Post-consolidation VsX Benchmarks

[Sturminator]

This is excellent work, bluesfan. I agree with TDMM that at least in the post-expansion era, comparing individual scoring against these benchmarks is probably the best way to get a sense of how players scored against some normalized league-wide standard. This is something approaching the OPS+/ERA+ system that has long been used in baseball, at least for the post-expansion era.

 

[BraveCanadian]

Not surprised to see 1992-93 as the highest benchmark - it really should be - that year was really wacked. I knew 1995-96 was a year where scoring spiked because the NHL had a major crackdown on obstruction during the regular season (only to forget about it entirely come playoffs), but it still surprises me to see it so closely resemble a typical season in the 1980s for scoring among top 18 players.

Right it was a very strange year where for some reason a lot of scoring concentrated to the top.

 

[Sturminator]

Right it was a very strange year where for some reason a lot of scoring concentrated to the top.

Were holding and obstruction called really tight that year, perhaps? Tighter rules enforcement would lead to more space for the superstars at even strength and more penalties called would lead to more powerplay minutes for those same superstars. I don't know if this was really the case, but it is the simplest explanation I can come up with. I'd be interested to see if the upper tier scorers in the league had disproportionate spikes in their powerplay scoring in 1992-93.

 

[BraveCanadian]

Were holding and obstruction called really tight that year, perhaps? Tighter rules enforcement would lead to more space for the superstars at even strength and more penalties called would lead to more powerplay minutes for those same superstars. I don't know if this was really the case, but it is the simplest explanation I can come up with. I'd be interested to see if the upper tier scorers in the league had disproportionate spikes in their powerplay scoring in 1992-93.

I still want to say it also had to do with the tv timeouts being introduced but that is probably just me

 

[Sturminator]

I still want to say it also had to do with the tv timeouts being introduced but that is probably just me

Oh, maybe so. The TV timeouts have allowed 1st liners to take more even-strength shifts, that's for sure. I'm not sure if this is an adequate explanation for the spike in scoring, but it may be a piece of the puzzle.

 

[bluesfan94]

I'll start work today on the number of scorers above each benchmark. I plan on using percents 150, 125, 110, 105, 100, 95, 90, etc all the way down to 50%. I'll also do raw data counts (i.e number of scorers with > 80 points) using 200, 165, 150, 145, 140, etc all the way down to 50 pts. I'm only going to do post expansion to start with, but I may continue all the way to early NHL if I get bored/interested enough

 

[Sturminator]

I'll start work today on the number of scorers above each benchmark. I plan on using percents 150, 125, 110, 105, 100, 95, 90, etc all the way down to 50%. I'll also do raw data counts (i.e number of scorers with > 80 points) using 200, 165, 150, 145, 140, etc all the way down to 50 pts. I'm only going to do post expansion to start with, but I may continue all the way to early NHL if I get bored/interested enough

Fantastic. seventies' spreadsheet data may save you some (or a lot of) time, if you're interested. I very much look forward to seeing the data.

 

[TheDevilMadeMe]

Were holding and obstruction called really tight that year, perhaps? Tighter rules enforcement would lead to more space for the superstars at even strength and more penalties called would lead to more powerplay minutes for those same superstars. I don't know if this was really the case, but it is the simplest explanation I can come up with. I'd be interested to see if the upper tier scorers in the league had disproportionate spikes in their powerplay scoring in 1992-93.

There have been several threads on the 1992-93 season on the history board and why scoring was so crazy:

http://hfboards.mandatory.com/showthread.php?t=251123&highlight=1992
http://hfboards.mandatory.com/showthread.php?t=363844&highlight=1992
http://hfboards.mandatory.com/showthread.php?t=509960&highlight=1992
http://hfboards.mandatory.com/showthread.php?t=679988&highlight=1992

Seems like a perfect storm of the league increasing the schedule from 80 to 84 games, Ottawa and San Jose getting shafted by the expansion rules and being fodder for the rest of the league, and a major obstruction crackdown on top of it.

 

[Sturminator]

Interesting. It seems my intuition was right, but I forgot about expansion and the increase in the length of the schedule to 84 games. Thanks, Devil.

 

[bluesfan94]

Hey Sturm, do you have benchmarks for VsX in terms of goals and assists?

 

[bluesfan94]

Fantastic. seventies' spreadsheet data may save you some (or a lot of) time, if you're interested. I very much look forward to seeing the data.

I would love the spreadsheet.

 

[Sturminator]

Hey Sturm, do you have benchmarks for VsX in terms of goals and assists?

I haven't done that work, sorry.

 

[Sturminator]

I would love the spreadsheet.

PM me your e-mail, and I'll forward you the spreadsheet he sent me. I promise I'll only register you for one hot housewives of the midwest newsletter.

 

[TheDevilMadeMe]

What would be the purpose of establishing benchmarks for goals and assists separately?

 

[bluesfan94]

What would be the purpose of establishing benchmarks for goals and assists separately?

You could rate goalscoring and playmaking somehow. I'm not sure if there is one necessarily, I just figured if he had it, I could include it.

 

[Sturminator]

Looking through the benchmarks I set, I have noticed that the benchmark is almost always attached to the name of an established offensive star - someone who is considered a high-end ATD scorer - with the exception of the following period:

1939-40:
2. Dumart/Bauer - 43

1940-41:
2. [various players] - 44

1941-42:
2. Patrick - 54

1942-43:
1. D. Bentley - 73
2. Cowley - 72
3. M. Bentley - 70
4. Patrick - 61
5. Taylor - 60
5. Carr - 60
...average as benchmark: 66

1943-44:
2. D. Bentley - 77

1944-45:
1. Lach - 80
2. M. Richard - 73
3. Blake - 67
4. Cowley - 65
5. Kennedy - 54
5. DeMarco - 54
5. Carveth - 54
5. Mosienko - 54
...average as benchmark: 63

1945-46:
2. G. Stewart - 52

1946-47:
1. M. Bentley - 72
2. M. Richard - 71
3. Taylor - 63
...benchmark = Taylor: 63

1947-48:
2. O'Connor - 60

What does this mean for our benchmarking? Well, many people have been suspicious of the 40's for a long time as a period when NHL hockey was simply not being played at a particularly high level, irrespective of the war. The basic assumption behind the VsX methodology is that after the true superstars/outliers in any given season, there is a tier of "star" scorers, and that the quality of this tier has remained relatively consistent throughout history, although the league has grown in size and depth. I think this assumption holds true pretty well in the VsX system for most of the period of post-consolidation NHL history, but I'm not sure it is functioning correctly in the 1940's.

We try to benchmark our scoring against a "star" scorer, but when that alleged star is Woody Dumart, Lynn Patrick, Gaye Stewart or Billy Taylor...I'm not sure if the idea behind the benchmark is really working as intended. One argument would be that these guys had career years in the seasons in which they serve as the benchmark and that their scoring feats are legitimately up to the star standard, but I'm not so sure. That's an awful lot of "career years" all within a short timeframe. Something still smells about the 1940's. I am considering a Vs1 standard for this timeframe to reflect what I believe was a lower standard of hockey. I realize that this is an arbitrary judgment and I don't like it much, but I don't know how better to deal with this timeframe, which I feel is still somewhat positively distorted in comparison to other eras due to the shallowness of the league talent pool.

Comments?

 

[TheDevilMadeMe]

As far as I know, nobody has addressed a big issue with using the percentage method - it is highly influenced by league size.

From the 2011 assassination thread:

To buy it as a valid comparison between eras, I want to see just how many players from the post expansion era rank ahead of certain benchmarks (5th, 10th, 20th) pre expansion.

Just to use 50% as a benchmark:

1930: 25
1940: 41
1950: 34
1960: 28
1970: 69
1980: 68
1990: 83
2000: 102
2010: 99

these were randomly-selected seasons based on nothing but the start of a decade. the results would likely look smoother if three-year periods were chosen, or something like that. (this is based on outliers being removed)

the league expanding has a lot to do with the results looking like this. However, the talent pool has been growing consistently as well, and the above numbers don't do too bad a job of approximating what that growth has looked like: about 50% since 1970, about 4X since 1930. (also, keep in mind that since 1970, more defensemen would meet this benchmark than before, so the numbers get a slight upward boost from that)

it still makes the most logical sense of any system - what percentage of the best did the player score? If the assumption is that 100 one year is as good as 80 in another, then it should follow that 50 and 40 are also equal in these two seasons.

Those numbers make perfect sense in the post expansion era on, but the 1960 numbers are just brutally low, considering the early 60s were actually a highpoint of league talent level.

The benchmark looks beautiful from 1970 on, considering the influx of European talent,, but kind of all over the place beforehand. I'm 99% sold on this being the best method to compare offense post-expansion (as long ad you dsicount Orr teammates just a tad), but not sold on it's applicability to a smaller league just yet.

I know seventieslord acknowledges the problem and basically adds something like a 7% fudge factor in favor of pre-expansion players (not on a season by season basis, but over their average).

 

[Sturminator]

As far as I know, nobody has addressed a big issue with using the percentage method - it is highly influenced by league size.

From the 2011 assassination thread:

Those numbers make perfect sense in the post expansion era on, but the 1960 numbers are just brutally low, considering the early 60s were actually a highpoint of league talent level.

The benchmark looks beautiful from 1970 on, considering the influx of European talent,, but kind of all over the place beforehand. I'm 99% sold on this being the best method to compare offense post-expansion (as long ad you dsicount Orr teammates just a tad), but not sold on it's applicability to a smaller league just yet.

I know seventieslord acknowledges the problem and basically adds something like a 7% fudge factor in favor of pre-expansion players (not on a season by season basis, but over their average).

First of all, I'm pretty sure the numbers seventies generated were using a straight Vs2 benchmark, which is obviously going to create some serious distortions in certain eras, the 50's being one of them. I would imagine that a good deal more players reach the 50% level when compared to the more liberal (and I think less distorting) VsX benchmark. We'll have to see when we get back the results of bluesfan's work.

Although I'm not sure how much it matters. One of the central assumptions of any percentage-based system is that it is only valid for comparing players who enjoyed equal opportunities to score - basically 1st liners who played on their respective teams' powerplay first units. 24 players crossing the 50% line in 1960 is still more than the 18 who were actually first liners in that season. I agree with you completely that Vs2 or VsX or VsAnything is not going to do a good job of measuring the output or talent of depth scorers in the O6 era, but that's been known for a long time. Guys like Ralph Backstrom and Phil Goyette just can't be fairly evaluated using this methodology; it's one of the constraints of the system.

 

[TheDevilMadeMe]

First of all, I'm pretty sure the numbers seventies generated were using a straight Vs2 benchmark, which is obviously going to create some serious distortions in certain eras, the 50's being one of them. I would imagine that a good deal more players reach the 50% level when compared to the more liberal (and I think less distorting) VsX benchmark. We'll have to see when we get back the results of bluesfan's work.

Although I'm not sure how much it matters. One of the central assumptions of any percentage-based system is that it is only valid for comparing players who enjoyed equal opportunities to score - basically 1st liners who played on their respective teams' powerplay first units. 24 players crossing the 50% line in 1960 is still more than the 18 who were actually first liners in that season. I agree with you completely that Vs2 or VsX or VsAnything is not going to do a good job of measuring the output or talent of depth scorers in the O6 era, but that's been known for a long time. Guys like Ralph Backstrom and Phil Goyette just can't be fairly evaluated using this methodology; it's one of the constraints of the system.

It still hurts first liners in the O6 who would have spent few seasons of their career on lower lines.

Don't you think at the height of the O6 there was a lot more scoring talent than there was available scoring positions on 6 teams?

I'm probably not explaining it well, but the increase in scoring opportunities in a larger league makes it harder to repeat high rankings, but makes it easier to repeat high percentage scores.

This is a simplification, but in the O6, there were 36 roster spots available for a forward on the top 2 lines, and only 18 of them got powerplay time since the first unit played the full PP (realistically, probably more like 20-24 due to forwards playing the point). In modern times, there are 180 roster spots on the top 2 lines and basically all of them receive PP time since teams roll 2 units.

 

[Sturminator]

It still hurts first liners in the O6 who would have spent few seasons of their career on lower lines.

Don't you think at the height of the O6 there was a lot more scoring talent than there was available scoring positions on 6 teams?

I'm probably not explaining it well, but the increase in scoring opportunities in a larger league makes it harder to repeat high rankings, but makes it easier to repeat high percentage scores.

This is a simplification, but in the O6, there were 36 roster spots available for a forward on the top 2 lines, and only 18 of them got powerplay time since the first unit played the full PP (realistically, probably more like 20-24 due to forwards playing the point). In modern times, there are 180 roster spots on the top 2 lines and basically all of them receive PP time since teams roll 2 units.

Yes, of course. It is really only accurate for the big stars of the pre-expansion era because, as you say, scoring opportunities were much more precious before expansion. This is most specifically a problem for Habs depth scorers, who were often denied opportunities due to the team's incredible offensive talent. This would also be problematic for depth scorers in Detroit who got stuck behind the Production Line at its peak. For Leafs, Bruins, Hawks and Rangers skaters, it seems much less problematic, as those teams were rarely deep enough to deny strong offensive players scoring opportunities. This data definitely has to be taken in context when dealing with non-elite pre-expansion scorers.

The thing is...without detailed icetime and special teams usage breakdowns, it is impossible to systematically measure offensive efficiency without the assumption of equal opportunity. Your point is that this assumption becomes increasingly narrow the smaller the league, and you are right. Without more data, however, I doubt this problem can be remedied in a systematic way. We must simply accept that the system does not apply to certain players. It is a constraint of the system, rather than an indication of its inaccuracy within the universe of players to which it is meant to apply.

 

[TheDevilMadeMe]

To put it another way, I think scores above a certain threshold (80%? 75%?) are mostly translatable across league size, but once you get below that... A scoring line player having an off-year in a 30 team league might score a 65, but in a 6 team league, he might get dropped to a lower line and score a 50 without being any different a player.

 

[Sturminator]

To put it another way, I think scores above a certain threshold (80%? 75%?) are mostly translatable across league size, but once you get below that... A scoring line player having an off-year in a 30 team league might score a 65, but in a 6 team league, he might get dropped to a lower line and score a 50 without being any different a player.

I think this is certainly true for scoringliners in Montreal. I really doubt it applies all that much to players in New York, Chicago and Boston for most of the O6 period, though. Detroit would be problematic for about half of the era, and the Leafs for maybe a little while. It is all quite context specific.

Actually, for most non-Hab players, I think the biggest problem of limited opportunity is a different one than what you have described. I don't see any reason to give a guy like Vic Stasiuk a mulligan for any of his seasons in Boston, for example, because the Bruins were never deep in scoring talent, and if he wasn't making it on a Bruins scoringline, it was because he was sucking. But Stasiuk was a relatively well-rounded player. I think it is more likely the somewhat crappy one-way guys like his sometimes linemate, Bronco Horvath, who were most hurt by the size of the league in the O6 era. Horvath got punted from the NHL pretty quickly for being a huge defensive liability, whereas in the post-expansion world, some team would have almost certainly picked him up and let him do his thing. Horvath simply didn't get as many opportunities, on a career level, as he would have gotten in a larger league, but that is not really a season-to-season issue which should cause distortions in a percentages system.

I'd be interested to see if there is some sort of plateau and then a dropoff in percent scoring during the O6 era - like a bunch of players score in the 90% and 80% range, and then very few in the 70% to 60% range before we start seeing a bunch of guys in the 50% and 40% range again. If this is the case, it would seem to support your argument that there were a few stars who got all of the scoring opportunities, and then a bunch of guys who were basically treated as grinders. If scoring levels were reached in a relatively smooth progression, then I dunno.

 

[bluesfan94]

This is the VsX Thresholds from expansion forward. I'll finish the rest of the years and add Vs18 Thresholds sometime this week

 

[seventieslord]

Looking through the benchmarks I set, I have noticed that the benchmark is almost always attached to the name of an established offensive star - someone who is considered a high-end ATD scorer - with the exception of the following period:



What does this mean for our benchmarking? Well, many people have been suspicious of the 40's for a long time as a period when NHL hockey was simply not being played at a particularly high level, irrespective of the war. The basic assumption behind the VsX methodology is that after the true superstars/outliers in any given season, there is a tier of "star" scorers, and that the quality of this tier has remained relatively consistent throughout history, although the league has grown in size and depth. I think this assumption holds true pretty well in the VsX system for most of the period of post-consolidation NHL history, but I'm not sure it is functioning correctly in the 1940's.

We try to benchmark our scoring against a "star" scorer, but when that alleged star is Woody Dumart, Lynn Patrick, Gaye Stewart or Billy Taylor...I'm not sure if the idea behind the benchmark is really working as intended. One argument would be that these guys had career years in the seasons in which they serve as the benchmark and that their scoring feats are legitimately up to the star standard, but I'm not so sure. That's an awful lot of "career years" all within a short timeframe. Something still smells about the 1940's. I am considering a Vs1 standard for this timeframe to reflect what I believe was a lower standard of hockey. I realize that this is an arbitrary judgment and I don't like it much, but I don't know how better to deal with this timeframe, which I feel is still somewhat positively distorted in comparison to other eras due to the shallowness of the league talent pool.

Comments?

You could be harder on those years by being easier on years where there was a huge concentration of top-end talent in the top-5 in scoring... like 1980-1996, especially.

First of all, I'm pretty sure the numbers seventies generated were using a straight Vs2 benchmark, which is obviously going to create some serious distortions in certain eras, the 50's being one of them.

that could be. I have no idea how long ago that post was from, and whether I did it manually or using my spreadsheet.

 

[Sturminator]

This is the VsX Thresholds from expansion forward. I'll finish the rest of the years and add Vs18 Thresholds sometime this week

I should have thanked you for this earlier, but anyway, thanks, and good work. I've gone through the data you generated and put together threshhold averages for each "decade" since expansion. Actually, I grouped the years in a somewhat queer fashion. I wanted to use nice, clean, ten year blocks, but as there have been 44 seasons since expansion, this was impossible. Instead, for the purposes of averaging, I elected to bundle the first 13 seasons together, then 10 seasons for the 80's and 90's, and then the last 11 seasons for the 00's. I don't think it really distorts the averages, but it should be noted. Anyway, here is the data:

Years|>150%|>125%|>110%|>105%|>100%|>95%|>90%|>85%|>80%|>75%|>70%|>65%|>60%|>55%|>50%
2000-01 -- 2011-12 |0|0|0|1|1|3|6|9|14|21|30|42|56|74|98
1990-91 -- 1999-00 |0|0|1|1|1|4|5|9|13|18|25|34|44|59|77
1980-81 -- 1989-90 |1|1|1|1|1|4|4|6|9|12|19|26|37|51|68
1967-68 -- 1979-80 |0|0|1|2|2|3|5|7|9|14|19|25|34|44|58

These are interesting results. I think the >50% numbers are the most useful for judging overall league depth. If you believe in the system, they suggest a couple of things:

- that the increase in league depth proceeded at a fairly smooth rate from the 70's through the 80's and 90's, but that league depth has increased quite a bit in the last decade. This actually makes a lot of sense, because it was well into the 90's before European stars became fully integrated into the NHL, and the 90's was a comparatively crappy era for Canadian talent. It is only in the past decade that averaged numbers would show the full result of the incoming European talent.

- that the increase in league depth from the 1980's until the present is approximately 50%. That is, that the influx of European talent has not actually doubled the NHL talent pool.

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

Interestingly, this data fits almost perfectly with my long-held opinion on the influence of European players on the size of the NHL talent pool. Of course, this coincidence may seem like evidence that I have simply built this system of benchmarks to confirm my own biases, though I would have to be pretty smart to reverse-engineer my methodology in that way. I assure you, I am capable of no such thing. seventies generally thinks that I am too hard on 80's players, but going easier on them would make the difference in the size of the league talent pool from the 1980's to present look even smaller, and I'm not sure if I buy that. An approximate 50% increase in the size of the NHL talent pool over that period smells about right to me. Anyway, your mileage may vary.