Comparing Top 3, 5 or 10 scoring and Hart finishes from different eras

[daver]

Sidney Crosby now has a scoring finish resume and Hart trophy nomination resume that rivals the two players who usually are viewed as the #5 player of all-time, Bobby Hull and Jean Beliveau.

Here are their best Hart finishes:

Crosby - 1, 1, 2 ,2, 2, 2, 3, 4, 5, 6 (total of 10)

Hull - 1, 1, 2, 2, 3, 3, 3, 3 (total of 8)

Beliveau - 1, 1, 2, 2, 2, 2, 3, 4, 4 (total of 9)


Take away the duplicates among the three players:

Crosby - 2, 2, 5, 6

Hull - 3, 3, 3

Beliveau - 2, 2, 4



IMO, hard to choose between Hull and Beliveau as to who has the most impressive Hart resume but Crosby has one more elite Hart finish so edge to him.


Here are their best Art Ross finishes:

Crosby - 1, 1, 2, 2, 3, 3, 3, 3, 5, 6, 10, 10 (total of 12)

Hull - 1, 1, 1, 2, 2, 2, 4, 5, 6, 7, 9 (total of 12)

Beliveau - 1, 2, 2, 3, 3, 3, 3, 4, 6, 8, 9, 10 (total of 12)


Take away the duplicates among the three players:

Crosby - 1, 3, 3, 3, 3, 5, 10, 10

Hull - 1, 1, 2, 4, 5, 7, 9

Beliveau - 3, 3, 3, 3, 4, 8, 9, 10


Hard to choose between the three players. Hull's higher end finishes are matched by Crosby's and Beliveau's more Top 3 finishes.

I would argue that Crosby's are statistically more impressive given he played in a league with five times as many players. In other words, I don't question Hull or Beliveau's 1st place finishes but once you get past first place, in general, 2nd or 3rd or 4th, etc... placements from a 30 team league are arguably stronger than from the O6.

This can be exemplified by looking at scoring finishes. From 1946/47 to 66/67, 3rd place in scoring was, on average, 18% behind the leader. From 1997/98 to 18/19, 3rd place was, on average, 13% behind the leader, a 30% difference. The same can be said about Top 5 scoring finishes where, on average, the 5th best scorer was 27% closer to the leader than the O6 5th place scorer.

I would say it is not unreasonable to say with some certainty that Crosby has the more impressive Hart and Art Ross finishes between the three players when this statistical reality is considered.

Click to read more...

 

[BigBadBruins7708]

The problem is finding a way to account for competition.

Crosby, through no fault of his own, has played in an era that lacks other consistently elite players outside of Ovechkin and Malkin.

The same thing is a driving factor behind Ovechkin having 8 Rockets. He is easily a top 3 goal scorer of all time, but gets a legacy boost from not having any consistent competition for the Rocket.

maybe this is just a side effect of the cap, guys who are borderline elite cant sustain it like they used to because of team turnover and not being able to hoard talent.

as for the 30 v 6 argument, you can't look at it purely on number of players. In the 6 team league, every line was like a 1st line in the current NHL, etc...it was the best 100-ish players.

Hell, from the late 50s into the mid 60s, 5 of the 6 teams iced a HOF goalie. So guys like Hull and Richard spent all year shooting against the likes of Hall, Plante, Sawchuk, Bower, Worsley

 

[daver]

as for the 30 v 6 argument, you can't look at it purely on number of players. In the 6 team league, every line was like a 1st line in the current NHL, etc...it was the best 100-ish players.

Hell, from the late 50s into the mid 60s, 5 of the 6 teams iced a HOF goalie. So guys like Hull and Richard spent all year shooting against the likes of Hall, Plante, Sawchuk, Bower, Worsley

The idea was to eliminate subjectivity altogether. I am not looking for pages and pages of debate over population or hockey outside of Canada or who played against more "HOF" talent.

Finishing in the Top 3 out of a group of 600 players > finishing Top out of a group of 120 players should be statistically inarguable. IMO, this should be a consensus opinion. The stats I presented clearly back up the assertion that comparing Top X finishes from dramatically different league sizes is very questionable statistically . For example, if there was consensus agreement that today's 30 team is as competitive the O6 then there would be consensus agreement that finishing Top 5 in a 30 team is statistically superior to finishing Top 5 in a six team league.

Does it mean that every 3rd place finish in the current league is superior to every 3rd place finish in the O6? No, but when you are looking at a large sample of Top 3 or 5 finishes, I don't think it is unreasonable to say that Crosby's multiple Top 5 finishes are superior to the same number of Top 5 finishes by a player from the O6.

 

[daver]

The problem is finding a way to account for competition.

Crosby, through no fault of his own, has played in an era that lacks other consistently elite players outside of Ovechkin and Malkin.

The same thing is a driving factor behind Ovechkin having 8 Rockets. He is easily a top 3 goal scorer of all time, but gets a legacy boost from not having any consistent competition for the Rocket.

maybe this is just a side effect of the cap, guys who are borderline elite cant sustain it like they used to because of team turnover and not being able to hoard talent.

Crosby has separated himself from the pack of the next best 10 - 20 scorers over his career in a way that is only clearly bettered by the Big Four. Having one or two more elite talents, even if that is a viable argument, would have very little effect on that.

 

[vadim sharifijanov]

The idea was to eliminate subjectivity altogether. I am not looking for pages and pages of debate over population or hockey outside of Canada or who played against more "HOF" talent.

Finishing in the Top 3 out of a group of 600 players > finishing Top out of a group of 120 players should be statistically inarguable. IMO, this should be a consensus opinion.

paragraphs #1 and 2 contradict each other

 

[daver]

paragraphs #1 and 2 contradict each other

Statistical realities are not subject to subjectivity. Finishing in the 99th percentile (Top 5 in today's league) > finishing in the 96th percentile (Top 5 in the O6).

That is the statistical reality and the numbers back it up. 5th place in today's league will usually be statistically superior to 5th place in the O6.

IMO, this argument should hold a lot more value than debating which eras were more competitive as that can never come close to being proven one way or the other.

I am not sure why this concept would not be common sense frankly.

 

[vadim sharifijanov]

I am not sure why this concept would not be common sense frankly.

because you are being highly subjective, obviously

 

[seventieslord]

daver, quantitative metrics like VsX exist today because a decade ago, guys in the ATD who were doing all the heavy duty work of comparing players and refining rankings over the years, realized that 10th place in 1910 or 1950 is not the same thing as 10th place in 1990.

It's definitely not perfect - I have my own problems with it - but I think it does a fine job of recognizing that quantity is more important than the placement. The question posed in this thread is legitimate, but the conclusion so far - that you have to make subjective mental adjusments - is not as strong as the answer(s) already developed IMO.

 

[daver]

because you are being highly subjective, obviously

"Finishing in a higher percentile is better" is 100% free of subjectivity.

 

[Midnight Judges]

Finishing in the Top 3 out of a group of 600 players > finishing Top out of a group of 120 players should be statistically inarguable.

It isn't though.

The league could double in size tomorrow - that wouldn't make the top 600 players in the world even....toppier.

What are the chances that someone outside the top 120 from the 60s or outide the top 600 in recent years was going to win an Art Ross, a goal scoring title, or a Hart? I'd say not impossible, but pretty slim.

The significant objective discriminator may be that far more players are getting top line TOI and preferential zone starts relative to the 60s.

 

[daver]

daver, quantitative metrics like VsX exist today because a decade ago, guys in the ATD who were doing all the heavy duty work of comparing players and refining rankings over the years, realized that 10th place in 1910 or 1950 is not the same thing as 10th place in 1990.

It's definitely not perfect - I have my own problems with it - but I think it does a fine job of recognizing that quantity is more important than the placement. The question posed in this thread is legitimate, but the conclusion so far - that you have to make subjective mental adjusments - is not as strong as the answer(s) already developed IMO.

What are the answers that are already developed? Can you provide those.

IMO, I have seen enough from the numbers to believe it is reasonable to give Crosby an edge over Hull and Beliveau given that their raw Art and Hart placings are extremely similar. I think if we rated all of their seasons mentioned in the OP using the VsX, Crosby clearly comes out on top in scoring finishes, which should also translate to a superior Hart record.

This is a similar edge I would give to players who were clearly carrying their team's offense, had clearly less talented linemates, faced harder defensive matchups, had a clearly superior defensive, or a clear physical edge.

 

[MadLuke]

Hell, from the late 50s into the mid 60s, 5 of the 6 teams iced a HOF goalie. So guys like Hull and Richard spent all year shooting against the likes of Hall, Plante, Sawchuk, Bower, Worsley

Like everyone else in the league did, that a good argument when talking for how many ppg they achieved to do, but less about their ability to finish in the league top 5/10.

It must almost in a trivial way have been easier to do in a 6 team league once you were a proven player getting first line/powerplay minute and on a good team versus in a 30 team league.

You were competing against just what 17 other first liner to get into the league top 10 (and if you were on the 3-4 good team with a good powerplay arguably with only 12 other first liner), that has I imagine a smaller effect on how hard to win the Ross, but at one point must have for getting in the top 5 and certainly top 10. And it was quite harder to achieve and prove yourself to get into that position, but once you were the chance of repeating it was higher.

 

[daver]

Here is a look at the 10th place finishers:

From 1954/55 to 68/69, 10th place in scoring was, on average, 50% behind the leader. From 1997/98 to 18/19, 10th place was, on average, 30% behind the leader, a 68% difference.

 

[Midnight Judges]

The significant objective discriminator may be that far more players are getting top line TOI and preferential zone starts relative to the 60s.

The other significant objective discriminator is the fact that hockey is international now, whereas it was almost only Canadian back then, and the result is that the size of the talent pool is way larger. The likely outcome is that there are more generational players, more elite players, more great players, more very good players, and far fewer weak links and pylons playing defense.

This makes it far more difficult to separate from the pack.

As a result, I would argue that Crosby's Hart finishes are actually far superior to Hull's and Beliveau's, and second only to Ovechkin out of this generation.

 

[MadLuke]

The other significant objective discriminator is the fact that hockey is international now, whereas it was almost only Canadian back then, and the result is that the size of the talent pool is way larger. The likely outcome is that there are more generational players, more elite players, more great players, more very good players, and far fewer weak links and pylons playing defense.

But that is in good part balanced by how small the elite talent pool got in many place at the same time (Canada specially).

 

[Midnight Judges]

But that is in good part balanced by how small the elite talent pool got in many place at the same time (Canada specially).

Nope. Not even close.

Canada's population has more than doubled since then. Some guys note that births have actually gone down, but births don't really measure it accurately either. For example, it wouldn't count Brett Hull.

Even if you do think births are indicative, Canada's birth number peaked at roughly 480,000 in 1960. (Note that practically nobody from that peak would have been old enough to play in the 50s and 60s - Beliveau's and Hull's time. Hull left the NHL in 1972, just as the youngest people from the beginning of the spike would have turned 18 and entered the NHL. He simply did not compete with them.)

Canada's average births from 1980-1996 (when the bulk of today's player were born) is roughly 380,000. A difference of 100,000 (roughly 50,000 of which are boys).

canada births by year

We know for a fact that the USA hockey playing population has been north of 420,000 since the turn of the century - with roughly 69% of them being youth boys (NHL feeders). The hockey growth in the USA alone blows the doors off the difference in Canadian births five to six times over (actually more than that), and we're not even counting Europe yet - which likely has a raw total of players equal to or greater than the US.

https://cdn1.sportngin.com/attachme...618.175602984.1558572104-813979025.1558572103

 

[MadLuke]

Population is not that relevant, playing age is the only population metric interesting and even that, population pool playing at a level/environment that make entering the NHL possible size is what matter and that seem to have shrinked a lot (more because of cost, popularity in different activity/sports than lest youth).

 

[Midnight Judges]

If cost is the primary factor, then participation rates would have increased, not decreased.

poverty rate canada by year

 

[seventieslord]

What are the answers that are already developed? Can you provide those.

The answer, you basically discovered on your own in your next post. There's no need to make a mental adjustment to what it means to be 10th place in 1960 compared to 2000, when you can see what the numbers say.

Here is a look at the 10th place finishers:

From 1954/55 to 68/69, 10th place in scoring was, on average, 50% behind the leader. From 1997/98 to 18/19, 10th place was, on average, 30% behind the leader, a 68% difference.

...however, this is a real interesting way to put it. If we are treating 1st place as the "standard" in this case, then we are talking about factors of 0.70 and 0.50, and the difference should, in no way, be described as 68%. It would be correct to say that one is 29% less impressive than the other, or one is 40% more impressive than the other.

That said, no one compares the resumes of Beliveau, Hull and Crosby based on the seasons they were ~10th in scoring. Those are Beliveau's 10t-13th best seasons, Hull's 11th-12th, and Crosby's 11th. We are concerned with their large numbers of top-3 finishes (they each have six to eight of them). And while it is a virtual guarantee that comparing a modern, larger-league player who was 10th in scoring to an older, smaller-league player who did the same, the modern player can be shown to have done something more statistically impressive. But that is by no means a guarantee when talking about top-3 finishes, thanks to the wider gaps that tend to exist at the top of any leaderboard. A quantitative measure (like VsX) is much more appropriate to gauge such a thing.

I mean your premise isn't absurd. These three players are now all in the conversation for the best player after the big-4. And I'm not saying "check the VsX, it shows you are clearly wrong", not by a long shot. Their 10 year scores are now 100 (Hull), 99 (Crosby) and 96 (Beliveau). So on the basis of prime regular season offense he really gives up nothing to these guys. Demonstrating Crosby to be better than either player probably requires a deeper analysis than "look how equal these achievements are, and one of them did it in a modern league".

 

[daver]

The answer, you basically discovered on your own in your next post. There's no need to make a mental adjustment to what it means to be 10th place in 1960 compared to 2000, when you can see what the numbers say.



...however, this is a real interesting way to put it. If we are treating 1st place as the "standard" in this case, then we are talking about factors of 0.70 and 0.50, and the difference should, in no way, be described as 68%. It would be correct to say that one is 29% less impressive than the other, or one is 40% more impressive than the other.

That said, no one compares the resumes of Beliveau, Hull and Crosby based on the seasons they were ~10th in scoring. Those are Beliveau's 10t-13th best seasons, Hull's 11th-12th, and Crosby's 11th. We are concerned with their large numbers of top-3 finishes (they each have six to eight of them). And while it is a virtual guarantee that comparing a modern, larger-league player who was 10th in scoring to an older, smaller-league player who did the same, the modern player can be shown to have done something more statistically impressive. But that is by no means a guarantee when talking about top-3 finishes, thanks to the wider gaps that tend to exist at the top of any leaderboard. A quantitative measure (like VsX) is much more appropriate to gauge such a thing.

I mean your premise isn't absurd. These three players are now all in the conversation for the best player after the big-4. And I'm not saying "check the VsX, it shows you are clearly wrong", not by a long shot. Their 10 year scores are now 100 (Hull), 99 (Crosby) and 96 (Beliveau). So on the basis of prime regular season offense he really gives up nothing to these guys. Demonstrating Crosby to be better than either player probably requires a deeper analysis than "look how equal these achievements are, and one of them did it in a modern league".

Thanks for the response.

I agree completely that Top 3 finishes hold a lot more weight than Top 10 finishes, and that the difference in Top 3 finishes is closer than the difference in Top 10 finishes over the two eras.

Are you able to provide a link to the 10 year VsX scores you referenced?

Cheers

 

[daver]

...however, this is a real interesting way to put it. If we are treating 1st place as the "standard" in this case, then we are talking about factors of 0.70 and 0.50, and the difference should, in no way, be described as 68%. It would be correct to say that one is 29% less impressive than the other, or one is 40% more impressive than the other..

You are absolutely right. In relation to 3rd place, the factors become 0.88 and 0.84, somewhat of a marginal difference but arguably an edge no less, generally speaking.

 

[unknown33]

Finishing in the Top 3 out of a group of 600 players > finishing Top out of a group of 120 players should be statistically inarguable. IMO, this should be a consensus opinion.

No, this is wrong. It's not like the 120th - 600th best player has a chance at finishing in the Top 3. You are assuming that NHL expansion automatically creates more elite players which there is no evidence for.

 

[daver]

No, this is wrong. It's not like the 120th - 600th best player has a chance at finishing in the Top 3. You are assuming that NHL expansion automatically creates more elite players which there is no evidence for.

I am assuming that the 600 players, or more specifically the 180 Top 6 forwards, are as equally elite as the 36 Top 6 forwards in the O6. This is the most reasonable way to approach this. An assumption that there are more elite players on a relative basis in today's league vs. the O6 is equally unreasonable.

Again, the point of the thread is to look at this in a strict statistical sense, free of opinion on which era was more competitive.

So there are six times as many elite forwards in the game today vs. the O6. Finishing in 3rd vs. a field of 180 places you in a higher percentile than finishing in 3rd vs. a field of 36. The former is clearly farther away from the median.

In other words, generally speaking, a 3rd place finish in today's league is more likely to finish in 3rd or better if the season had been played against a field of 36.

I am positive that a VsX analysis would confirm that 3rd place in the O6 from 1947 to 1967 scores lower than 3rd place from 1997 to 2019.

 

[Michael Farkas]

This might not respond to anything in particular...but this is vaguely related to the 180 top six forwards thing...I just went through the rosters, there are about 17-20 #1 d-men in the league. I'm not sure that there are 180 top six forwards in the league...I'm pretty sure there isn't and I don't think it's gonna be terribly close...

 

[daver]

This might not respond to anything in particular...but this is vaguely related to the 180 top six forwards thing...I just went through the rosters, there are about 17-20 #1 d-men in the league. I'm not sure that there are 180 top six forwards in the league...I'm pretty sure there isn't and I don't think it's gonna be terribly close...

Are you referring to your subjective assessment of whether a player should or should not be considered #1 D-man or a Top 6 forward?

Setting aside that this is simply your opinion, one which could be countered with an opinion that there were even less than 36 Top 6 forwards in the O6, this does not change the statistical reality that finishing 3rd out of a larger group is more impressive. Even if you opined there were only 120 Top Six forwards, this doesn't close the door to statistical considerations.