1912 to 1926 Consolidation Numbers

[Dreakmur]

As has been discussed, it's time to share my imperfect project for destruction

Here is the link to the document: Consolidation Project - Google Sheets

The process:
Since this was accumulated many years ago, with much less access to scoring records, I had to do this in a somewhat patchwork method. I added each player in manually. It is possible that there are some players who I have missed, especially guys who were much lower profile and had only 1 or 2 good seasons.

The math that was involved:
#1 - Games played equalized. Points were essentially counted as a per game stat and I made each league play the same number of games.

#2 - Goals per game. This is based on the possibly faulty assumption that each league was exactly equal. I took the difference in scoring rates, and gave all the players from the lower scoring league a boost equal to the difference in league-wide rates.

#3 - Adjusted assists. This was implemented after I noticed how may more assists were awarded in the PCHA one season. I thought that would most likely be due to differences in statistical tracking rather than differences in how may assists actually happened. This calculates how many assists per goal were awarded in each league, and makes the adjustment to make them equal. Again, this adjustment is based on the assumption that the leagues were equal.

Fire away!

Click to read more...

 

[TheDevilMadeMe]

Thank you for posting this

So the biggest issue I've always had with your post-consolidation VsX equivalent is that it doesn't take into account the fact that assists were handed out very rarely back then, much less than 1 assist per goal.

When 1.7-1.8 assists have been handed out per game basically since WW2, I think you would either need to normalize assists to a modern ratio, or just ignore assists entire and make this a VsX for goals only.

As such, I've always felt that it overrated goal scorers (like Malone and Lalonde) and underrated playmakers (like Taylor and Nighbor, though I guess Taylor was somewhat balanced)

 

[ResilientBeast]

My criticisms I guess would be

1) 6 man vs 7 man hockey, the dynamics of it were just different so trying to equate the two on one scoring table just feels wrong to me (despite my previous efforts to do so)
2) Sure the assists were handed out in a different frequency...but the adjustment is kinda creating them out of thin air to try and equate the two.

 

[Dreakmur]

Thank you for posting this

So the biggest issue I've always had with your post-consolidation VsX equivalent is that it doesn't take into account the fact that assists were handed out very rarely back then, much less than 1 assist per goal.

When 1.7-1.8 assists have been handed out per game basically since WW2, I think you would either need to normalize assists to a modern ratio, or just ignore assists entire and make this a VsX for goals only.

As such, I've always felt that it overrated goal scorers (like Malone and Lalonde) and underrated playmakers (like Taylor and Nighbor, though I guess Taylor was somewhat balanced)

There was no forward passing at this time, right? That led to much more individual rushes, I would assume.

 

[Dreakmur]

My criticisms I guess would be

1) 6 man vs 7 man hockey, the dynamics of it were just different so trying to equate the two on one scoring table just feels wrong to me (despite my previous efforts to do so)

What impact does this have on scoring?

2) Sure the assists were handed out in a different frequency...but the adjustment is kinda creating them out of thin air to try and equate the two.

So, do you think one league had more passing, and as a result more assists?

 

[ResilientBeast]

What impact does this have on scoring?



So, do you think one league had more passing, and as a result more assists?

I'll respond to both of these questions with more depth than I have available to me right now soonish.

 

[BenchBrawl]

Assists are the heart of the problem.

Starting with a Goal VsX is not a bad idea, just to get our feet wet. Using #1 and #2, then implementing the classic VsX formula, we could come up with this in no time.

The assists problem must be discussed more throughoutly.

 

[seventieslord]

I personally don't have a problem with the infrequency of assists in those times. The total points numbers as we have them are the best approximation we have for what kind of contribution a player made to the offense. that said, I think dreakmur's strategy of adjusting assists to normalize between the PCHA and NHA/NHL is probably a good idea.

My main criticism is about the assumption that the leagues were of equal competition level. Generally they were, from a top-to-bottom standpoint, but particularly at the top-end, with the players we really care about, the short schedules led to frequent outliers in one league or another. By adjusting the league GPG to normalize them to eachother, I think you are unwittingly calling the leagues equal. Just like I could do the same thing to the NHL and QMJHL for 2020 - the resulting blended list would be a fair summation of who dominated their respective league the most, but would say nothing about the actual competition levels.

It's ok to be subjective about this. Just by looking at the top-3 scorers in the east and the west, you can usually get a good idea of which player was the outlier who really tore their league apart and deserves a score above 100. Some years there isn't one - just give the league leaders 100 each.

I've seen dreakmur post his results more than a few times, and compared to my own, and even though we went about it differently, they're never materially different. A few higher here, a few lower there, typically about the same 7-year score. I prefer my method but I'm also not all that invested in this because if you guys put his numbers through the ringer and it passes muster then I know I can trust the results that become standardized for ATD use are practically the same as they'd be if I debated this very strongly and had an impact on them.

 

[TheDevilMadeMe]

The problem with taking the assist numbers at the time at face value is that it makes it look like nobody was a good playmaker, not when comparing them to NHL-era players, especially post-war players.

At the very minimum, you have a situation with zero secondary assists, that alone disadvantages certain players.

You also have a situation where Frank Nighbor looks like a really awful pick in the top 50.

This is Sturminator's post where he used top 10 finishes in goals and assists separately to compare Nighbor to Bobby Clarke and concluded that their offense is similar:

After all you now know about Frank Nighbor, if you still think Bobby Clarke's offensive resume is so much better that's probably because you've never taken a critical look at Clarke, himself. When I equated Nighbor's offensive accomplishments to Clarke's long ago, I assumed that the reader was intimately familiar with Clarke's career, although I see that in this case that is not true. I suppose if I want to make this comparison, it was always going to come to this.

Here is Bobby Clarke's complete regular season top-20 scoring resume, with top-10 finishes bolded:

Points: 10th (71-72), 2nd (72-73), 5th (73-74), 6th (74-75), 2nd (75-76), 8th (76-77), 8th (77-78)

Goals: 13th (71-72), 16th (72-73), 12th (73-74)

Assists: 17th (71-72), 3rd (72-73), 10th (73-74), 1st (74-75), 1st (75-76), 6th (76-77), 4th (77-78), 9th (78-79), 8th (79-80), 5th (82-83)

...and now for Frank Nighbor, I'll be as uncharitable to his scoring finishes as possible, and simply multiply his actual placement by two for every season in which he finished in the top-10 in any category (completely ignoring finishes outside of the top-10). This means that in the cases in which Nighbor finished first, he'll have a 2nd place finish recorded on the following list. Here it is:

Points: 8th (12-13), 6th (14-15), 14th (15-16), 2nd (16-17), 18th (17-18), 4th (18-19), 6th (19-20), 10th (20-21), 16th (23-24), 16th (25-26)

Goals: 8th (12-13), 6th (14-15), 14th (15-16), 2nd (16-17), 6th (18-19), 6th (19-20), 10th (20-21), 20th (23-24)

Assists: 12th (14-15), 12th (17-18), 4th (18-19), 2nd (19-20), 4th (20-21), 16th (21-22), 6th (23-24), 2nd (25-26)

*the reader should note that no assist totals are available (to me, at least; I am not an SIHR member) other than PCHA totals (one season for Nighbor - 14-15) before the inaugural NHL season in 1917-18. Nighbor's "scoring" finishes before that time reflect only his goal-scoring. No adjustment has been made to reflect where he would have placed in a modern-style scoring table given that he was clearly the second greatest playmaker of his era behind Taylor.*

They come out quite nearly even. Clarke has one more top-10 scoring season, but Nighbor 4 more in the top-20, and this is using only goal-scoring to calculate Nighbor's points for the better part of his prime years. Nighbor kills Clarke in goals, and probably after correction for the missing data comes fairly close though short of Clarke in assists. This is using very stringent criteria for Nighbor - multiplying all of his finishes by 2 and not adjusting for how badly he was hurt in the points race by the lack of assist totals - and he still comes out with nearly identical offensive value to that of Bobby Clarke.

Now to address questions of competitive quality. The top peak forwards against whom Nighbor competed were:

Lalonde, Malone, Taylor, Foyston, MacKay, Broadbent, Denneny, Pitre, Noble, Darragh, Dye, Cleghorn and Hyland - leaving out the Joliat/Morenz generation against whom Nighbor competed towards the end of his career (including his last assists crown in 25-26). Now, not all of these players were active during every season of Nighbor's career, but most of them were active and at their peaks during Nighbor's prime scoring years. As we've discussed before through the course of ATD#11 (a discussion to which Spit was not a party, though he could have joined in), this is quite a big generation of talent - on an entirely different level from the pre-NHA generation. When discussing top-10 placements, this is in no way a thin universe of scoringline forwards, and I see no reason to devalue scoring achievements from this era once the two-league effect is accounted for (which I do quite brutally for Nighbor by multiplying his scoring finishes by two).

It's not like Bobby Clarke's era was so much stronger. The mid-70's were a lowpoint for the NHL between the O6 and 80's eras, and Bobby Clarke, himself, got outscored during his prime by names like Pete Mahovlich, Tim Young, Terry O'Reilly, Ken Hodge, Wayne Cashman, Vic Hadfield, etc. Does it make Bobby Clarke any less an offensive force because he sometimes fell short of guys who weren't all time greats? No, just like it doesn't make Nighbor any less a scorer because he occasionally got outscored by the Corb Dennenys of hockey. If these guys had been able to do it for as long as Nighbor and Clarke, they'd be all-time greats, as well.

Look at the top-end competition in Nighbor and Clarke's respective eras. Was Lafleur better than Lalonde? The biggest difference in their careers seems to be Lafleur's playoff performances, not his regular season numbers. Esposito vs. Taylor? I'd probably give it to Espo, but it's debatable. Perreault vs. Malone. Again, highly debatable, and I think Malone probably wins this one. Broadbent vs. Lemaire; Foyston vs. Martin; MacKay vs. Ratelle; Dye vs. Shutt; Denneny vs. Barber, etc. Compare the forward talent between the eras, and it doesn't look particularly different unless you're the kind of person who doesn't realize how great a scorer Mickey MacKay was. The high-end forward talent was probably a bit thicker in the 70's than it was in the teens and Bobby Orr was playing, but a non-biased comparison of eras does not show huge differences in the competitive level among the top scorers. There were a lot more players in the 70's, but the top layer of talent doesn't appear to have been particularly superior.

Only an extremely biased and unfair appraisal of Nighbor's scoring credentials can place him much below Clarke in this area, and there is certainly an argument that Nighbor was actually the better scorer (value of goal-scoring vs. playmaking, etc.). Clarke's goal-scoring credentials are quite weak. How would pure playmaker Bobby Clarke have looked in NHA scoring tables that only counted goals? Offensively, Nighbor and Clarke are on the same level, and defensively, they exist in their own special class, which includes only Bobby Clarke and Frank Nighbor. There simply are no other scoring forwards in hockey history whose defensive games approach this level of dominance.

...

None of these players, including Milt Schmidt, have a defensive value even close to that of Bobby Clarke. Frank Nighbor does. The biggest thing holding Nighbor back from more recognition in the ATD was not his defensive value, on which there is virtually unanimous agreement. The biggest thing keeping Nighbor off of ATD 1st lines and out of the HOH top-50 was a lack of perspective on his scoring feats, due mostly to limited information on NHA scoring and postseason contests during the era. That blind spot has been corrected, and now we know why so many of Nighbor's contemporaries considered him the best player in hockey. Now it makes sense. Now we understand why he got more votes than Lalonde in the "Player of the Half-Century" poll, why he won the 1st Hart trophy in a season in which he placed 8th in points.

I really think that the ideal system would do something similar except with percentile VsX-like system. Otherwise, we are back to the days when Bobby Clarke is a significantly better offensive player than Frank Nighbor, at least partly because Clarke played in an era when it was much easier for a playmaker to get credited with an assist.

 

[Dreakmur]

The problem with taking the assist numbers at the time at face value is that it makes it look like nobody was a good playmaker, not when comparing them to NHL-era players, especially post-war players.

At the very minimum, you have a situation with zero secondary assists, that alone disadvantages certain players.

You also have a situation where Frank Nighbor looks like a really awful pick in the top 50.

This is Sturminator's post where he used top 10 finishes in goals and assists separately to compare Nighbor to Bobby Clarke and concluded that their offense is similar:



I really think that the ideal system would do something similar except with percentile VsX-like system. Otherwise, we are back to the days when Bobby Clarke is a significantly better offensive player than Frank Nighbor, at least partly because Clarke played in an era when it was much easier for a playmaker to get credited with an assist.

I’m not sure it was just a difference in recording assists. Based on game descriptions I have read, it seems like there would have actually been fewer assists.

 

[Sturminator]

There remains the enormous problem of comparing leagues which were playing different games with different rules. This includes the number of men on the ice (6 vs. 7 man hockey), but also includes forward passing, and penalties resulting in power plays. These are all huge differences in their effect on scoring opportunities. Quality of competition was also quite uneven during the pre-consolidation era, and there is no objective means of squaring this circle.

I have always considered this a noble effort, Dreak, but ultimately these numbers will always need to be supplemented with a lot of granular information about league rules, who was on the leaderboards, etc.

 

[Dreakmur]

There remains the enormous problem of comparing leagues which were playing different games with different rules. This includes the number of men on the ice (6 vs. 7 man hockey), but also includes forward passing, and penalties resulting in power plays. These are all huge differences in their effect on scoring opportunities. Quality of competition was also quite uneven during the pre-consolidation era, and there is no objective means of squaring this circle.

Yes, this whole thing assumes that the leagues are equal.

Rule differences, for me, are irrelevant. There are plenty of rule differences between eras.

I have always considered this a noble effort, Dreak, but ultimately these numbers will always need to be supplemented with a lot of granular information about league rules, who was on the leaderboards, etc.

The leagues are all colour-coded, so you can see the respective leader boards.

 

[ImporterExporter]

I still maintain Nighbor is underrated offensively.

He was the creme de la creme defensively speaking and still managed to put up some strong scoring totals over the course of his career (single season as well), being a play making slanted C, which is likely hurt by very sporadic assists records. Doing the Pete Green bio, I got to investigate a ton of game reports from the 1900-1925 range and it's pretty clear there are a lot of assists that didn't get recorded, at least if you believe the words in the newspapers to be accurate descriptions of play. And I'm talking the primary kind. Trying to gauge secondary assists is nearly impossible in most cases but I'm sure they happened. We simply don't know how much unfortunately.

I think play makers are universally underappreciated in this era.

This was a great effort Dreak. I've always enjoyed your consolidated (NHA/NHL/PCHA, etc) number crunching as well. Gives you at least a ballpark idea where a player ranks worldwide, even if the rules were quite different between leagues.

Pre-consolidation really needs a top something project in the HoH forum. There is so much to unearth.

 

[Dreakmur]

I still maintain Nighbor is underrated offensively.

The numbers don't support that.

This was a great effort Dreak. I've always enjoyed your consolidated (NHA/NHL/PCHA, etc) number crunching as well. Gives you at least a ballpark idea where a player ranks worldwide, even if the rules were quite different between leagues.

This thread went exactly as I expected. A bunch of posters pointing out problems and nobody coming with any solutions. I'm used to it after 20+ years coaching minor hockey lol.

 

[ImporterExporter]

The numbers don't support that.



This thread went exactly as I expected. A bunch of posters pointing out problems and nobody coming with any solutions. I'm used to it after 20+ years coaching minor hockey lol.

Huh? I think you took my post the wrong way sir.

I was complimenting you on your efforts. The only issue I have with that era, in terms of calculating offensive value is lack of assist data. That's not your fault and I'm not going to state I have any clear cut solutions as I'm not as well versed in the numbers game as some of you are.

The only thing I'm fairly certain on is players who had a reputation for setting up others, are impacted negatively because assists weren't regularly tracked. We can quantify goal scoring. We can't do that for players who were setting others up, which Nighbor seemed to do more than most. Again, just an opinion, not fact by any means.

 

[ResilientBeast]

My criticisms I guess would be

1) 6 man vs 7 man hockey, the dynamics of it were just different so trying to equate the two on one scoring table just feels wrong to me (despite my previous efforts to do so)
2) Sure the assists were handed out in a different frequency...but the adjustment is kinda creating them out of thin air to try and equate the two.

So @Sturminator kinda of touched my two points

1) Rule changes to you are irrelevant year to year which I don't have an issue with since a VsX style system is only comparing players scoring from within a single season. But your system attempts to equate two leagues playing in someway fundamentally different versions of "hockey" into one leaderboard. This is different than comparing VsX scores from 1929 (was that Cooney's big year where they changed the rules) and 1930 for example. The rules changed sure, but you're only comparing players against their peers playing under the same rules. So combining into a single 1915 leaderboard just feels off personally.

2) I still struggle with creating assists devoid of any context surrounding the goals their own team scored. By trying to make assist rates "even" between the leagues you ignore the big rule differences that forced players to earn assists in specific ways. But by inflating NHA/NHL assists to match the amount given in the PCHA I think it probably over-rewards players who aren't traditionally good playmakers and wouldn't find nearly the same success under PCHA rules.

I still like the system and use it, and I've definitely swung violently back and forth over the years but think I've finally settled into a place where their numbers have value but require a lot of context.

 

[Dreakmur]

So @Sturminator kinda of touched my two points

1) Rule changes to you are irrelevant year to year which I don't have an issue with since a VsX style system is only comparing players scoring from within a single season. But your system attempts to equate two leagues playing in someway fundamentally different versions of "hockey" into one leaderboard. This is different than comparing VsX scores from 1929 (was that Cooney's big year where they changed the rules) and 1930 for example. The rules changed sure, but you're only comparing players against their peers playing under the same rules. So combining into a single 1915 leaderboard just feels off personally.

If one league had rules that impacted scoring, it would be accounted for in the scoring equalization formula.

2) I still struggle with creating assists devoid of any context surrounding the goals their own team scored. By trying to make assist rates "even" between the leagues you ignore the big rule differences that forced players to earn assists in specific ways. But by inflating NHA/NHL assists to match the amount given in the PCHA I think it probably over-rewards players who aren't traditionally good playmakers and wouldn't find nearly the same success under PCHA rules.

The NHA/NHL assists were inflated about as much as the PCHA/WCHL were.

 

[TheDevilMadeMe]

The numbers don't support that.



This thread went exactly as I expected. A bunch of posters pointing out problems and nobody coming with any solutions. I'm used to it after 20+ years coaching minor hockey lol.

I thought about this, and here's what I would consider doing:

Normalize assists in every season to 1.5 assists per goal. That's still less than the actual 1.8 assists per goal in modern times.

Alternatively, what is the ratio of assists-per-goal in the first decade after the 1926 consolidation? I would consider normalizing assists to that number, though there are those of us who think that the officially recorded assists in the 1930s STILL disadvantage playmakers, though far less so than they did in the split league era.

Hell, even normalizing assists to a 1:1 ratio with goals would be a major improvement.

____________

Going through all that - the first decision is clearly going to be to decide what multiplier to give officially recorded assists.

_______________

And yes, I realize that I am asking you to do a lot of work that I myself am not going to be doing any time soon.

 

[Dreakmur]

I thought about this, and here's what I would consider doing:

Normalize assists in every season to 1.5 assists per goal. That's still less than the actual 1.8 assists per goal in modern times.

Alternatively, what is the ratio of assists-per-goal in the first decade after the 1926 consolidation? I would consider normalizing assists to that number, though there are those of us who think that the officially recorded assists in the 1930s STILL disadvantage playmakers, though far less so than they did in the split league era.

Hell, even normalizing assists to a 1:1 ratio with goals would be a major improvement.

____________

Going through all that - the first decision is clearly going to be to decide what multiplier to give officially recorded assists.

_______________

And yes, I realize that I am asking you to do a lot of work that I myself am not going to be doing any time soon.

I set it up on excel because doing that formula would actually be very little work.

I’m still not sure exactly what we normalize to, since I think there were actually far fewer assists due to the forward passing forward rules.