Playoff point allocation (1980-2017)

[Hockey Outsider]

Over the weekend I spent some time tackling a project that I've wanted to do for years - applying Iain Fyffe's "Point Allocation" methodology to the NHL playoffs (from 1980 to 2017).

The purpose of the method is to allocate points to each player on a team, in accordance with who helped generate them. (Since this is the playoffs, I've decided to allocate wins; points in the standings are obviously a regular season concept). Iain's method (which is in turn borrowed from baseball writer Bill James's "Win Shares" system) can be read here - https://web.archive.org/web/20030705023433/http://www.puckerings.com:80/research/ptalloc.html.
The most significant departure from his approach is I allocate defensive points based on estimated even-strength ice time. Ultimately this method ranks all players, regardless of their position or what season they played in, based on a standard currency - wins.

I spot-checked some of the results to make sure the system "works". For example, if you add up all of the player scores on the 2002 Detroit Red Wings, you'd get exactly 16 wins. If you add up all of the scores for every single player who's played for the Washington Capitals, you'd get exactly 116 (the number of playoff games the franchise has won - prior to 2018).

Some might ask - isn't this the same thing as hockey-reference.com's widely-criticized "Point Shares"? Conceptually there are similarities, but the biggest issue with that system is they fundamentally don't understand how to evaluate defensive play. This system is far from perfect (see my list of self-criticisms), but it has a much better starting point. (I also have consistent data going back to 1980 - sometimes that website's results are wonky because they have to make so many simplifying assumptions for the years prior to expansion that the distribution gets flattened).

For the most part, the results make sense. It's always tough to create a purely statistical system (since obviously the computer doesn't care about a player's name or reputation). Bill James once said something to the effect of "If a model is never surprising, it's probably not useful. If a model is consistently surprising, it's probably wrong".

Click to read more...

 

[Hockey Outsider]

TOP HUNDRED PLAYERS - PLAYOFF POINT ALLOCATION (1980-2018)

1	Wayne Gretzky*	16.04
2	Patrick Roy*	15.61
3	Mark Messier*	13.84
4	Martin Brodeur	12.20
5	Nicklas Lidstrom*	11.70
6	Jari Kurri*	11.38
7	Glenn Anderson*	10.59
8	Joe Sakic*	10.10
9	Paul Coffey*	9.89
10	Chris Chelios*	9.79
11	Brett Hull*	9.61
12	Jaromir Jagr	9.55
13	Ed Belfour*	9.38
14	Claude Lemieux	9.23
15	Steve Yzerman*	9.19
16	Sidney Crosby	9.11
17	Peter Forsberg*	9.07
18	Scott Stevens*	8.94
19	Larry Murphy*	8.88
20	Ray Bourque*	8.73
21	Sergei Fedorov*	8.45
22	Marian Hossa	8.31
23	Grant Fuhr*	8.16
24	Evgeni Malkin	8.13
25	Mike Modano*	8.02
26	Doug Gilmour*	8.02
27	Scott Niedermayer*	8.01
28	Mark Recchi	7.98
29	Mario Lemieux*	7.92
30	Bryan Trottier*	7.87
31	Chris Pronger*	7.79
32	Bobby Smith	7.48
33	Al MacInnis*	7.47
34	Esa Tikkanen	7.33
35	Patrick Marleau	7.28
36	Marc-Andre Fleury	7.24
37	Brendan Shanahan*	7.22
38	Mike Vernon	7.18
39	Patrick Kane	6.84
40	Henrik Lundqvist	6.81
41	Sergei Zubov	6.78
42	Curtis Joseph	6.77
43	Ron Francis*	6.65
44	Chris Osgood	6.63
45	Rod Brind'Amour	6.63
46	Billy Smith*	6.63
47	Andy Moog	6.62
48	Brian Propp	6.57
49	Dominik Hasek*	6.56
50	Denis Savard*	6.52
51	Brian Rafalski	6.51
52	Adam Oates*	6.47
53	Patrik Elias	6.35
54	Ryan Getzlaf	6.31
55	Guy Carbonneau	6.22
56	Joe Nieuwendyk*	6.21
57	Mike Bossy*	6.21
58	Henrik Zetterberg	6.14
59	Jonathan Toews	6.05
60	Charlie Huddy	6.03
61	Brad Richards	5.99
62	Denis Potvin*	5.95
63	Joe Thornton	5.94
64	Jeremy Roenick	5.94
65	Steve Thomas	5.91
66	Alex Ovechkin	5.80
67	Chris Drury	5.79
68	Eric Desjardins	5.77
69	Rick Tocchet	5.76
70	Sandis Ozolinsh	5.73
71	Daniel Briere	5.73
72	Brian Bellows	5.72
73	Tom Barrasso	5.70
74	Justin Williams	5.59
75	Tomas Holmstrom	5.57
76	John Tonelli	5.56
77	Patrick Sharp	5.55
78	Zdeno Chara	5.49
79	Luc Robitaille*	5.48
80	Pavel Datsyuk	5.48
81	Igor Larionov*	5.44
82	John LeClair	5.43
83	Steve Larmer	5.43
84	Duncan Keith	5.42
85	Joe Mullen*	5.41
86	Chris Kunitz	5.38
87	Corey Perry	5.38
88	Vincent Damphousse	5.35
89	Kevin Stevens	5.32
90	Larry Robinson*	5.29
91	Jamie Langenbrunner	5.25
92	Petr Sykora	5.22
93	Dino Ciccarelli*	5.19
94	Mike Keane	5.16
95	Martin St. Louis	5.08
96	Valtteri Filppula	5.07
97	Kris Letang	5.03
98	Sergei Gonchar	5.02
99	Kevin Lowe	4.99
100	Daniel Alfredsson	4.98

[TBODY] [/TBODY]

Remember, the system isn't saying that Glenn Anderson is a better playoff performer than Joe Sakic (or Sergei Fedorov, or Mario Lemieux). But the system is a career-based metric, so it rewards longevity. Even though he was only a distant fourth (among forwards) on the Oilers dynasty, the system reads his contributions as significant given the amount of offense he generated, and the amount of ice time he received.

Ovechkin makes a huge leap on this list thanks to his 2018 campaign. He jumps from 128th to 66th due to a Smythe calibre run.

Given that he lacks a "signature" playoff run, Jagr fares remarkably well.

Since this system allocates team victories, you need to contribute to winning teams to earn points. The only player on the top fifty without a Cup is Propp.

 

[Hockey Outsider]

Table removed - no longer relevant

 

[Hockey Outsider]

Here are the top ten single seasons by position:

• Forwards - Gretzky 1985, Gretzky 1988, Malkin 2009, Lemieux 1991, Sakic 1996, Bure 1994, Richards 2004, Lemieux 1992, Gretzky 1984, St. Louis 2004.
• Defense - Leetch 1994, Coffey 1985, Keith 2015, Pronger 2006, Lidstrom 1998, Doughty 2012, MacInnis 1989, Niedermayer 2003, Blake 2001, Coffey 1984.
• Goalie - Roy 1986, Brodeur 1995, Roy 1993, Giguere 2003, Belfour 1999, Quick 2012, Brodeur 2003, Vernon 1997, Brodeur 2000, and Luongo 2011.

Top twenty five-years peaks (not consecutive):

• Gretzky, Roy, Brodeur, Messier, Kurri, Sakic, Lemieux, Belfour, Crosby, Malkin, Smith, Kane, Fleury, Anderson, Vernon, Trottier, Hull, Lidstrom, Forsberg, Fedorov

Here are the ten players who have contributed the most to a single franchise:

• Brodeur NJD, Lidstrom DET, Gretzky EDM, Sakic COL/QUE, Kurri EDM, Messier EDM, Yzerman DET, Crosby PIT, Anderson EDM, Forsberg COL/QUE.

Best playoffs from players who didn't win the Stanley Cup:

• Giguere 2003, Bure 1994, Luongo 2011, Hull 2000, Gretzky 1993, Iginla 2004, Belfour 2000, Hextall 1987, Modano 2000, Lindros 1997

Best playoffs from players who didn't win the Conn Smythe:

• Brodeur 1995, Bure 1994, Gretzky 1984, Coffey 1985, St. Louis 2004, Belfour 1999, Crosby 2009, Brodeur 2003, Messier 1994, Brodeur 2000

 

[Hockey Outsider]

Let me be the first to offer some self-criticisms of this method:

• Point Allocation implicitly assumes that each team has a schedule that's equally difficult. Even in the regular season, that assumption is wrong, but it's probably close enough to being true that it doesn't materially affect the results. In the playoffs, when half the teams only face a single opponent before bowing out, that assumption is clearly false.
• This method can be merciless to players who play well on bad teams. For example, in 1991, the Bruins were swept in the first round. Thus, there were no points to allocate. Tough luck for Adam Oates, who assisted on nine of his team's twelve goals - he got nothing.
• On the very few occasions when a team's offense operates at a sub-marginal level, the results get wonky. Adam Oates (again!) gets penalized because he was the leading scorer for the 2002 Flyers, which was the worst offensive team dating back to 1980 (two goals in five games). The system allocates a greater proportion of the team's negative score to Oates - counter-intuitively, he would have helped his Point Allocation had he simply not scored at all! (This isn't a major concern as only a few teams have ever scored at a sub-marginal rate).
• My approach for allocating points to defense is different than Iain's (both have their own weaknesses). Essentially I allocate a team's overall defensive points to skaters in proportion to their estimated even-strength ice time. This means 1) penalty killing is effectively ignored and 2) this approach essentially assumes that all teammates are roughly equally good defensively (on a per-minute basis). Obviously these are weaknesses in the approach - but I still think the end results look reasonable. (The other issue is I use estimated ES ice time, even in years when actual ice time was recorded - it was simply too challenging a task to try to merge different data sets with different parameters).
• This is a subjective observation, but I don't think there's enough spread in the points allocated to goalies on the same team. This isn't a major problem, as most teams ride one goalie for the playoffs, but there are exceptions. (For example, it looks like Hasek earned too many points in 2008, and Osgood not enough).
• Prior to 1987, the first round of the playoffs was a best-of-five. Thus, teams won the Stanley Cup with fifteen victories (rather than the sixteen needed today). I haven't attempted to correct for this - so the members of the Islanders dynasty (and other strong teams from the early to mid eighties, including the Oilers) are at a slight disadvantage.
• I haven't subtracted a team's own statistics when using league averages. In a regular season with 31 teams, that's probably not necessary. In the playoffs, when the winning team might play in around 15% of the total games, an adjustment might be significant.
• There are no adjustments for overtime (goals and shots are calculated on a per-game basis, rather than a per-minute basis, which would probably be more accurate). A team that spent a lot of time in overtime will have too many of their wins allocated to offense, and not enough allocated to defense & goaltending (since goals for & goals against are calculated on a per-game basis, rather than per-minute).
• I'm not sure that I have the right split between defense and goaltending. (More accurately, I think I have the right split overall, but not enough variance between individual teams). For example, the system reads the 2003 Ducks as having both brilliant goaltending and defense. In reality they were a solid defensive team but it really was the goaltending that carried them. Giguere's performance is therefore understated, and the defensive scores of all the skaters are overstated.

Are these issues, individually or in the aggregate, serious enough to render this project meaningless? The thought crossed my mind - but I figure it's better to share the results so at least it starts a conversation.

 

[bathdog]

Interesting, thanks.

Are these issues, individually or in the aggregate, serious enough to render this project meaningless? The thought crossed my mind - but I figure it's better to share the results so at least it starts a conversation.

Can't get away from flaws, unique methodologies that present reasonable results are almost always interesting. I think it was summed up pretty well in the quote in the first post.

If a model is never surprising, it's probably not useful.

 

[The Panther]

Best playoffs from players who didn't win the Stanley Cup:

• Giguere 2003, Bure 1994, Luongo 2011, Hull 2000, Gretzky 1993, Iginla 2004, Belfour 2000, Hextall 1987, Modano 2000, Lindros 1997

Best playoffs from players who didn't win the Conn Smythe:

• Brodeur 1995, Bure 1994, Gretzky 1984, Coffey 1985, St. Louis 2004, Belfour 1999, Crosby 2009, Brodeur 2003, Messier 1994, Brodeur 2000

No Gretzky in 1983 for either?

I was pleased to see Bure '94 rated so high on "best players who didn't win Cup" list, though. That incredible playoff by him tends to get passed over too quickly, I think. Full marks to Leetch, but by strict "mvp" wording I think Bure was the most valuable player that spring.

 

[Canadiens1958]

TOP TWENTY-FIVE GOALIES - PLAYOFF POINT ALLOCATION (1980-2017)

Rank	Player	PA
1	Patrick Roy*	45.20
2	Martin Brodeur	35.38
3	Ed Belfour*	27.15
4	Grant Fuhr*	23.66
5	Mike Vernon	20.86
6	Henrik Lundqvist	19.73
7	Curtis Joseph	19.57
8	Chris Osgood	19.28
9	Andy Moog	19.22
10	Billy Smith*	19.20
11	Dominik Hasek*	18.98
12	Marc-Andre Fleury	16.61
13	Tom Barrasso	16.53
14	Ron Hextall	13.92
15	Jonathan Quick	13.34
16	Corey Crawford	12.84
17	Evgeni Nabokov	12.43
18	Felix Potvin	11.54
19	Mike Richter	11.36
20	Pekka Rinne	10.71
21	Roberto Luongo	10.54
22	Braden Holtby	10.40
23	Jean-Sebastien Giguere	10.38
24	Nikolai Khabibulin	10.26
25	Antti Niemi	9.91

[TBODY] [/TBODY]

I decided to show goalies separately. Over the course of a single season, a truly dominant skater can match the value of an excellent goalie. But goalies generally earn more Points/Wins Allocated over the course of their careers simply due to how much ice time they receive. (In other words - given the importance of their position, a goalie can achieve a great career score by being above average, or even just average, for a long time - the same isn't true for forwards or defensmen).

Vernon,Osgood and Moog are surprising.

 

[quoipourquoi]

Based on Martin Brodeur in 1995, I'm assuming it is reflecting the field of goaltenders in the 1995 playoffs as opposed to the league average expectation? Not saying one is necessarily better than the other, especially for point allocation, but he was always my red flag in that he has the best run by the playoff standard and is #15 of the four-round era by the regular season standard.

As always, great stuff!

 

[Hockey Outsider]

No Gretzky in 1983 for either?

I was pleased to see Bure '94 rated so high on "best players who didn't win Cup" list, though. That incredible playoff by him tends to get passed over too quickly, I think. Full marks to Leetch, but by strict "mvp" wording I think Bure was the most valuable player that spring.

Gretzky's 1983 playoffs weren't in the top ten among those who didn't win the Cup, but it was in the top twenty. His team "only" won 11 playoff games, so there weren't enough wins to go around to propel him higher (though obviously he was very important to his team).

Vernon,Osgood and Moog are surprising.

They surprised me too at first. I'm not saying they didn't have their moments, but they're generally considered average goalies on strong teams. Some systems look at how a goalie performs compared to the average - in which case, these goalies would have scores near zero. This approach recognizes the value in playing, even at a league-average level, over long periods of time.

Based on Martin Brodeur in 1995, I'm assuming it is reflecting the field of goaltenders in the 1995 playoffs as opposed to the league average expectation? Not saying one is necessarily better than the other, especially for point allocation, but he was always my red flag in that he has the best run by the playoff standard and is #15 of the four-round era by the regular season standard.

As always, great stuff!

Thanks! Yes, the save percentage benchmarks are based on the playoff average, rather than the regular season average. (This was inevitable since I wanted the system to "balance" - ie every player's score had to add up to the team's wins in each postseason - so only playoff results were included).

 

[ImporterExporter]

These studies never cease to amaze me. Great job HO, as always and thank you for taking the time to put these together.

 

[ImporterExporter]

I'd also like to say, yawn, yet another study Patrick Roy utterly dominates.

Nice to see MAF on their as well. He often gets criticized far to much by the Pens main board folks for 2010-2013 but outside of those horrendous years (I personally put a big chunk of the blame on Bylsma and Shero) he's been a very reliable and sometimes flat out amazing goalie in the postseason. And I have to think if his (and Vegas) magical run continues he'll climb the list even further.

 

[Canadiens1958]

They surprised me too at first. I'm not saying they didn't have their moments, but they're generally considered average goalies on strong teams. Some systems look at how a goalie performs compared to the average - in which case, these goalies would have scores near zero. This approach recognizes the value in playing, even at a league-average level, over long periods of time.

Average very often simply does not exist. This season per HR an average team had 92 points. th 95 did Three teamswith 92,94,96 points did not make the playoffs. One with 95 points did.

Sense we have growing evidence that RS norms for goalies do not apply in the playoffs and this pre dates 1980.

 

[Hockey Outsider]

These studies never cease to amaze me. Great job HO, as always and thank you for taking the time to put these together.

Thanks! I've been sick with the flu all week (not fun this time of year), and I figured this was a semi-productive way to spend the time.

I'd also like to say, yawn, yet another study Patrick Roy utterly dominates.

A study that shows that Roy was the best playoff goalie of the past 40 years is about as interesting as a study that says that Gretzky was a good playmaker.

Nice to see MAF on their as well. He often gets criticized far to much by the Pens main board folks for 2010-2013 but outside of those horrendous years (I personally put a big chunk of the blame on Bylsma and Shero) he's been a very reliable and sometimes flat out amazing goalie in the postseason. And I have to think if his (and Vegas) magical run continues he'll climb the list even further.

The interesting thing about the playoffs is any failures are self-limiting. By that I mean, even if a goalie (or skater) plays catastrophically poorly, they lose four games and then they're out. The regular season is tougher in the sense that no matter how bad a team is, it has to play a full schedule.

 

[ImporterExporter]

Thanks! I've been sick with the flu all week (not fun this time of year), and I figured this was a semi-productive way to spend the time.

A study that shows that Roy was the best playoff goalie of the past 40 years is about as interesting as a study that says that Gretzky was a good playmaker.

The interesting thing about the playoffs is any failures are self-limiting. By that I mean, even if a goalie (or skater) plays catastrophically poorly, they lose four games and then they're out. The regular season is tougher in the sense that no matter how bad a team is, it has to play a full schedule.

Darn, hope you feel better soon.

How much of a gap do you think there really is between Roy and Gretzky all time as playoff performers?

Obviously Gretzky's raw numbers are insane. Adjusted numbers are also ridiculous but I've often wondered if Roy isn't 99's equal in the playoffs. I mean in all reality I highly doubt if either Montreal team (86/93) wins the title or even gets that close to one if Roy isn't standing on his head time and time again. Colorado was a different story in that those Avs teams were simply more talented (it's not like Roy wasn't great to amazing then either) than most.

Roy has the 3 Smythe's and almost surely would have gotten a 4th had the Habs not lost to Calgary in 89. One can certainly argue 4 or even 5 for Gretzky as well if things fell slightly differently. Just about any metric you/overpass/Q has done over the years puts Roy so far and away above other goalies from the past 4 decades in the postseason it really does mirror what Gretzky did among skaters.

 

[quoipourquoi]

Can’t speak for HO, but from my 4-round evaluations (EvE, scoring adjusted to opposition strength), Roy is dominant relative to any one other goaltender, whereas Gretzky is dominant relative to the field as a whole.

I credit it to the nature of the positions in that there is no cap to what a skater can produce, whereas goaltending by nature starts at perfection and is a constant push and pull to see who can hold ground the best. It’s less likely for a skater to get hot and produce like Gretzky than it is for a goaltender to get hot and stop at the same rate as Roy. The latter happens all of the time and even from non HOFers (think Giguere, Quick, Lindbergh, Vanbiesbrouck), but few are going to rattle off 35-40 points regardless of opponent strength and GP.

That five of the top-20 EvE four-round playoff runs belong to Roy while no one other goaltender appears more than once indicates that while his top-level isn’t necessarily better than everyone (though his 1993 does rank 1st), the frequency by which he can reach it is unmatched in the era from the goaltending position. For Gretzky, I believe he had the very best 4-5 runs followed by multiples from the usual suspects: Lemieux, Forsberg, Sakic, etc. So skaters are having multiple great all-time runs, just not ones as productive as Gretzky’s.

In essence, anyone can be Patrick Roy once, but no one can be Wayne Gretzky through a ~16-28 game sample size in the era. However, because anyone can be Patrick Roy once, what does it say that so few have done it twice in this era? Harder to distinguish yourself as a goaltender than as a skater?

Another consideration: consecutive years of success vs. success spread out over time.

Razor thin gap. I’ll take a random year of Gretzky over a random year of Roy, but a random 5 or 10 years of Roy over a random 5 or 10 years of Gretzky. It’s harder to lose with hot goaltending than it is with one hot scoring line. And in retrospect, the 100th anniversary of the Stanley Cup: people were thinking Gretzky vs. Lemieux or Montreal vs. Toronto, but unless Maurice Richard is available, Wayne Gretzky and Patrick Roy might be the two most appropriate players for that series.

 

[Zuluss]

Can we have a similar thing done on a per-game or per total TOI basis?
In addition to punishing players that do well on bad teams (which the per-game/TOI adjustment will not fix), the system seems to be rewarding folks for just sticking around on good (or at least PO-bound) teams long enough and playing a lot of PO games.
Per-game ranking would not probably slot Mario Lemieux between Crosby and Hossa.

 

[Hockey Outsider]

How much of a gap do you think there really is between Roy and Gretzky all time as playoff performers?

I agree with quoipourquoi's reply. In terms of career value, they were probably equal. More players approached Roy's level of performance over a single playoff run, but I think that's largely due to the nature of the position.

Can we have a similar thing done on a per-game or per total TOI basis?
In addition to punishing players that do well on bad teams (which the per-game/TOI adjustment will not fix), the system seems to be rewarding folks for just sticking around on good (or at least PO-bound) teams long enough and playing a lot of PO games.
Per-game ranking would not probably slot Mario Lemieux between Crosby and Hossa.

The top twenty-five skaters on a per-game basis (minimum 50 games) are:

DELETED - old version of system used

Of course the downside with this metric is players who hang around past their prime hurt their scores. For example, the system reads Bryan Trottier as a slightly better playoff performer than Mike Bossy from 1980 to 1987. Bossy retired due to injuries, while Trottier went on to be a useful depth player on Lemieux's Penguins. So it's literally true that Bossy was better on a per-game basis, but it's also misleading.

 

[Yurog]

Claude Lemieux (0.032) is not in top 25

 

[Czech Your Math]

Players' individual plus-minus data aren't being used in the equation?

 

[The Panther]

...in retrospect, the 100th anniversary of the Stanley Cup: people were thinking Gretzky vs. Lemieux or Montreal vs. Toronto, but unless Maurice Richard is available, Wayne Gretzky and Patrick Roy might be the two most appropriate players for that series.

Thinking back to that '93 Final, if Patrick Roy and Kelly Hrudey had switched teams, it might have gone down to overtime in game 7...

 

[quoipourquoi]

Thinking back to that '93 Final, if Patrick Roy and Kelly Hrudey had switched teams, it might have gone down to overtime in game 7...

Wouldn’t that be something? Especially if Montreal found a way to win against a team with both?

Completely unscientific: Hrudey allowed just 69.5% of what would be expected to be allowed on 177 shots by Montreal; Roy allowed 60.1% of what would be expected to be allowed on 155 shots by Los Angeles. Now assuming their EvE remains the same but their rates are applied against each other’s expectation:

LA Expectation: 155 shots x 11.8% = 18.35 goals
MTL Expectation: 177 shots x 12.2% = 21.62 goals

Hrudey Prevention: 69.5% x 18.35 = 12.75 GA
Roy Prevention: 60.1% x 21.62 = 12.99 GA

It would definitely depend on distribution, but it sure looks like it would be an extremely tight series. Hrudey had a great Final, even if the rest of his playoff run was less memorable. Funny that Montreal ended up playing all three 1987 Canada Cup goaltenders in those playoffs - not that Roy was crossing names off of a list like Arya Stark or anything...

 

[Hockey Outsider]

Players' individual plus-minus data aren't being used in the equation?

That’s correct. It’s based on estimated ice time rather than plus/minus.

 

[ImporterExporter]

Can’t speak for HO, but from my 4-round evaluations (EvE, scoring adjusted to opposition strength), Roy is dominant relative to any one other goaltender, whereas Gretzky is dominant relative to the field as a whole.

I credit it to the nature of the positions in that there is no cap to what a skater can produce, whereas goaltending by nature starts at perfection and is a constant push and pull to see who can hold ground the best. It’s less likely for a skater to get hot and produce like Gretzky than it is for a goaltender to get hot and stop at the same rate as Roy. The latter happens all of the time and even from non HOFers (think Giguere, Quick, Lindbergh, Vanbiesbrouck), but few are going to rattle off 35-40 points regardless of opponent strength and GP.

That five of the top-20 EvE four-round playoff runs belong to Roy while no one other goaltender appears more than once indicates that while his top-level isn’t necessarily better than everyone (though his 1993 does rank 1st), the frequency by which he can reach it is unmatched in the era from the goaltending position. For Gretzky, I believe he had the very best 4-5 runs followed by multiples from the usual suspects: Lemieux, Forsberg, Sakic, etc. So skaters are having multiple great all-time runs, just not ones as productive as Gretzky’s.

In essence, anyone can be Patrick Roy once, but no one can be Wayne Gretzky through a ~16-28 game sample size in the era. However, because anyone can be Patrick Roy once, what does it say that so few have done it twice in this era? Harder to distinguish yourself as a goaltender than as a skater?

Another consideration: consecutive years of success vs. success spread out over time.

Razor thin gap. I’ll take a random year of Gretzky over a random year of Roy, but a random 5 or 10 years of Roy over a random 5 or 10 years of Gretzky. It’s harder to lose with hot goaltending than it is with one hot scoring line. And in retrospect, the 100th anniversary of the Stanley Cup: people were thinking Gretzky vs. Lemieux or Montreal vs. Toronto, but unless Maurice Richard is available, Wayne Gretzky and Patrick Roy might be the two most appropriate players for that series.

As usual, great rundown Q.

To the bolded, this is something I've been arguing here and especially over in the ATD for quite some time. No one position is on more of an island and can tilt a series (one way or the other) than the netminder. Just my humble opinion of course.

 

[Czech Your Math]

That’s correct. It’s based on estimated ice time rather than plus/minus.

I think that's an important piece of the puzzle.