengler Cup.

That's not true. Hašek faced the Soviet National Team at the Izvestia as well as the best non-NHLers from Sweden and Finland. The Spengler Cup is hardly comparable.

 

[quoipourquoi]

And if you keep saying that he was better than Ed Belfour in 1990-91 because he had chunks of three games against three teams that averaged a .500 record with a higher save percentage (.914 to .910) than a 74 GP league leader, then I'll probably stop responding to you.

I apologize if this was a little aggressive, but how is this any different than someone saying that Akim Aliu was "statistically better" than Evgeni Malkin? Seriously.

 

[TheDevilMadeMe]

Also, his save percentage and consistency particularly in 1986, 1989, and 1993; his tendencies to extend a series - rather than putting on less consistent performances when his team is down; his overtime performances; his leadership (which hasn't been discussed thoroughly because we've simply run out of time); his determination to never stop playing in a series despite injury; and a certain amount of sustained relevance. There's a lot of arguments outside of the unexpected Stanley Cups with anemic offensive support.

Leadership is definitely a plus for Roy and Brodeur, and something Hasek really didn't provide. Hasek was quite aloof from his teammates - he was something of an aloof genius that everyone put up with because he was so good at stopping the puck. Brodeur (with some help from Elias) was basically the guy giving the Devils peptalks between periods in the 2012 playoffs, and I know the famous stories about Roy.

 

[ContrarianGoaltender]

Final Post on Hasek vs. Roy in the Playoffs

I wanted to use the playoff projection method to focus a bit on some of Hasek's other playoff seasons, as some have criticized him for getting injured in 2006 and replaced in 2008. Again, note that equivalent seasons to Hasek's 2008 were not included for Sawchuk, Hall and Plante, and that Plante was the only other guy even playing in the playoffs past the age of 40 (Hasek was 43 in 2008). However, I'll run the numbers anyway.

The other question is how exactly to account for a goalie's absence. The 2006 Sens were expected to win 1.57 series and won 1. The 2008 Red Wings were expected to win 2.20 and won all 4. This suggests that counting the expectation against Hasek in seasons he didn't play or got replaced would substantially overestimate the actual cost to his teams of not having Hasek , considering they both won at least one playoff series and the '08 Red Wings did better than expected (obviously, since they won the Cup). A similar argument can actually be made for '97 as well, even if you assume that Hasek was completely faking injury, given that the Sabres were heavy underdogs against the Flyers and would have been longshots to win the Cup even with Hasek.

The options seem to be to either exclude those seasons from his numbers (as was done in the prior post), to count all team results for Hasek, or to give him zeroes on the team success side and fully credit him for the team expectations on the other. I have included a number of scenarios in the following table, with the first ones giving the total team results and the last ones penalizing Hasek for the full team expectancy in the given seasons. I also ran the numbers for Hasek for those who aren't willing to penalize him for either of these seasons or 1997.

Description|Series W|Exp Series|Diff|%|Cups|Exp Cups|Diff|%
All Hasek teams|18|12.25|+5.75|+47%|2|1.15|+0.85|+74%
Hasek + team in '06|13|10.05|+2.95|+29%|1|0.83|+0.17|+20%
Individual w/'06|12|10.05|+1.95|+19%|1|0.83|+0.17|+20%
Individual w/'08|12|10.68|+1.32|+12%|1|0.99|+0.01|+0%
Individual w/both|12|12.25|-0.25|-2%|1|1.15|-0.15|-13%
Individual w/o '97|12|7.34|+4.66|+63%|1|0.60|+0.40|+67%

Adjusting for Regular Season Goalie Achievement

One additional argument for Hasek is that during his regular season peak his individual contribution was so high that it made his teams look much better than they actually were, and that this helps underrate how much he himself did to help his teams succeed in the postseason relative to expectations. Helping a team grab a playoff spot is a vital first step that is not always guaranteed (Ken Dryden was the only goalie in this group to play on teams that always made the playoffs), and in general the better the regular season ranking, the easier the path to the Cup come playoff time.

I ran the numbers again for Roy and Hasek, to try to determine how strong their teams would have been with average goaltending. The win/loss record of the backups was left as is, and the team's winning percentage in the remaining games was estimated under the assumption that they had a league average starting goalie, based on the Pythagorean expected win formula of: (Team Goals per Game ^ 2) / (Team Goals per Game ^ 2 + ((1 - LgAvg save %) * Shots Against / 60 Minutes) ^ 2). This did not take into account shot quality or special teams factors, but it should still provide a better estimate of how good those teams would have been if they had an ordinary netminder instead of a superstar.

Doing this for both Roy and Hasek had a sizable impact on their expected team success. Because Roy was nearly always above average throughout his career, his team winning percentages dropped much across the board with the exception of 1986. Roy's 1994 Canadiens team were expected to finish 9th in the Eastern Conference with an average goalie, so Roy's expected playoff numbers for that year were revised to zero since he may not have been there in the first place if he wasn't a great goalie.

Hasek's effect on regular season winning was even greater than Roy's. The Sabres would not have been expected to make the playoffs in any year from 1995 to 2001 with an average starter, producing expected winning percentages that would have ranked them 11th, 12th, 10th, 11th, 10th and 9th in the Eastern Conference over that period.

Here are the full yearly numbers for Roy and Hasek, for both the actual team results and the average replacement starter scenario:

Goalie|Year|Actual Win%|Series W|Cup|Exp Series|Cup %|Avg Goalie Win%|Exp Series|Cup %
Roy|1986|.544|4|1|1.04|.082|.564|1.19|.110
Roy|1987|.575|1|0|1.27|.053|.552|1.10|.035
Roy|1988|.644|1|0|1.95|.220|.578|1.34|.088
Roy|1989|.719|3|0|2.66|.318|.594|1.49|.059
Roy|1990|.581|1|0|0.77|.062|.499|0.37|.012
Roy|1991|.556|1|0|1.01|.083|.501|0.69|.031
Roy|1992|.581|1|0|1.83|.175|.521|1.33|.073
Roy|1993|.607|4|1|1.19|.155|.551|0.74|.065
Roy|1994|.571|0|0|0.75|.033|.480|0.00|.000
Roy|1996|.634|4|1|1.42|.044|.653|1.54|.061
Roy|1997|.652|2|0|2.17|.248|.569|0.77|.015
Roy|1998|.579|0|0|1.11|.052|.519|0.77|.015
Roy|1999|.598|2|0|1.33|.069|.566|1.11|.039
Roy|2000|.585|2|0|0.89|.032|.549|0.69|.015
Roy|2001|.720|4|1|2.35|.306|.631|1.44|.088
Roy|2002|.604|2|0|0.95|.045|.571|0.75|.023
Roy|2003|.640|0|0|1.27|.095|.541|0.61|.013
Roy|Total|.620|32|4|23.95|2.07|.565|15.94|0.74

Goalie|Year|Actual Win%|Series W|Cup|Exp Series|Cup %|Avg Goalie Win%|Exp Series|Cup %
Hasek|1994|.565|0|0|0.60|.029|.550|0.52|.021
Hasek|1995|.531|0|0|0.50|.006|.432|0.00|.000
Hasek|1997|.561|0|0|1.14|.067|.427|0.00|.000
Hasek|1998|.543|2|0|0.78|.032|.429|0.00|.000
Hasek|1999|.555|3|0|0.59|.015|.448|0.00|.000
Hasek|2000|.518|0|0|0.35|.007|.466|0.00|.000
Hasek|2001|.598|1|0|0.82|.018|.524|0.00|.000
Hasek|2002|.707|4|1|2.16|.351|.619|1.22|.113
Hasek|2006|.689|0|0|1.57|.161|.657|1.28|.097
Hasek|2007|.689|2|0|1.54|.146|.625|1.00|.047
Hasek|2008|.701|0|0|2.20|.315|.696|2.14|.298
Hasek|Starter|.607|12|1|7.34|0.60|.525|2.74|0.18
Hasek|Teams|.625|18|2|12.25|1.15|.553|6.16|0.58

Based on these results, an average goalie on Roy's teams would still have been expected to win 16 playoff series. An average goalie on Hasek's teams from 1994 to 2008 would have been expected to win just 6, and really only 3 if we assume that an average goalie wouldn't be playing in the NHL past the age of 40. Hasek's prime numbers are particularly strong, as based on this measure his Buffalo teams won six playoff series while an average starter with the same teammates would be expected to win 0.5, and went to game 6 OT in the Finals despite having almost a zero chance of winning a Cup in that period with an average starting goalie in net (0.02).

The case can be made that Hasek let his teammates down some in the 1997 playoffs, but the Sabres were always a Cup longshot that year and it can be easily argued that Hasek was the main reason that the team with the third-worst win threshold in the league during the regular season was even in that playoff position in the first place. I am less willing to criticize Hasek for his injuries or struggles past the age of 40 when few other goalies were even able to play in the NHL, much less hold down a starting job in the playoffs.

And that's the sort of thing that's going to bother me here with this metric. Anytime I'm looking at a formula that tells me that Jacques Plante didn't win enough Stanley Cups, my skepticism really kicks in. It's like an unusually high expectation to be placed on a team with regular season success.

Ken Dryden managed to outperform expectations on great teams. And if you look at the entire original six era, I don't think it is out of line to suggest that Jacques Plante's teams won about what they were supposed to. He won 4 Cups with 7 regular season champions (57%). The other 18 regular season champs won 10 Cups (56%).

I know that Patrick Roy's 58.3% chance of beating the 1997 Detroit Red Wings every game wasn't exactly reflective of the 201-110 shot disparity that Roy faced compared to Vernon over six games.

Sure, just like everyone who watched the 1996 Avalanche-Red Wings series would have given the Avs a higher than 13.1% chance of advancing, but that's what I'm crediting Roy with for that series. I agree that regular season record does not perfectly encapsulate team strength, but I do think there is great value in systematic approaches, and I don't think they are effectively countered by a few examples where they don't necessarily exactly match up.

 

[Canadiens1958]

Outside the Box

Leadership is definitely a plus for Roy and Brodeur, and something Hasek really didn't provide. Hasek was quite aloof from his teammates - he was something of an aloof genius that everyone put up with because he was so good at stopping the puck. Brodeur (with some help from Elias) was basically the guy giving the Devils peptalks between periods in the 2012 playoffs, and I know the famous stories about Roy.

Ken Dryden - academic and Dominik Hasek did not fit the profile of an NHL goalie. Teammates accommodated both as recognition of what each brought to the team.

Dryden definitely understood the culture and inner working of the hockey teams he played with - evident in his published works and interviews. Dryden also understood when it was time to move to the next phase of his life.

Do not feel that Hasek ever allowed the media to get a true glimpse of the non-goalie or off ice person.

 

[Rhiessan71]

Hasek's effect on regular season winning was even greater than Roy's. The Sabres would not have been expected to make the playoffs in any year from 1995 to 2001 with an average starter, producing expected winning percentages that would have ranked them 11th, 12th, 10th, 11th, 10th and 9th in the Eastern Conference over that period.

The only reason it looks this way is because of Hasek's peak during this time and the fact that it accounts for so much of his actual career, almost half compared to Roy in fact.

That Roy was able to maintain such a high level of play in the playoffs no matter the year, team or age over almost his entire 18 year career is the reason he is the Playoff King and holds the title of best modern money goalie.

Even when Roy went to the Av's, he still stood out, was always counted amongst the best on that team, a huge leader on those teams and still managed to out perform the likes of Sakic and Forsberg more often than not in the post season.
That was definitely not the case with Hasek once he went to a team as good or better than Roy's Av teams.

Breaking down that data, it doesn't take much to see the disparity in Hasek's numbers after leaving the weaker Sabres. Where Roy's numbers don't have any where near the same kind of disparity after leaving the weaker Habs.

So for me I see it like this...
Is Hasek's peak enough to overcome Roy in the regular season...yes.
Is Hasek's peak enough to overcome Roy in the playoffs...not even remotely.

And honestly, anyone that says that players aren't capable of raising their level of play in playoffs is out to lunch and IMO quite simply haven't played competitive hockey at a high level before. Going that extra mile, sacrificing like you wouldn't do in the regular season, bearing down and upping your level of concentration because of what each game means in the playoffs are all very real.
I mean hell, the entire Oiler teams from the mid 80's proves this fully and completely.

 

[Ohashi_Jouzu*]

He posted better numbers after he played himself out of contention for being a starter at the beginning of both Chicago seasons, and his save percentage compared to Puppa and Fuhr was almost negligible given the fact that they barely recorded enough games to be on the official record (Hasek was literally one fewer relief performance away from not counting).

You're arguing with me for the sake of arguing. Hasek's 1992-93 - an injured season of 28 games - is not part of his prime. It's barely an official rookie season. And yes, Roy held a similar edge in GAA over Craig Billington (0.58 in the DPE) seven years after his prime had ended.

And if you keep saying that he was better than Ed Belfour in 1990-91 because he had chunks of three games against three teams that averaged a .500 record with a higher save percentage (.914 to .910) than a 74 GP league leader, then I'll probably stop responding to you.

Right, so you keep on detracting from Hasek because he "didn't win the starting role" in those situations. Why again, exactly?

But let's back up and explore what you mean by "playing himself out of contention". In '90/91, his first appearance was a 28 save 1-1 tie, and his last 3 appearances were partial games in the playoffs, where he stopped 36 of the 39 combined shots he faced (no decisions). Certainly that doesn't qualify.

I think '91/92 is as close as you're going to get. Even then, though, what was he really supposed/expected to do in those appearances? He got plenty of decisions during the regular season, and ended the regular season with two strong appearances. Finally gets called upon again in the playoffs against St. Louis, and in his first "real" game in almost a month keeps his team in a double overtime game. Didn't get the win, though, of course. Then another month later, with Pittsburgh already up 3-0 in the series and looking to complete the sweep at home, Lemieux/Stevens/Tocchet dominated a 6-5 game of balls out offense. Come on, there's little any goalie can really prove/stamp on the game in that little work to usurp the job of a guy playing at Belfour's level, so why the demerit points for Hasek, hmm? I really don't think this qualifies, either.

And then there's '92/93, still basically platooning with Grant Fuhr and Puppa. Posts much better numbers than them over pretty much evenly split appearances. Puppa isn't around to start '93/94 (nor is Draper, the other of the young goalies they tried that year), so obviously Hasek proved something. The team's record over the second half of '93/94 compared to the first, and the statistical difference between Hasek and Fuhr over the course of the season, provide a decent amount of quantitative proof that Hasek was about to be a secret no longer. That epic 70 save battle against Brodeur (and rewardless heroics the next game) in the playoffs pretty much sealed it.

So again, where did you manage to find enough criteria with the weight to cast such a disparaging eye on Hasek's break-in to the NHL?

edit:

And to clarify some of your misunderstandings that I didn't bold and separate, when I say Hasek's "prime" starts in '93, I mean the season that starts in '93 - i.e. '93/94, not '92/93. And for the end points, if I say '01, that typically would mean the season ending in '01 (i.e. '00/01), and applied there, too.

Also, like I said, Roy didn't always distinguish himself statistically from Billington (sample size obviously different magnitudes, but...), and when he did, Marc Denis' brief appearances were negligibly different statistically. Denis had pretty fluffy opponents, but still. Just saying that I think Hasek's backups have had similarly fluffy opponents on his off days, and they still usually failed to post anything remotely similar in terms of stats. I mean, as good as Biron's numbers were in fluffy spot duty in '98/99, his numbers still lagged behind Hasek's to the same degree that, say, Damian Rhodes lagged behind him. That's all I'm saying there. It's not the be-all, end-all of anything, just a small piece that contributes context on a couple of levels ("replacement value", strength/style of team in front of them, etc).

 

[Ohashi_Jouzu*]

Breaking down that data, it doesn't take much to see the disparity in Hasek's numbers after leaving the weaker Sabres. Where Roy's numbers don't have any where near the same kind of disparity after leaving the weaker Habs.

Breaking down the numbers, even though Hasek's numbers may look a lot lower relative to his own numbers, his ranking relative to his peers (in terms of GAA and SV%) compares just fine with those of Roy. The top 10s have already been enumerated in this thread, and the rest is easily checked.

So for me I see it like this...
Is Hasek's peak enough to overcome Roy in the regular season...yes.
Is Hasek's peak enough to overcome Roy in the playoffs...not even remotely.

Wow, you claim to have flip-flopped in the past on this one, and yet now it's "not even remotely". I find that hard to believe - given the amount of information available here and how it seems to stack up. Totally gonna be one of those agree to disagrees when it comes to interpretation of everything we know and saw (of all these first candidates, really).

And honestly, anyone that says that players aren't capable of raising their level of play in playoffs is out to lunch and IMO quite simply haven't played competitive hockey at a high level before. Going that extra mile, sacrificing like you wouldn't do in the regular season, bearing down and upping your level of concentration because of what each game means in the playoffs are all very real.
I mean hell, the entire Oiler teams from the mid 80's proves this fully and completely.[/QUOTE]

 

[Michael Farkas]

did you get that from me? If so, I have a 2012 update now.

https://dl.dropbox.com/u/28241305/Complete Goalie Stats from 52-53 to 82-83 - updated.xls

Maybe a silly question. I see the formula for Column U, but how would you describe what that's measuring exactly? (if you don't have it handy, it's [LgAvgSv%-ActualSv%]*Minutes). That's the thing that caught my eye the most in that impressive spreadsheet.

 

[Rhiessan71]

Breaking down the numbers, even though Hasek's numbers may look a lot lower relative to his own numbers, his ranking relative to his peers (in terms of GAA and SV%) compares just fine with those of Roy. The top 10s have already been enumerated in this thread, and the rest is easily checked.

I honestly don't even know what to say to this. Like you realise my post referred to CG's numbers and I specifically cited the difference in playoff numbers between Hasek's numbers after Buffalo vs Roy's numbers after Montreal right?

Wow, you claim to have flip-flopped in the past on this one, and yet now it's "not even remotely". I find that hard to believe - given the amount of information available here and how it seems to stack up. Totally gonna be one of those agree to disagrees when it comes to interpretation of everything we know and saw (of all these first candidates, really).

Again, I didn't think my post was that confusing yet here you are confused.
All I said there was what I have said since day 1. That I had Hasek over Roy for the regular season, mostly due to peak and it's Roy over Hasek (and everyone else) for the playoffs. The playoffs is where it is not even remotely close and that still stands IMO.
The only thing that has changed for me through this thread is that the gap I had between Hasek and Roy in regular season peak is not as large as I thought it was.
That change is what has me back in the Roy camp and I believe I made that extremely clear at least twice, now three times apparently.

I hope that is clear enough for you.

 

[seventieslord]

Maybe a silly question. I see the formula for Column U, but how would you describe what that's measuring exactly? (if you don't have it handy, it's [LgAvgSv%-ActualSv%]*Minutes). That's the thing that caught my eye the most in that impressive spreadsheet.

Good question! I had forgotten there was anything in there that needs explaining. that is a weighted +/- off the league average. So, if you want to know a goalie's career weighted number of points he was above or below the league average, just sum up all those scores, then divide by career minutes. Also really handy for isolating blocks of seasons (i.e. from 1994-2001 Hasek averaged x points above the league average) or for choosing a player's best 5 (i.e. over Roy's best 5 seasons he averaged x points above the league average) or taking out a few outliers (i.e. aside from these two spikes/years spent on the Islanders, ____ had a career average of x points below/above the league average)

 

[Michael Farkas]

Good question! I had forgotten there was anything in there that needs explaining. that is a weighted +/- off the league average. So, if you want to know a goalie's career weighted number of points he was above or below the league average, just sum up all those scores, then divide by career minutes. Also really handy for isolating blocks of seasons (i.e. from 1994-2001 Hasek averaged x points above the league average) or for choosing a player's best 5 (i.e. over Roy's best 5 seasons he averaged x points above the league average) or taking out a few outliers (i.e. aside from these two spikes/years spent on the Islanders, ____ had a career average of x points below/above the league average)

I like the idea of it, I really do. The first thing I did was sort by that column. As you're already aware or quickly becoming aware, I'm not statistics guy...but would you - or the other math-magicians - say that the metric would be more or less useful, especially for elite goaltenders (such as, but not limited to, this project) if LgAvgSv% was scrapped (or put in a separate category, even better) and LgAvgSSv% (League Average Starter Save Pct.) was introduced. Not that it would have a tremendous effect career-wise, but when taking "best seasons" or whatever, would the original LgAvgSv% category not be influenced by poor expansion team's backup netminder(s), for instance?

Hypothetical: If Roy had actually posted elite numbers in 1993 regular season, he would have gotten the benefit of Jeff Hackett's .856 job with San Jose, Berthiaume's .871 season with Ottawa, Wendel Young's .872 in Tampa and even a non-expansion blunder like Frank Pietrangelo's .858 save pct. (to be fair, he's only remembered for one save, these numbers suggest he made as many as three more!) in Hartford...while Brodeur and Hasek would not...so Roy would have had an accentuated "best season" perhaps for no good reason really...

Does that concept hold any water or am I just talking out of the back of my pants here...?

 

[Bear of Bad News]

Hypothetical: If Roy had actually posted elite numbers in 1993 regular season, he would have gotten the benefit of Jeff Hackett's .856 job with San Jose, Berthiaume's .871 season with Ottawa, Wendel Young's .872 in Tampa and even a non-expansion blunder like Frank Pietrangelo's .858 save pct. (to be fair, he's only remembered for one save, these numbers suggest he made as many as three more!) in Hartford...while Brodeur and Hasek would not...so Roy would have had an accentuated "best season" perhaps for no good reason really...

It's certainly possible.

On the other hand, Hackett, Berthiaume, Young, and Pietrangelo's performances that season give evidence that it was harder to put together an "elite" season that year (as does Roy's performance that year).

 

[seventieslord]

I like the idea of it, I really do. The first thing I did was sort by that column. As you're already aware or quickly becoming aware, I'm not statistics guy...but would you - or the other math-magicians - say that the metric would be more or less useful, especially for elite goaltenders (such as, but not limited to, this project) if LgAvgSv% was scrapped (or put in a separate category, even better) and LgAvgSSv% (League Average Starter Save Pct.) was introduced. Not that it would have a tremendous effect career-wise, but when taking "best seasons" or whatever, would the original LgAvgSv% category not be influenced by poor expansion team's backup netminder(s), for instance?

Hypothetical: If Roy had actually posted elite numbers in 1993 regular season, he would have gotten the benefit of Jeff Hackett's .856 job with San Jose, Berthiaume's .871 season with Ottawa, Wendel Young's .872 in Tampa and even a non-expansion blunder like Frank Pietrangelo's .858 save pct. (to be fair, he's only remembered for one save, these numbers suggest he made as many as three more!) in Hartford...while Brodeur and Hasek would not...so Roy would have had an accentuated "best season" perhaps for no good reason really...

Does that concept hold any water or am I just talking out of the back of my pants here...?

I get your point, but I'm thinking this would probably wash out from year to year, because the percentage of games backups play is more or less a constant, and I think the percentage they lag behind their starters is more or less a constant as well. Only one way to test that out though, I guess.

One other thing that doesn't sit right with me - Sawchuk was often below the league average during the last 2/3 of his career and his career average, I think, is below the league average. But that was the average of the world's 6 best goalies (more or less). I was thinking of making a new "league average" column that uses that "top 6" as a standard for the league average, and extends it out gently over time... like perhaps top-7 throughout the 7s, top-8 in the 80s, top-9 in the 90s, and top-10 through present day. And compare all goalies' save percentages to that average.

Also - not sure why I never posted this file before.

https://dl.dropbox.com/u/28241305/Persev7.xls

Same thing, but for playoffs.

Also includes the "perseverance" stat which I was a believer in thanks to The Hockey Compendium, back when I started it in 2008. You can ignore all that stuff. Most of the stuff on the right, I forget what I was doing now. If you have any questions, though, let me know. That goes for anyone.

 

[Jabroni]

Just a quick question from a non-voter: when will the final voting for Vote 1 be done?

 

[overpass]

I wanted to use the playoff projection method to focus a bit on some of Hasek's other playoff seasons, as some have criticized him for getting injured in 2006 and replaced in 2008. Again, note that equivalent seasons to Hasek's 2008 were not included for Sawchuk, Hall and Plante, and that Plante was the only other guy even playing in the playoffs past the age of 40 (Hasek was 43 in 2008). However, I'll run the numbers anyway.

The other question is how exactly to account for a goalie's absence. The 2006 Sens were expected to win 1.57 series and won 1. The 2008 Red Wings were expected to win 2.20 and won all 4. This suggests that counting the expectation against Hasek in seasons he didn't play or got replaced would substantially overestimate the actual cost to his teams of not having Hasek , considering they both won at least one playoff series and the '08 Red Wings did better than expected (obviously, since they won the Cup). A similar argument can actually be made for '97 as well, even if you assume that Hasek was completely faking injury, given that the Sabres were heavy underdogs against the Flyers and would have been longshots to win the Cup even with Hasek.

I appreciate the value of a systematic approach but I think this method understates the strength of the 2006 Senators. That team was better than their regular season points percentage for several reasons.

1. They had the fewest games go to OT/SO and the fewest OT/SO wins in the regular season, under a set of rules that are not used in the playoffs. They led the East in regulation points by 9.

2. They had the best GF/GA ratio in the league, and the best of any team since the lockout.

3. They had a substantial lead in the East for most of the year until a late-season slide where multiple key defencemen were all out at the same time. And they did that with Marty Havlat in the lineup for most of the season. All these skaters were back for the playoffs.

4. They had a track record of being a very strong team in previous seasons, unlike Buffalo, Carolina, etc. In hindsight we also know they were strong enough to go to the Finals in the following year despite losing Chara and Havlat.

5. Look at the talent on that roster. Alfredsson, Spezza, Heatley, Havlat, Chara, Redden on the top end. Vermette and Kelly on the fourth line. Volchenkov and Pothier on the third pairing. They were very strong top to bottom, with lots of stars and no marginal players.

After Detroit went down in the first round Ottawa was the best team left, healthy outside of the goal crease, and a clear favourite against any remaining opponent. They lost to Buffalo (who were pretty good themselves, I admit) in five close games during which they outshot their opponent but their rookie backup posted a sub-0.900 save percentage. Hasek certainly had an opportunity to make a difference if he was willing to play.

Edit: It would be interesting to see the Vegas odds for Ottawa winning the Cup that year. Anyone know if it's possible to find that?

 

[overpass]

Another question about your playoff results vs expectations, CG. Presumably the 1967 Toronto win over Chicago was rated as a major upset. Did you include this series for both Sawchuk and Hall? Neither played the full series.

 

[TheDevilMadeMe]

Just a quick question from a non-voter: when will the final voting for Vote 1 be done?

Votes are due tonight. I didn't give an exact time, so I guess the voters have until midnight EST to get them in.

 

[Master_Of_Districts]

I appreciate the value of a systematic approach but I think this method understates the strength of the 2006 Senators. That team was better than their regular season points percentage for several reasons.

1. They had the fewest games go to OT/SO and the fewest OT/SO wins in the regular season, under a set of rules that are not used in the playoffs. They led the East in regulation points by 9.

2. They had the best GF/GA ratio in the league, and the best of any team since the lockout.

3. They had a substantial lead in the East for most of the year until a late-season slide where multiple key defencemen were all out at the same time. And they did that with Marty Havlat in the lineup for most of the season. All these skaters were back for the playoffs.

4. They had a track record of being a very strong team in previous seasons, unlike Buffalo, Carolina, etc. In hindsight we also know they were strong enough to go to the Finals in the following year despite losing Chara and Havlat.

5. Look at the talent on that roster. Alfredsson, Spezza, Heatley, Havlat, Chara, Redden on the top end. Vermette and Kelly on the fourth line. Volchenkov and Pothier on the third pairing. They were very strong top to bottom, with lots of stars and no marginal players.

After Detroit went down in the first round Ottawa was the best team left, healthy outside of the goal crease, and a clear favourite against any remaining opponent. They lost to Buffalo (who were pretty good themselves, I admit) in five close games during which they outshot their opponent but their rookie backup posted a sub-0.900 save percentage. Hasek certainly had an opportunity to make a difference if he was willing to play.

Edit: It would be interesting to see the Vegas odds for Ottawa winning the Cup that year. Anyone know if it's possible to find that?

Not sure about the Vegas Odds, but I had them at a 38% shot (if win probabilities are based on regular season goal ratio) or a 25.5% shot (if a bayesian approach is used), both of which figures are really impressive, given that a very good team in a typical year might have a 15% shot.

 

[Ohashi_Jouzu*]

Edit: It would be interesting to see the Vegas odds for Ottawa winning the Cup that year. Anyone know if it's possible to find that?

Not sure if officially "Vegas odds" or not, but: something to look at. Ottawa 3rd spot at 11-1.

 

[pluppe]

No, it means he didn't record enough games to have had a particularly large impact in seven NHL seasons since he first came to North America. Plus an extra season in which he was retired. You're acting as though he didn't have the chance to not be an injury-prone player.

I don´t know, did he?

Which strikes me as especially funny, because it's the exact same gap that you refused to believe could exist between Roy and Hasek in their respective quality games percentage when trailing in a playoff series.

And what strikes me as funny (though I am not surprised) is that you don´t see the difference between 360 consecutive games and 28 biasely selected games.

I have never said that the numbers don´t exist. But it´s funny that you pull them out again but twice has avoided (unless I missed something in which case I apologize) to respond when I questioned their relevance.

Originally Posted by quoipourquoi
I'm not sure how Hasek's consistency when he's in an equal or advantageous position in a series is a good counter for an argument that he struggles when faced with adversity (competition at his position, playing with injuries), but sure. Subtracting the trailing numbers from the total numbers gives us this:

Roy (Leading/Tied): 134-67; 66.7%
Roy (Trailing): 35-10; 77.7%
Roy (Difference): +10%

Hasek (Leading/Tied): 64-26; 71.1%
Hasek (Trailing): 15-13; 53.6%
Hasek (Difference): -18%


So, as you can see, Hasek's consistency falls significantly when his team is in a bad position, whereas Roy's consistency when down in a series offers his team the best chance to get back into the series.

Originally Posted by pluppe
I think you are very biased here. Your way is one way of looking at it but couldn´t somebody claim something like "Winning the first game or a game 7 is the most important thing in the playoffs so performing when tying is the best quality in a goaltender. That Hasek performed worse when trailing probably has to do with that Buffalo as a team realized they were not good enough to overcome the situation, gave up and did not support their goalie."

And if you believe that Roy actually lifted his performance +10% when trailing you have to answer why he did not do this in the other playoff games? Do you really want a goalie who lacks the motivation to give it his best when tying a series in the most important time of the year.

Now I want to point out that I don´t believe in these statements. I´m just showing that it is very easy to choose arguments that suits your agenda.

In this case I am not surprised that the numbers vary so much when it is such a small sample. I find it highly unrealistic that these numbers shows such clear clutchness when larger studies are having problems finding it at all. That is a hot topic and I am not actually sure what I believe but I am sure that if clutchness exists you are overrestimating it in this example because I find it very hard to believe that a goalie that has an amazing "cluthgap" of a staggering +24% when trailing suddenly are -4% in all other games in the playoffs where every game is a high pressure game. A 28% clutch swing. That would be the most selective clutch ability I have ever seen. In this case random variance as part of the explanation is a much better hypothesis in my eyes.

I would like to say though that I would guess there are some truth to the numbers. I think Roy might have had a little better resilience when down. I think Hasek might have been a little more at risk of breaking down and I would guess that on the whole these differences are small and that they are very close in the playoffs with an edge to Roy because of his slightly better stats over a larger sample that resulted in a phenomenal trophy case. This is what the numbers also tell me. Just not to the simplified extent your very selective analysis claims.

If Hasek's European career was his prime as well, then he has little excuse for not winning a job in 1990-91, 1991-92, or distancing himself from Puppa in 1992-93.

Yes, how wonderful if the world was that simple.

 

[ContrarianGoaltender]

I appreciate the value of a systematic approach but I think this method understates the strength of the 2006 Senators. That team was better than their regular season points percentage for several reasons.

I largely agree with your assessment. At least one sportsbook had them at 10-3 to win the Cup heading into the playoffs, an implied Cup probability of 23%, compared to my number of 16%. I don't think these types of examples necessarily invalidate the usefulness of the overall method, but it would certainly be useful to make adjustments where necessary.

Another question about your playoff results vs expectations, CG. Presumably the 1967 Toronto win over Chicago was rated as a major upset. Did you include this series for both Sawchuk and Hall? Neither played the full series.

Yes, I credited that one to both Sawchuk and Hall. I'd be open to suggestions about how to handle partial series like that because they aren't always straightforward, and will be even less so as we move down the list to get more goalies who weren't always the clear #1 starters.