Top Ten goalscorers of all-time

[Zuluss]

But that's where dominance and competition come into play. If we look at GPG and compare to the 15th place as a baseline (minimum 41 games for everyone including Lemieux), we get them leading by these percentages:

Lemieux:
96.5, 82.5, 70.7, 65.3, 62.3, 49.2, 38.2, 37.5, 32.7, 7.0, 3.0

Ovechkin:
75.6, 67.5, 60.5, 58.5, 54.8, 53.7, 47.9,
28.0, 14.3, 11.4

I think it's pretty clear that when healthy, Lemieux was a more prolific goalscorer. Ovechkin is arguably a greater goalscorer, but that's based on health

Maintaining the pace for the whole season is the trick.
Ovechkin had a fair number of hot starts (e.g., 41 goals in 53 games in 09/10 and going about goal-per-game in the first half of 13/14). Had he been injured after those hot starts, his gpg numbers would have been up there with Lemieux'. But he stayed healthy, and now the gpg comparison is punishing him for that.

 

[Blade Paradigm]

*AS OF December 11, 2017, the system outlined in this post has been replaced by a newer system. This post is here for archival purposes. Refer to Post #81 for the new Adjusted % Lead Over 10th system.*

I don't want to post all of this data by such gradual means, but I have calculated the conversion multiplier to convert 1917-1923 era numbers to their 1997-2017 equivalents for % Lead Over 10th.

Percentage Lead Over 10th:

The conversion multipliers (with more to be added to this list shortly):
(1917-18 to 1922-23 -> 1997-98 to 2016-17): 0.32628
(1923-24 to 1940-41 -> 1997-98 to 2016-17): 0.62218
(1941-42 to 1946-47 -> 1997-98 to 2016-17): 0.64499
(1947-48 to 1969-70 -> 1997-98 to 2016-17): 0.609415

Using the 1917-1923 multiplier, we can, for example, calculate Joe Malone's career. We also have the multiplier for the era encapsulating the careers of Charlie Conacher and Howie Morenz. Some changes are significant, such as Maurice Richard's 1944-45 season; with the more precise multiplier, his peak is defined as a 76% lead (whereas, in the preliminary model it was 70%). Bobby Hull's peak season, 1966-67, is a 66% lead over 10th (whereas, in the preliminary model it was 65%). Howe's peak, his 1952-53 season, is an 88% lead over 10th (whereas, in the preliminary model it was 87%).

Percentage Lead Over 10th:

Joe Malone: 78-34-24-13
Charlie Conacher: 50-49-48-34-27
Howie Morenz: 52-41-40-35-27-25-19-16-11

*this list is ongoing

Heatley: 25-25-11-11-3-0
Iginla: 41-26-25-24
Patrick Kane: 39-10-0
Stamkos: 67-46-38-32-30-9
Ovechkin: 63-61-52-52-50-44-43-30-15-6.

More to come.

P.S. I would also like to change the baseline for above-average % Lead Over 10th seasons to 46%. Using exact numbers to the 12th and 13th decimal places, the average % Lead Over 10th in the 1997-2017 era is actually 45.58217041189178%.

Based on my calculation of the 1997-2017 average % lead, I think we have determined a source of error in the original % Lead Over 10th model. During the original calculations, it is possible that too much rounding of the figures was done. The more rounded the figures are during the calculation process, the less precise the resultant numbers are.

While I think that the original model provides a great basis for comparison as a source of evidence, a more precise model would most certainly alter the % lead figures by a few percentage points. This has significant implications. The conversion mulipliers for the years between 1941-42 and 1996-97 should be redone. I would be glad to do this.

Although additional work would be required to develop a revised % Lead Over 10th model, I strongly feel that this should be done for the sake of precision.

As of right now, I can say that the raw 1997-2017 percentage leads for both the % Lead Over 10th and % Lead Over 15th models are sufficiently precise because they require no conversion multiplier. Additionally, the 1970-71 to 1996-97 conversion multiplier for the % Lead Over 15th model presented on Page 2 of this thread, and the 1917-18 to 1922-23 conversion multiplier for the % Lead Over 10th model presented in this post are sufficiently precise, calculated from numbers rounded to the 12th and 13th decimal places.

% Leads vs 15th-Ranked Player:

Highlighted in green are the finishes that were at or above the 55% baseline (the average % lead in the modern era of the #1 ranked player over the 15th ranked player):

Bure: 71-65-55-35-19-0
Ovechkin: 86-71-65-60-60-56-52-37-24-12-0-0
Brett Hull: 84-58-57-30-20-13-10-3-3-0
Bossy: 70-44-35-30-30-27-25-19-16
Gretzky: 86-83-53-44-42-21-17-16-14-0

*ongoing updates to this list with figures from more players:

Selanne: 68-45-32-31-30-5
Stamkos: 76-55-45-41-39-13
Lafleur: 53-44-39-34-33-15
Lemieux: 71-46-45-33-30-27-8-7-7-0
Kurri: 50-42-27-20-0-0
Marcel Dionne: 49-46-27-21-20-19-10-6-5-4

Iginla: 53-43-34-32-5-0
Jagr: 42-37-34-30-24-23-22-13-0-0
Bondra: 68-37-21-18-16-15
Yzerman: 39-36-19-16-10-10-3
Robitaille: 24-20-16-10-8-8-7-6-6-3-0

Of note, Mike Bossy has the fewest above-average seasons with just one. By comparing the players' finishes to the 15th-ranked player, it has been determined that he actually led the field by less than the others in this group.

Additionally, Wayne Gretzky actually had a higher % lead over the 15th-ranked player in his 87-goal season (((87 / 41) - 1) * 0.7637) than in his 92-goal season (((92 / 44) - 1) * 0.7637). This is the same result as one would find in his % leads over 10th, where the 77% represents 1983-84, and the 76% represents the 1981-82 season. His 87-goal season is tied with Alexander Ovechkin's 2007-08 season as the largest % lead over 15th of this group.

We can compare with the % Lead Over 10th data:

Many of the numbers relative to one another in this first group stay similar between the % Lead Over 10th and % Lead Over 15th models with the exception of some of the very high-end peaks.

Brett Hull's 86-goal season drops to become the third-most dominant season listed in the % lead over 15th model. Bossy's 69-goal season becomes the lowest peak season of the five players listed. Ovechkin and Gretzky both become tied for the top season in terms of % lead dominance.

In the second group, Lafleur falls beneath the above-average threshold in his % Lead Over 15th results. Kurri also falls below the threshold. Jarome Iginla sees an increase relative to these two players.

Mario Lemieux has the most significant drop in peak % lead (his 85-goal season in 1988-89); this is because the difference between 10th and 15th place that year was minuscule. The 10th-ranked goal scorer in 1988-89 had 46 goals, and the 15th-ranked player had 44 -- six players scored between 46 and 44 goals that season. Several scorers were bunched up at that total, whereas in other years there would be a sharper decline in scoring from 10th to 15th (for example, in 1983-84, the 10th-placed finisher had 47 goals while the 15th-placed finisher had 41 goals). Selanne's 76-goal season remains below-average relative to the threshold, as the 15th-placed finisher scored 48 goals that year. More players were recording large goal totals during those season; we have an understanding of the degree to which scoring was easier across the league from year to year. When Bossy scored 69 goals in 1978-79, the 15th-placed scorer had 36 goals -- in terms of raw numbers, a slightly lesser degree of dominance than Lemieux's 1988-89 season.

Thank you, Zuluss, for your work. You laid the foundation for this endeavor, and I will always be grateful for your input.

 

[Regal]

Maintaining the pace for the whole season is the trick.
Ovechkin had a fair number of hot starts (e.g., 41 goals in 53 games in 09/10 and going about goal-per-game in the first half of 13/14). Had he been injured after those hot starts, his gpg numbers would have been up there with Lemieux'. But he stayed healthy, and now the gpg comparison is punishing him for that.

Perhaps. Goalscoring is typically more streaky than points, so it's not as easily projected. Still, Lemieux had some large leads that had the potential for more in all three of his wins, and some of his worse years are the ones with lower games played, likely as he played through injury. It's possible he would do even better or more games in several years. I still believe that Lemieux was a better goalscorer when healthy, but I also think Ovechkin belongs in HO's tier 1, despite the playoffs.

 

[bobholly39]

Maintaining the pace for the whole season is the trick.
Ovechkin had a fair number of hot starts (e.g., 41 goals in 53 games in 09/10 and going about goal-per-game in the first half of 13/14). Had he been injured after those hot starts, his gpg numbers would have been up there with Lemieux'. But he stayed healthy, and now the gpg comparison is punishing him for that.

When I first read your post I thought you' be talking about Crosby and I was going to agree.

Hot starts doesn't mean you can keep the pace to the end and Crosby particularly has been guilty of that a couple of times at least.

Lemieux is the one player in history for whom I disagree however.

He's maintained incredible paces for full seasons and his numbers usually go up in full seasons. I think if there's one player in hockey history to whom you give the benefit of the doubt to in terms of being able to maintain a scoring pace over a longer period of time to - it has to be Lemieux.

 

[Big Phil]

This is a little rough, and I am not even going to put it in the exact order but from a goal scoring standpoint if I were to pick my top 10 guys it might look like this:

Gretzky
Howe
Bobby Hull
Lemieux
Richard
Bossy
Ovechkin
Esposito
Conacher
Brett Hull

I have to leave Bure off this list because his career was cut short and in comparison to the rest of these guys that's glaring. Even to Bossy. With the case of Bossy he scored 50 9 straight years and Bure had just 5 elite goal scoring years, Bossy 9. Plus the playoffs where he just simply kept going. Bure has a 0.62 GPG which is incredible but I like Hull's 0.58 with a 20 year career better.

The rest are obvious. Conacher may not be on everyone's list, but only three players led the league in goals more times than him.

Ovechkin has to make this list, it is pretty hard to keep him off of here.

I tend to favour the players who were great goal scorers AND playmakers because to me it makes the goal scoring seem all that more impressive. But either way this is my personal top 10.

 

[Blade Paradigm]

@Zuluss,

I have an update. I am proposing a more nuanced system based on the preliminary % lead model. I re-assessed your original post and realized that the 0.9 modifier used for the 1970-97 era was determined by dividing the 2nd place over 10th place averages. While the 2nd place over 10th percentage leads between the two eras are more similar than the 1st over 10th percentage leads, the same applies to all 2nd place over 10th leads -- the further down the list, the more similar the margins in many cases. It would not be precise solely to pick and choose the multiplier based on whichever-placed finish has the most similar % leads compared to those of the modern era.

I have, thus, devised a model that takes the multipliers of the 1st, 2nd, and 3rd placed percentage leads over 10th of each era and averages them out to create a new multiplier.

Here are the % leads and % lead multipliers of each era for 1st, 2nd, and 3rd place over 10th:

---------------------------------------------------------------
1917-18 to 1922-23:

1st place over 10th:
Average: 139.70137096134%
Multiplier (1997-2017 era / 1917-1923 era): 0.3262829140345794

2nd place over 10th:
Average: 121.0213741366992%
Multiplier (1997-2017 era / 1917-1923 era): 0.24754920144493

3rd place over 10th:
Average: 91.18465706120505%
Multiplier (1997-2017 era / 1917-1923 era): 0.272941223574166
---------------------------------------------------------------
1923-24 to 1940-41:

1st place over 10th:
Average: 73.2620751799498%
Multiplier (1997-2017 era / 1923-1941 era): 0.622179624313544

2nd place over 10th:
Average: 52.11469769003309%
Multiplier (1997-2017 era / 1923-1941 era): 0.5748617156621746

3rd place over 10th:
Average: 41.98403386354883%
Multiplier (1997-2017 era / 1923-1941 era): 0.5927980134153855
---------------------------------------------------------------
1941-42 to 1946-47:

1st place over 10th:
Average: 70.67110711561513%
Multiplier (1997-2017 era / 1941-1947 era): 0.6449901844231923

2nd place over 10th:
Average: 32.33627547128691%
Multiplier (1997-2017 era / 1941-1947 era): 0.9264748054212349

3rd place over 10th:
Average: 27.1068068722531%
Multiplier (1997-2017 era / 1941-1947 era): 0.9181476810148292

You'll notice that the 1st-over-10th % lead is so much higher than the 2nd and 3rd place over 10th % leads. In fact, the multipliers are >0.9 for 2nd and 3rd place, making clear that the league's players, for the most part, did not dominate their peers in the goal-scoring leader boards to a degree much greater than in the modern era. We must recall that this takes into account dominance over peers. We will take the average of the 1st, 2nd, and 3rd place multipliers in each era to determine our final modifiers.
---------------------------------------------------------------
1947-48 to 1969-70:

1st place over 10th:
Average: 74.79654266560749%
Multiplier (1997-2017 era / 1947-1970 era): 0.6094154727936524

2nd place over 10th:
Average: 47.12100941712862%
Multiplier (1997-2017 era / 1947-1970 era): 0.6357831654263735

3rd place over 10th:
Average: 35.9512521282787
Multiplier (1997-2017 era / 1947-1970 era): 0.6922721851431556
---------------------------------------------------------------
1970-71 to 1996-97:

1st place over 10th:
Average: 57.32026999682465%
Multiplier (1997-2017 era / 1970-1997 era): 0.795219045800323

2nd place over 10th:
Average: 32.78734923766489
Multiplier (1997-2017 era / 1970-1997 era): 0.9137287771617871

3rd place over 10th:
Average: 24.75942837229907
Multiplier (1997-2017 era / 1970-1997 era): 1.005194929997691
---------------------------------------------------------------
1997-98 to 2016-17:

1st place over 10th:
Average: 45.58217041189178

2nd place over 10th:
Average: 29.95874452530799

3rd place over 10th:
Average: 24.88805186947602
---------------------------------------------------------------
We notice with these figures that the differences of the multipliers between the 1st, 2nd, and 3rd place over 10th within each era are different compared to the differences of those in other eras; for example, in the 1970-1997 era, the multipliers are each separated by approximately 0.1. In the 1947-1970 era, they are separated by ~0.05. In the 1941-1947 era, we see the effects of WWII, where the 2nd and 3rd placed finishers led the 10th-placed finisher by a much smaller margin than the 1st-placed finisher. This means that, in this era, a 50 goal over 23 goal lead by 1st over 10th (Maurice Richard, 1944-45) would be than a 52 goal over 25 goal lead by 1st over 10th in the 1947-1970 era (Bobby Hull, 1966-67). Such a lead in the WWII era was an anomaly and out of the ordinary, whereas in the subsequent era it was much more standard; the frequency with which the top players recorded large goal totals was much greater in the 1947-1970 era, making such leads less significant and more of a product of the era.

In order to determine final multipliers to be used for determining the adjusted % leads of any player, we determine the average of the 1st, 2nd, and 3rd over 10th percentage leads of each era. The resultant numbers take into account the climate of goal-scoring within the era, whereas the preliminary model was much less clearly defined in its estimation of the climate.

Here are the final multipliers:

FINAL MULTIPLIERS (AVERAGE OF 1st, 2nd, 3rd OVER 10th):

====================================================

1917-18 to 1922-23:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.2822577796845585

====================================================

1923-24 to 1940-41:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.5966131177970347

====================================================

1941-42 to 1946-47:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.8298708902864188

====================================================

1947-48 to 1969-70:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.6458236077877272

====================================================

1970-71 to 1996-97:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.9047142509866004

====================================================

To calculate the adjusted % lead of any player's finish over 10th, use the formula: [([the player's goal total / the goal total of the 10th-placed finisher that year] - 1) * (final multiplier for the era that the season took place within)] * 100 =

We can use whole numbers for the final % leads, as the difference between tenths is less than one goal.

For example, Bobby Hull's 1966-67 season % lead (52 goals / 25 goals) is a 70% adjusted lead. Maurice Richard's 1944-45 season % lead (50 goals / 23 goals) is a 97% adjusted lead.

*Note: You can copy and paste numbers into the standard Calculator program in Windows. This ensures precision.

-----------------------------------------------------------------------------------------

The average % lead of the 1st-place finisher over 10th in the modern era (1997-2017) was 45.58217041189178%. Those seasons with % leads higher than the 46% baseline are to be deemed at the level of an above-average 1st-place finish by modern era standards. One could deem these, for our purposes, "elite goal-scoring seasons" -- better than the average first-place finish in the modern era.

Adjusted Percentage Leads Over 10th:

Richard: 97-66-65-56-49-47-38-29-26-24-12-11-8
Howe: 94-68-62-54-38-37-34-30-21-19-19-17-14-12-11-10-5-2-0
Esposito: 101-80-67-47-34-28-20-11-5
Gretzky: 77-76-53-43-43-16-13-12-8
Lemieux: 77-42-41-26-25-17-4-4-0
Bobby Hull: 70-60-60-51-46-40-36-30-29-20-12-12-10
Ovechkin: 63-61-52-52-50-44-43-30-15-6

Bure: 61-55-48-28-10
Brett Hull: 82-60-54-22-10-0-0-0
Charlie Conacher: 48-47-46-33-26
Stamkos: 67-46-38-32-30-9
Joe Malone: 67-39-30-12
Bossy: 66-38-30-29-25-24-23-8-8
Bernie Geoffrion: 60-47-25-16-16-13-6-0

Selanne: 58-37-21-20-19-0
Bondra: 58-28-13-10-9-5
Howie Morenz: 50-40-39-34-26-24-18-16-11
Lafleur: 50-45-29-27-27-6
Kurri: 49-43-26-10
Marcel Dionne: 43-43-19-16-15-12-0-0-0
Iginla: 41-26-25-24

Patrick Kane: 39-10-0
Yzerman: 37-34-12-7-6-4
Jagr: 35-30-29-21-17-13-11-6
Sakic: 35-8-6-5-0
Heatley: 25-25-11-11-3-0
Neely: 20-12-8-3
Robitaille: 15-14-9-6-4-0-0-0-0

 

[Sentinel]

As I said in another thread, Ovechkin is absolutely on the shortlist of eight greatest goalscorers of all time. However, he comes up really short where the other seven don't. Even Brett Hull, who is not exactly known as the best playoff performer, has two Cups and scored 10+ goals in playoffs four times. Ovechkin had one such playoffs.

 

[Blade Paradigm]

As I said in another thread, Ovechkin is absolutely on the shortlist of eight greatest goalscorers of all time. However, he comes up really short where the other seven don't. Even Brett Hull, who is not exactly known as the best playoff performer, has two Cups and scored 10+ goals in playoffs four times. Ovechkin had one such playoffs.

If one is to come up with a shortlist based on the new, revised Adjusted % Leads Over 10th system above, I think the group would be:

Maurice Richard
Gordie Howe
Phil Esposito
Wayne Gretzky
Mario Lemieux
Bobby Hull
Alexander Ovechkin
Pavel Bure
Brett Hull
Charlie Conacher

Based on the Adjusted % Leads Over 10th system, we see that Steven Stamkos' five best seasons are better than Mike Bossy's. Bossy played in an era when there were many 40-and-50-goal scorers every year. He only dominated his peers once, and that is because there was a sudden influx of 40-goal scorers from 1979-80 onward through the next decade-plus. Additionally, he played on a dynasty team -- his numbers are within the context of the best possible situation for him to thrive.

I would leave Bossy off of the list.

Conacher did not peak as high as Bossy did, but he had three seasons that are above the 46% average of 1st-place over 10th % leads in the modern era (1997-2017). In other words, he had three seasons that would be considered above-average 1st-place finishes by today's standards. Bure had as many Top 10 finishes as Conacher, but peaked much higher; contextual evidence shows that he achieved his peak seasons with very little assistance and while suffering from debilitating knee problems.

Stamkos vs Bossy:

Stamkos: 67-46-38-32-30-9
Bossy: 66-38-30-29-25-24-23-8-8

By the time Stamkos finishes his career, he might have padded out his stats with another low Top-10 finish or two. I have highlighted their Top 5 % lead seasons in purple.

Here are the number of 40-goal scorers each year in the NHL from the start of Bossy's career to the end:

1977-78: 11
1978-79: 10
1979-80: 24
1980-81: 19
1981-82: 22
1982-83: 23
1983-84: 24
1984-85: 23
1985-86: 23
1986-87: 17

Bossy dominated his peers to a much greater extent in 1978-79 than in any other year. In every year aside from 1978-79, the field was much closer to him in the goal-scoring race.

 

[Hockey Outsider]

20-25% lead over 2nd is nothing special, Ovechkin leads in all Richard-winning seasons but two were about that big. (The exceptions are the lockout-shortened 12/13, when Stamkos was 2nd, and 15/16, Kane's career year).

If you look at leads over 5th/10th, which should be immune from who was whose linemate and who had a career year/off-season, Lemieux leads in 87/88 and 95/96 were 37% and 35% (46% and 47%). That's a bit below par for the course for a goal-scoring title winner of 1970-96 (38% and 59% leads) and 1.5-2 times better than the lead of a typical runner-up over 5th/10th (16% / 33%).

Ovechkin has 5 seasons when his lead over 5th is 38% or better (plus another 32% lead) and 5 seasons when his lead over 10th is 50% and more (plus two more with 44% and 43% leads). The average leads of #1 and #2 over 5th/10th have also shrunk during his time, making all goal-scoring titles by Ovechkin better than average (oftentimes significantly so) and roughly twice better than a typical runner-up campaign for the era.



That's true, but are they worth more than 700 games or so, which are the goal-scoring prime of Lemieux and Ovechkin? I doubt that.
Plus, we do not know how Ovechkin would have fared in the playoffs had he spent his peak years on a team comparable to Pens of the early 90s. Lemieux was on the right team at the right time. Ovechkin was not. Of course, you cannot credit Ovechkin for the games he did not play, but it is also not as if he played them and did not do anything. He just did not have a chance.

First, I appreciate the thoughtful and detailed response.

I'll start with the playoff discussion. I think we fundamentally disagree on how heavily the post-season should be weighed. This will probably need to be in the "agree to disagree" category.

====
To recap the "percentage lead over 10th place" argument. You said that Ovechkin generally has larger leads over 5th place, 10th place, etc., compared to Lemieux. My response is Lemieux faced tougher competition, which made it more difficult to lead the league in goals by larger margins. Your response to that is, generally, the league leaders during Ovechkin's era are smaller than during Lemieux's era, which means Lemieux couldn't have faced tougher competition.

My response to that you can't compare the leading scorer to 5th/10th place, look at the margin of victory, and draw a conclusion on the talent of the league.

It's often (widely?) accepted that high end talent in NHL was relatively weak leading up towards the 2005 lockout. If you look at how many points the Art Ross winner scored compared to 10th place, for the three years leading up to the lockout, the margins were 25%, 25% and 19%. That seems to be fairly low by historical standards.

If we look at, say, 1987 and 1988, the margin between the league leader and 10th place was much higher- 93% and 58% respectively. I don't intend to mis-represent your position (so tell me if I misunderstood you), but I think you'd tell me that competition couldn't have been tougher in 1987 or 1988, since someone was able to achieve such a big margin over your benchmark (10th place).

But that defies common sense. The top ten scorers in 1987 include nine Hall of Famers all more-or-less in their prime (Gretzky, Kurri, Lemieux, Messier, Gilmour, Ciccarelli, Hawerchuk, Goulet and Bourque). In 1988, the list consist of eight Hall of Famers in their prime (Lemieux, Gretzky, Savard, Hawerchuk, Stastny, Robtaille, Messier and Goulet). The competition from 2002 to 2004 was clearly weaker (not that there weren't some great players, but I have a hard time imagining someone like Cory Stillman, Alex Tanguay, Robert Lang or a 39-year-old Adam Oates placing in the top ten in scoring in the late eighties).

In summary, my response is it's wrong to assess the quality of the league simply by looking at the percentage between 1st and 10th (or 5th, or any other position). It's obvious, from going through a list of names, that Lemieux faced tougher goal-scoring competition than Ovechkin, and that's why comparing their margin over second place (although interesting) isn't a strong argument.

 

[cupcrazyman]

Bossy for sure. Hull Jr deserves a mention too. Here is all time goals per game.

NHL All-Time Goals per Game Leaders

A few guys with shortened careers due to injuries.

 

[daver]

First, I appreciate the thoughtful and detailed response.

I'll start with the playoff discussion. I think we fundamentally disagree on how heavily the post-season should be weighed. This will probably need to be in the "agree to disagree" category.

====
To recap the "percentage lead over 10th place" argument. You said that Ovechkin generally has larger leads over 5th place, 10th place, etc., compared to Lemieux. My response is Lemieux faced tougher competition, which made it more difficult to lead the league in goals by larger margins. Your response to that is, generally, the league leaders during Ovechkin's era are smaller than during Lemieux's era, which means Lemieux couldn't have faced tougher competition.

My response to that you can't compare the leading scorer to 5th/10th place, look at the margin of victory, and draw a conclusion on the talent of the league.

It's often (widely?) accepted that high end talent in NHL was relatively weak leading up towards the 2005 lockout. If you look at how many points the Art Ross winner scored compared to 10th place, for the three years leading up to the lockout, the margins were 25%, 25% and 19%. That seems to be fairly low by historical standards.

If we look at, say, 1987 and 1988, the margin between the league leader and 10th place was much higher- 93% and 58% respectively. I don't intend to mis-represent your position (so tell me if I misunderstood you), but I think you'd tell me that competition couldn't have been tougher in 1987 or 1988, since someone was able to achieve such a big margin over your benchmark (10th place).

But that defies common sense. The top ten scorers in 1987 include nine Hall of Famers all more-or-less in their prime (Gretzky, Kurri, Lemieux, Messier, Gilmour, Ciccarelli, Hawerchuk, Goulet and Bourque). In 1988, the list consist of eight Hall of Famers in their prime (Lemieux, Gretzky, Savard, Hawerchuk, Stastny, Robtaille, Messier and Goulet). The competition from 2002 to 2004 was clearly weaker (not that there weren't some great players, but I have a hard time imagining someone like Cory Stillman, Alex Tanguay, Robert Lang or a 39-year-old Adam Oates placing in the top ten in scoring in the late eighties).

In summary, my response is it's wrong to assess the quality of the league simply by looking at the percentage between 1st and 10th (or 5th, or any other position). It's obvious, from going through a list of names, that Lemieux faced tougher goal-scoring competition than Ovechkin, and that's why comparing their margin over second place (although interesting) isn't a strong argument.

Is this not a product of extremely low scoring seasons that seem to promote parity and allow seemingly inferior players to creep into the Top Ten? Many would argue that the seasons immediately preceding 2002-2004 was still the golden age with the big name players (Jagr, Sakic, Forsberg, Lindros, Selanne, Bure, Kariya) dominating the Top Ten scorers.

Yes, goals and points took a noticeable hit after 2001, mainly as PPs went down, but does mean that the Top 20 or 30 scorers, which is the size of peer group that I would suggest in this analysis, is that much worse than in any other season?

I would have trouble believing that a Top Ten goalscorer from years 2002-2004, could not have had a chance to finish in the Top Ten in any other season.

 

[Sentinel]

Based on the Adjusted % Leads Over 10th system, we see that Steven Stamkos' five best seasons are better than Mike Bossy's. Bossy played in an era when there were many 40-and-50-goal scorers every year. He only dominated his peers once, and that is because there was a sudden influx of 40-goal scorers from 1979-80 onward through the next decade-plus. Additionally, he played on a dynasty team -- his numbers are within the context of the best possible situation for him to thrive.

I would leave Bossy off of the list.

Which is, of course, ridiculous and torpedoes the entire method. When numbers-crunching goes against common sense, we can throw the former in the dustbin. Bossy is and will forever remain a greater goalscorer than Steven Stamkos.

 

[Fantomas]

My response is Lemieux faced tougher competition, which made it more difficult to lead the league in goals by larger margins.

I am not saying that you're wrong, but can one really prove "tougher competition"? All that you can show with numbers is that some players are better than others in a particular period, but comparing across eras is an exercise fraught with problems.

You can show, for example, that a particular era has more high scoring players or more hall of famers. But that doesn't mean the competition is stronger. What about other variables such as league-wide depth? Is it more impressive for a player to dominate a top-heavy league or a league in which scoring is more spread out?

I am skeptical of any claims that argue that the NHL had tougher competition going back to the cold war days and even the 1990s. Isn't the NHL more deep and global than it has ever been?

 

[daver]

I am not saying that you're wrong, but can one really prove "tougher competition"? All that you can show with numbers is that some players are better than others in a particular period, but comparing across eras is an exercise fraught with problems.

You can show, for example, that a particular era has more high scoring players or more hall of famers. But that doesn't mean the competition is stronger. What about other variables such as league-wide depth? Is it more impressive for a player to dominate a top-heavy league or a league in which scoring is more spread out?

I am skeptical of any claims that argue that the NHL had tougher competition going back to the cold war days and even the 1990s. Isn't the NHL more deep and global than it has ever been?

I agree. Big numbers from the 80 and 90s seem to get an unwarranted amount of praise when there really is no reason to believe that a Top 5 - 10 scorer from any era would outside that range in another era.

 

[Fantomas]

Let's not forget also the fact that the NHL has pretty much phased out goonery from the league. In the past you'd have guys playing with barely any talent aside from dropping the gloves, fourth lines were mostly filler and top lines as a result played 25 minutes a game.

Lemieux also had a fun time demolishing expansion teams in the 1990s. Remember how awful they were? But look at Vegas now. Is this not evidence of the parity of modern-day hockey?

I don't think there has ever been tougher competition than what we're seeing now. There has never been better goaltending, never been better defense. It's never been tougher to score than it is now. And I think it's never been tougher to dominate a league by a wide margin than today. Unfortunately quantifying any of this with numbers is probably impossible because numbers provide very limited information.

And I say this with no disrespect to 80s and 90s hockey. Those eras were much more fun to watch and I miss them, but more competitive they weren't.

 

[Neutrinos]

I'm pretty sure Bure's combination of speed + stickhandling + shot power puts him in my top 5, so I'm a little surprised he's not being universally pegged as top 10

 

[Canadiens1958]

@Zuluss,

I have an update. I am proposing a more nuanced system based on the preliminary % lead model. I re-assessed your original post and realized that the 0.9 modifier used for the 1970-97 era was determined by dividing the 2nd place over 10th place averages. While the 2nd place over 10th percentage leads between the two eras are more similar than the 1st over 10th percentage leads, the same applies to all 2nd place over 10th leads -- the further down the list, the more similar the margins in many cases. It would not be precise solely to pick and choose the multiplier based on whichever-placed finish has the most similar % leads compared to those of the modern era.

I have, thus, devised a model that takes the multipliers of the 1st, 2nd, and 3rd placed percentage leads over 10th of each era and averages them out to create a new multiplier.

Here are the % leads and % lead multipliers of each era for 1st, 2nd, and 3rd place over 10th:




1941-42 to 1946-47:

1st place over 10th:
Average: 70.67110711561513%
Multiplier (1997-2017 era / 1941-1947 era): 0.6449901844231923

2nd place over 10th:
Average: 32.33627547128691%
Multiplier (1997-2017 era / 1941-1947 era): 0.9264748054212349

3rd place over 10th:
Average: 27.1068068722531%
Multiplier (1997-2017 era / 1941-1947 era): 0.9181476810148292

You'll notice that the 1st-over-10th % lead is so much higher than the 2nd and 3rd place over 10th % leads. In fact, the multipliers are >0.9 for 2nd and 3rd place, making clear that the league's players, for the most part, did not dominate their peers in the goal-scoring leader boards to a degree much greater than in the modern era. We must recall that this takes into account dominance over peers. We will take the average of the 1st, 2nd, and 3rd place multipliers in each era to determine our final modifiers.



We notice with these figures that the differences of the multipliers between the 1st, 2nd, and 3rd place over 10th within each era are different compared to the differences of those in other eras; for example, in the 1970-1997 era, the multipliers are each separated by approximately 0.1. In the 1947-1970 era, they are separated by ~0.05. In the 1941-1947 era, we see the effects of WWII, where the 2nd and 3rd placed finishers led the 10th-placed finisher by a much smaller margin than the 1st-placed finisher. This means that, in this era, a 50 goal over 23 goal lead by 1st over 10th (Maurice Richard, 1944-45) would be than a 52 goal over 25 goal lead by 1st over 10th in the 1947-1970 era (Bobby Hull, 1966-67). Such a lead in the WWII era was an anomaly and out of the ordinary, whereas in the subsequent era it was much more standard; the frequency with which the top players recorded large goal totals was much greater in the 1947-1970 era, making such leads less significant and more of a product of the era.



====================================================

1917-18 to 1922-23:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.2822577796845585

====================================================

1923-24 to 1940-41:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.5966131177970347

====================================================

1941-42 to 1946-47:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.8298708902864188

====================================================

1947-48 to 1969-70:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.6458236077877272

====================================================

1970-71 to 1996-97:

Average of Multipliers for 1st + 2nd + 3rd Averages Over 10th: 0.9047142509866004

====================================================

To calculate the adjusted % lead of any player's finish over 10th, use the formula: [([the player's goal total / the goal total of the 10th-placed finisher that year] - 1) * (final multiplier for the era that the season took place within)] * 100 =

We can use whole numbers for the final % leads, as the difference between tenths is less than one goal.

For example, Bobby Hull's 1966-67 season % lead (52 goals / 25 goals) is a 70% adjusted lead. Maurice Richard's 1944-45 season % lead (50 goals / 23 goals) is a 97% adjusted lead.

Limiting obseravtions to the 1941 - 1970 era of the NHL since it appears that you have made some interesting misrepresentations and ommissions.

Let's start with the 1941 to 1947 period. Bolded you seem to attribute the scoring differentials for 1st thru 10th to WWII.

What about the introduction of the Red Line at the start of the 1943-44 season which opened up the game offensively and the growth within seven seasons of the schedule from 48 to 60 games.

Likewise during the O6 era, the schedule grew from 60 to 70 games, requiring increases in player rosters plus the rule change before the start of the 1956-57 season limiting the number of PPG to 1 per penalty. Then we have the 1967 expansion, longer schedule and an unbalanced schedule.

Please explain the validity of your calculations given these oversights.

 

[Zuluss]

Which is, of course, ridiculous and torpedoes the entire method. When numbers-crunching goes against common sense, we can throw the former in the dustbin. Bossy is and will forever remain a greater goalscorer than Steven Stamkos.

Another method would be to give "common sense" a check - had humanity not done that, we would have still thought the Sun orbits around the earth

If you look at VsX, which is not something people on this board would throw in the dustbin any time soon, Bossy's VsX goals for the best 7 seasons are 52.9, Stamkos' are 50.4. That's a difference, but given the fact that Stamkos' seven-highest goals total in a season is 25 goals in 37 games (13/14), he will probably close the gap this year or next.

Moreover, if we look at the modified leads over #10 Blade Paradigm presents, I do not think they support Stamkos over Bossy yet.

Stamkos: 67-46-38-32-30-9
Bossy: 66-38-30-29-25-24-23-8-8

Best season is a perfect tie, best three and best five go Stamkos' way, but Bossy has two more good seasons and then one more decent one outside of that, so Stamkos has ground to cover, and he is almost 28.
Just to put numbers in context, "two more good seasons" feature leads over 10th which are two goals less than the lead Crosby won the Rocket with last season. So, two more close 2nd places behind a weak Rocket winner is what Stamkos needs to pass Bossy's career. Neither easy, nor impossible.

 

[Canadiens1958]

I am not saying that you're wrong, but can one really prove "tougher competition"? All that you can show with numbers is that some players are better than others in a particular period, but comparing across eras is an exercise fraught with problems.

You can show, for example, that a particular era has more high scoring players or more hall of famers. But that doesn't mean the competition is stronger. What about other variables such as league-wide depth? Is it more impressive for a player to dominate a top-heavy league or a league in which scoring is more spread out?

I am skeptical of any claims that argue that the NHL had tougher competition going back to the cold war days and even the 1990s. Isn't the NHL more deep and global than it has ever been?

Interesting comments. Problem is doing an all inclusive study. The beautiful girl phenomena. In other words falling into the trap of looking at the attractive scoring data and nothing else.

No one looks at goaltending. What were the results of O6 goaltenders playing 3 games in 4 nights vs modern goaltenders doing the same? Etc

Nobody looks at the roster size aspects, the rule change impacts, etc.

Looking at superficial wow factors like scoring produces invalid and disappointing results.

 

[Zuluss]

My response to that you can't compare the leading scorer to 5th/10th place, look at the margin of victory, and draw a conclusion on the talent of the league.

It's often (widely?) accepted that high end talent in NHL was relatively weak leading up towards the 2005 lockout. If you look at how many points the Art Ross winner scored compared to 10th place, for the three years leading up to the lockout, the margins were 25%, 25% and 19%. That seems to be fairly low by historical standards.

If we look at, say, 1987 and 1988, the margin between the league leader and 10th place was much higher- 93% and 58% respectively. I don't intend to mis-represent your position (so tell me if I misunderstood you), but I think you'd tell me that competition couldn't have been tougher in 1987 or 1988, since someone was able to achieve such a big margin over your benchmark (10th place).

That's a good illustration why one cannot use select years instead of decade/two-decade averages, and why I also brought up #2 margins over #10. Most wins within any decade or two are "normal" wins - either by a very good goal-scorer having a career year, or by an all-time great looking human during this particular year. The same is true, and to a much greater extent, about #2 finishers.
Besides, "league-wide competition" is likely to be a long-term thing, it should not change all that much year over year.

Going back to your original post, your list of competitors in 1985-1997 has a lot of big names, but that is not synonymous with tough competition

• Hull 3x
• Yzerman 3x
• Gretzky 2x
• Bossy 2x
• Bure 2x
• Kurri 2x
• Kerr 2x
• Mogilny
• Selanne
• Robitaille
• Jagr
• Turgeon

Gretzky, Kurri, Kerr, and Bossy were largely done as goal-scorers after 1987, Bure was injured most of the time, Mogilny had one big season and that was more or less it, and the rest is not enough to fill top10. So, #10 most years in 1985-1997 would not be a tough guy to beat. He might be a big name guy in some years, but big name guys also have a lot of so-so seasons.

 

[Puck Dogg]

Bossy played in an era when there were many 40-and-50-goal scorers every year. He only dominated his peers once, and that is because there was a sudden influx of 40-goal scorers from 1979-80 onward through the next decade-plus. Additionally, he played on a dynasty team -- his numbers are within the context of the best possible situation for him to thrive.

I would leave Bossy off of the list.

Gretzky had also great team to play with and it doesn't diminish his accomplishments. Mario had also two- time Stanley Cup team in 1990-92 to play with. For his defense, Bossy has the highest goal per game ratio of all goal scorers in the NHL. Lemieux comes close, but falls short slightly.

Bossy scored 50-60 goals over nine seasons - starting allready at his rookie season with 53, which was a record until Selanne broke it. Bossy also had to retire early, only at 30 years of age. He should be a lock for top ten scorers of all time.

 

[Zuluss]

First, I appreciate the thoughtful and detailed response.

I'll start with the playoff discussion. I think we fundamentally disagree on how heavily the post-season should be weighed. This will probably need to be in the "agree to disagree" category.

As for the playoff discussion, here is the list of top10 (actually top12, because of a tie for #10) players in terms of GPG post-lockout (in order):

Ovechkin, Stamkos, Kovalchuk, Crosby, Malkin, Hossa, Heatley, Tarasenko, Nash, Kucherov, Selanne, Tavares

Here is a similar ranking on playoff GPG (again, in order):

Ovechkin, Zetterberg, Alfredsson, Kessel, Briere, Iginla, Gagne, Crosby, Franzen, Kane, Malkin, Marleau

The first list looks very close to how anyone would rank post-lockout goal-scorers, though one can quibble with a thing or two (is Malkin too high? should we drop Hossa? where is Iginla? do we really need a ranking outside top3, or are they all alike?)

The second list is just god-awful, it is not even clear where to start to make it right.

Those two rankings are measuring the same thing (goal-scoring), so it is clear the second one has a huge, huge amount of noise. People think that hot playoff runs (like the one that made Justin Williams playoff MVP) average out over a sample of 100 playoff games - well, look at Briere and Franzen, they do not.

A well-known rule from statistics is that the weights you should place on two signals are inversely proportional to the noise in them. Well, that is telling me to put much more weight on regular season. Emotions from the playoff wins of your favorite team and "the game is played to win the Cup" mottos tell you otherwise, but I think you need to remind yourself that those things are swaying you towards taking Franzen over Tavares and Briere over Heatley for goal-scoring.

 

[Hockey Outsider]

A bunch of responses, so I'll try to make one general reply.

To clarify, I wasn't commenting on the overall quality of the league. I was talking specifically about high-end competition. That's an important distinction.

It seems obvious to me that the average player from Ovechkin's era is better than the average player from Lemieux's era. It's also obvious to me that the difference is even more pronounced the deeper you go into the depth charts. Of course there will be plenty of exceptions to the principle, but in general I'm quite confident in making those statements.

The argument was made that "Ovechkin is a better goal-scorer than Lemieux because he beat the 2nd/5th/10th best goal-scorer by a larger margin". I said that's a bad argument, because it (implicitly) assumes that the high-end competition was the same in both eras. I grant that Ovechkin plays in a deeper/tougher/more competitive league on average. But since we're comparing these players to the 2nd/5th/10th best player in the league, why do we care about the league average? Even if Lemieux faced weaker goalies/defensemen than Ovechkin did, so did every other star in Lemieux's era, so he didn't have a relative advantage to his peers, which is what's relevant in this comparison.

On the topic of using mutli-decade averages - why? One of the blocks spans everything from 1971 (a year where two of the 14 teams were expansion teams, with highly imbalanced conferences, and hardly any European players) to 1997 (fairly balanced conferences, a few years out from expansion, only 5 of the 12 year-end all-stars were Canadian). Drawing conclusions about Lemieux's 1997 Art Ross, based on data from Orr's domination in the early seventies, is puzzling.

When I have time I'll post a summary of the top five or ten goal-scorers annually from Lemieux and Ovechkin's area. That should make it obvious why "Ovechkin beat his peers by a larger percentage than Lemieux" is a bad argument (since the underlying assumption - that both peer groups are equal - is false).

====
As for the playoffs - nobody is suggesting that Johan Franzen should be ranked over Ovechkin. But Ovechkin has by far the weakest playoff resume of the consensus top eight goal-scorers. Hull probably has the next weakest playoff resume, and I'd still rank him well ahead of Ovechkin in the postseason. Why should we ignore valid evidence?

Jude Drouin scored more per game in the playoffs than Marcel Dionne. Dionne is still, of course, a far superior scorer and player. But we shouldn't dismiss the playoffs altogether. Dionne's reputation is rightfully lessened because of his personal inability to score in the playoffs (relative to his lofty regular-season standards). Drouin's (far lesser) legacy is enhanced because he played well in the spring (even though he never won a Stanley Cup).

Just because some evidence doesn't provide a perfect ranking, it doesn't mean that evidence should be dismissed. Randy Carlyle won more Norris trophies than Borje Salming (or Brad Park, or Scott Stevens, etc) - but that doesn't mean we shouldn't at least consider the Norris trophy when evaluating a player's legacy. Ranking players solely based on playoff scoring (whether per-game or in total) isn't a perfect ranking, but that doesn't mean it should be ignored.

 

[Zuluss]

On the topic of using mutli-decade averages - why? One of the blocks spans everything from 1971 (a year where two of the 14 teams were expansion teams, with highly imbalanced conferences, and hardly any European players) to 1997 (fairly balanced conferences, a few years out from expansion, only 5 of the 12 year-end all-stars were Canadian). Drawing conclusions about Lemieux's 1997 Art Ross, based on data from Orr's domination in the early seventies, is puzzling.

Because, first of all, we need a benchmark of what constitutes a "dominant" or "weak" goal-scoring title win depending on an era. The fact that the benchmark (in terms of margin over 5th/10th) would be different for the O6 era and, say, the past decade seems obvious. The question is, how big is the difference?
The benchmark should be the "average goal-scoring title win". One way to get to that is to hand-clean the sample from obvious outliers (Gretzky, Ovechkin, Brett Hull's 90/91 and analyze that. Another way would be to take a longer sample and let the outliers average out, thus removing the subjectivity of defining "outliers".
The by-product of that approach is that we can test "quality of high-end competition" hypotheses.

As for the playoffs - nobody is suggesting that Johan Franzen should be ranked over Ovechkin. But Ovechkin has by far the weakest playoff resume of the consensus top eight goal-scorers. Hull probably has the next weakest playoff resume, and I'd still rank him well ahead of Ovechkin in the postseason. Why should we ignore valid evidence?

You should not ignore it, but you should question the degree of validity.
You cannot take a measure that ranks Franzen's goal-scoring ahead of Stamkos', compare Ovechkin and Lemieux on this measure and ask everyone to believe the latter comparison, and not the former.
Everyone would agree that Franzen vs. Stamkos gpg in post-season is mostly about Stamkos not playing enough games, and Franzen getting hot during two Cup runs (07/08 and 08/09). Had those two Cup runs been cut short because of the goalie laying an egg and had Tampa made playoffs in 11/12 and 12/13, the picture could have been vastly different.
More careful people will also remark that Stamkos' PO shooting % is 11.8 (vs. RS career 17%) and Franzen's PO shooting % is 12.5 (vs. RS career 11.6), and it was as high as 16.7 and 18.6 during the two Cup runs. So the difference in PO GPG between Stamkos and Franzen is mostly luck, and see how much it matters (PO GPG 0.39 vs 0.3 Franzen, RS GPG 0.54 vs. 0.31 Stamkos).
But then all comparisons of Ovechkin and Dionne in PO with anyone else are done with straight face as if there is no noise/luck in the numbers. Even more, this comparison is supposed to matter more than RS.

 

[Blade Paradigm]

Limiting obseravtions to the 1941 - 1970 era of the NHL since it appears that you have made some interesting misrepresentations and ommissions.

Let's start with the 1941 to 1947 period. Bolded you seem to attribute the scoring differentials for 1st thru 10th to WWII.

What about the introduction of the Red Line at the start of the 1943-44 season which opened up the game offensively and the growth within seven seasons of the schedule from 48 to 60 games.

Likewise during the O6 era, the schedule grew from 60 to 70 games, requiring increases in player rosters plus the rule change before the start of the 1956-57 season limiting the number of PPG to 1 per penalty. Then we have the 1967 expansion, longer schedule and an unbalanced schedule.

Please explain the validity of your calculations given these oversights.

One must note that the % lead model is not the be-all-end-all of analysis; it offers one aspect of a player's performance that could be used as part of a larger argument. Its purpose is to measure one's numerical dominance over their peers -- it does not account for the strength of the competition, but it considers the scoring trends of the top scorers and adjusts the % leads in accordance with the trends of each era. Likewise, the more goals that are scored by the competition, the less substantial each additional goal is for the player whose season is being measured. If Player A scores 70 goals, but many players are scoring 50, Player A is not as dominant as he would be in a season when he scores 50 and his competition scores 25. The method also measures the trend of top-end players scoring by large margins. For example, during the WWII era the second-and-third places did not separate themselves from the pack to a very great extent, so Maurice Richard's 1944-45 goal-scoring season relative to his peers is measured as a much more dominant season than a similar season in an era where the top players perennially dominated their peers -- Richard's season went against the trend, whereas such seasons were not as rare in the subsequent era and thus not as significant; scoring was more top-heavy in general across a 23 year period.

To summarize, the model takes into account two factors: 1) the raw percentage leads; 2) the trend of dominance during each particular era. #2 can be broken down into two additional factors: a) margins over 10th within an era; b) margins between #1, #2, and #3 over 10th.

The trend factor is what the multiplier represents. One can not adjust the margins from era to era by dividing the 1st place over 10th % lead averages alone; one must also account for the degree to which the second-and-third placed finishers dominated. If all three were dominant, then the adjustment reflects a very top-heavy scoring trend where dominant finishes were more commonplace and not as rare or significant as in other eras (1947-1970); if the 1st-placed finisher led by a wide margin but the second-and-third placed finishers showed more parity relative to their peers, then more weight is given to dominance during that era (1941-1947; 1970-1997). The second and third placed finishers reflect the league-wide trend and serve to shape the multiplier, since they comprise 2/3rds of the multiplier.

I am always open to suggestions. After all, our shared objective is to come to the most accurate conclusion about these players from a quantitative point of view. Nothing about this model is definitive. It offers a numerical proof of dominance over one's field in terms of raw goal totals, and it should be viewed as such. One still needs to interpret these numbers with contextual evidence.

What I did was develop the original model presented by Zuluss for Adjusted % Lead Over 10th with the intention of finding more precise numbers based on the eras originally outlined by him.

The eras were further broken down by categorizing them based on a determination of which groups of seasons featured GF averages of >3.0 and <3.0.

NHL League Averages | Hockey-Reference.com

1917-1918: 4.75
1918-19: 4.13
...
1991-92: 3.96
1922-23: 3.26
------------------------
1923-24: 2.66
1924-25: 2.50
...
1939-40: 2.49
1940-41: 2.68
------------------------
1941-42: 3.12
1942-43: 3.61
...
1945-46: 3.34
1946-47: 3.16
------------------------
1947-48: 2.93
1948-49: 2.72
...
1968-69: 2.98
1969-70: 2.90
-----------------------
1970-71: 3.12
1971-72: 3.07
...
1995-96: 3.14
1996-97: 2.92 (*the leap between this season and the next is substantial enough to separate the two)
-----------------------
1997-98: 2.64
1998-99: 2.63
...
2015-16: 2.71
2016-17: 2.77
-----------------------
These eras were not arbitrarily defined. The numbers themselves outline the eras for us. In most cases, the increase or decrease in GF averages from the last season of one era to the first season of the next is quite substantial.

If someone here wants to offer an alternative, please feel free to.

Again, the Adjusted % Lead model is not a wholesale answer to which players were better than others; it simply measures the dominance of players over their field in terms of goal totals. This is useful because goals in each era do not have the same value as in other eras (e.g., 50 goals in 2016-17 vs 50 goals in 1992-93). To compare face-value goal totals across eras is a fruitless exercise that too many people engage in. To quantify the degree to which one measures against their field is to provide a numerical reference as to the player's degree of dominance, which is a statistic that can be used for cross-era comparisons.

Context is still required in any discussion; this model simply serves as a numerical measure of dominance in terms of goal totals, adjusted by the league-wide trends of dominance.

I would use an Adjusted % Leads Over 15th model as well so that the results from the APLO10 can be compared with the results from the APLO15 model. An Adjusted % Leads Over 15th model would give us a broader scope of each player's career (Top 15 is closer to what one might consider a minimally-sufficient finish for a high-end scorer) and also compare it with a more evenly-spread field. Once the leader board reaches the 15th or so place in any given year, we usually start to see many players compile the same totals -- an even better idea of an "average" NHL season at any time than the 10th-placed finish.