HFBoards

Go Back   HFBoards > General Hockey Discussion > By The Numbers
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
By The Numbers Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

Top 3 Adjusted Years

View Poll Results: How many games constitutes a 'season' rather than a 'pace'
35 0 0%
40 0 0%
45 0 0%
50 1 14.29%
55 2 28.57%
60 3 42.86%
65 1 14.29%
70 0 0%
Voters: 7. You may not vote on this poll

Closed Thread
 
Thread Tools
Old
08-11-2012, 03:16 PM
  #1
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
Top 3 Adjusted Years

Hi guys. Looking at some of the other threads onhere, this might come across as caveman-like, but I find it interesting, regardless.

I went through hockey-ref's adjusted stats, starting with all the guys who have had at least 1 100 point adjusted season, found the each guy's best 3 years of adjusted points, and made a total.

I made a compromise between healthy players, and the oft-injured ones. Any single season HAD to have involved at least 60 games in order to count, then it was prorated to a full season. So, there are a lot of very big seasons here, on my personal list, that may not have shown up in Art Rossville. This, I feel, is a good compromise. Crosby, Forsberg, Lemieux, Lindros, and MOST of all, Pat Lafontaine, lose some good 'on pace' seasons, and the entire 94-95 season is eliminated - but, I feel 60 games constitutes a season, and anything less is more of an 'on pace' sample size.

I did these numbers with some beer, and got a bit excited, so I flew through it. Many numbers are rounded by season, so several of these totals may be off by a point, or maybe even 2.

Lastly, I removed everyone pre-1980 except Trottier, Bossy, and Dionne. There are a few reasons for this, but mostly I feel that era scoring is so skewed in the first years after expansion, especially in 70-75 when they started playing the 06 teams against the expansion teams an equal amount of games that it wrecks the numbers. League scoring was kinda low, because so many teams were very weak - yet 06 team scoring was similar to 80's scoring, so I feel 06 guys had an enormous adjustment advantage. It was easy to make the 1980 split because only those 3 guys carried their careers strongly into the 80's - the other 70's producers did not. Those 3 also have most of their big years in the 80's.

No further ado:

1. Mario Lemieux - 532
2. Wayne Gretzky - 512
3. Jaromir Jagr - 427
4. Alex Ovechkin - 373
5. Joe Sakic - 370
6. Evgeni Malkin - 365
7. Peter Forsberg - 364
8. Teemu Selanne - 363
9. Sidney Crosby - 358
9. Joe Thornton - 358
11. Adam Oates - 355
12. Eric Lindros - 351
13. Steve Yzerman - 346
14. Paul Kariya - 342
15. Bryan Trottier - 337
16. Marcel Dionne - 335
16. Jari Kurri - 335
18. Brett Hull - 330
19. Jarome Iginla - 327
20. Theoren Fleury - 326
21. Mike Bossy - 325
22. Pavel Bure - 324
22. Markus Naslund - 324
24. Jason Spezza - 321
25. Martin St. Louis - 320
26. Daniel Alfredsson - 319
26. Ron Francis - 319
28. Daniel Sedin - 318
29. Sergei Fedorov - 317
29. Danny Heatley - 317
31. Alex Mogilny - 315
32. Pierre Turgeon - 314
33. Zigmund Palffy - 311
33. John Leclair - 311
35. Peter Stastny - 310
35. Steven Stamkos - 310
35. Henrik Sedin - 310
38. Mark Messier - 309
38. Paul Coffey - 309
38. Pat Lafontaine - 309
41. Pavol Demitra - 304
42. Denis Savard - 303
42. Alexei Kovalev - 303
44. Burning Nipples - 301
44. Ilya Kovalchuk - 301
46. Vincent Lecavalier - 300
46. Todd Bertuzzi - 300
46. Pavel Datsyuk - 300
46. Mark Recchi - 300

tombombadil is offline  
Old
08-11-2012, 04:25 PM
  #2
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Nice list. I'm not sure why you would eliminate the '95 season though. The biggest surprise to me is that Lindros isn't higher under such a system, and that likely relates to the '95 season being eliminated. Your instincts about the 70s are correct, it was much easier to get adjusted points during those years than in any other era in the past 50+ years.

I think this highlights a difficult problem: How to credit players who played most of a season but missed ~1/4 season or less.

Matnor has done some work on this with his "on-pace" lists. My concern with those is that he actually gives a player more credit for playing any games at a top 5 or top 10 pace, than he does for players who are displaced by these players playing shortened seasons. IOW, if Joe actually finished 5th in points, but Mike played 20 games with a higher PPG, Joe will not receive ANY credit for top 5 pace, but Mike will receive 20 games credit.

Someone else came up with "points over PPG" or something, where they took (Points-Games).

My only other thought on such a metric is using harmonic mean of some kind:

A = actual points
P = pace points (adjusted to full season)

So the metric would be some form of 2 * (A * P) / (A + P)

Let's use Crosby's unadjusted 2011 season as an example:

2 * (66 * 132) / ( 66 + 132) = 88

So it gives him credit for 88 points, which more balances how good he was when healthy, against his actual value for the season.


Last edited by Czech Your Math: 08-11-2012 at 04:46 PM.
Czech Your Math is offline  
Old
08-11-2012, 07:00 PM
  #3
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
thanks, Czech. This was more or less posted for you. I thought you might expand on it.

94-95 was left of for the simple reason of being under 60 games. I think it falls into a 'pace' or sample size rating. Many guys had a big adjusted season, and had poor seasons on either side of it. It kind of feels like how the scoring race looks after 40 games... not everyone can keep the pace up. Anyways, doesn't matter if that's right r wrong, it failed to meet one of the requirements.

Have you ever adjusted Orr or Esposito's numbers in say, 70-71, or 71-72, by taking the scoring average of the O6 teams instead of the league average - the way reference does it?

tombombadil is offline  
Old
08-11-2012, 07:43 PM
  #4
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Quote:
Originally Posted by tombombadil View Post
thanks, Czech. This was more or less posted for you. I thought you might expand on it.

94-95 was left of for the simple reason of being under 60 games. I think it falls into a 'pace' or sample size rating. Many guys had a big adjusted season, and had poor seasons on either side of it. It kind of feels like how the scoring race looks after 40 games... not everyone can keep the pace up. Anyways, doesn't matter if that's right r wrong, it failed to meet one of the requirements.

Have you ever adjusted Orr or Esposito's numbers in say, 70-71, or 71-72, by taking the scoring average of the O6 teams instead of the league average - the way reference does it?
I've looked at O6 team scoring after expansion, but could take another look. After doing the study of fixed group of players from year to year, I can use those results. I still don't think it captures all the expansion effect, but it's the closest I have found. I'll try to post some results here at a later time.

I just think playing most or all of 48 games in at least as reliable as playing 60/82 or something. One player who really needs '95 is Lindros and the issue with him was he couldn't play a full schedule. The other one is Zhamnov, and that may have been a case where he likely doesn't keep up the pace. Jagr and Sakic did just fine the next year. Players like Forsberg, Kariya, Selanne and Gretzky didn't do so well due to being rookies or injuries. Who else put up big numbers that wouldn't be sustainable over 82 games?

Czech Your Math is offline  
Old
08-11-2012, 08:03 PM
  #5
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
Quote:
Originally Posted by Czech Your Math View Post
I've looked at O6 team scoring after expansion, but could take another look. After doing the study of fixed group of players from year to year, I can use those results. I still don't think it captures all the expansion effect, but it's the closest I have found. I'll try to post some results here at a later time.

I just think playing most or all of 48 games in at least as reliable as playing 60/82 or something. One player who really needs '95 is Lindros and the issue with him was he couldn't play a full schedule. The other one is Zhamnov, and that may have been a case where he likely doesn't keep up the pace. Jagr and Sakic did just fine the next year. Players like Forsberg, Kariya, Selanne and Gretzky didn't do so well due to being rookies or injuries. Who else put up big numbers that wouldn't be sustainable over 82 games?
great. we are on the same page. Zhamnov was actually the big one, in my mind.

Lindros, for sure. And I guess I agree, he needs that, and it is right in the middle of his prime.

Ron Francis and Paul Coffey had big years.... you know what, I will go through it some time tonight, and get back to you. It seemed like there were tons...

I had a lot of guesses as to why... generally goals come easier when the pace is less. I wonder if they weren't at each other's throats as much after striking together... or if perhaps, they were all a little out of shape. I may not actually have a point, after i look at it - but I originally put it down simply due to lack of size, as per the guidelines i had set out.

Why not Pat Lafontaine's huge season the year before his BIG season, if we include 48 game stretches? He played less than 60 and had somewhere close to 100 adj points.

I'll take a look.

tombombadil is offline  
Old
08-11-2012, 08:16 PM
  #6
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
ya, **** it - all i got is Coffey and Zhamnov, as far as massive outliers. Francis just got better. Lindros' isn't actually an outlier, as soon as i pro rate his next year to 82 games. I dunno, still a small sample size! I think I'm going to look at 70-71 now.

tombombadil is offline  
Old
08-11-2012, 08:22 PM
  #7
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
It's your list, so you do it however you want. It's fine the way it is, and you have a good rationale. You don't need to change it. I'm just saying that playing almost every game out of 48 is different than playing 45 out of 82. It's all the NHL's fault for these stupid lockouts. That's also why it's good you used 3 seasons. That's about the minimum for establishing a level of performance. The more seasons you use, the less one season will affect things. It's not fair to exclude that season... but then it's not 100% fair to include it either, so there's no perfect solution.

I'll post a list of best 3, 5, etc. simple adjusted seasons and adjusted using my study, just for comparison purposes. HR.com uses a slightly different method, which doesn't matter much in more recent years, but as you go back to pre-80s and esp. O6, it does change the results a lot. Although I have a good idea which players to look at, your list will help a lot and save me some time.


Last edited by Czech Your Math: 08-11-2012 at 08:28 PM.
Czech Your Math is offline  
Old
08-11-2012, 08:37 PM
  #8
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
great, i agree with all of this. I would like to see your 5 year study. I also agree that 3 is about the minimum for eliminating outliers. The more the better, like you said, but the more you go, the more it hurts guys who were injured. 5 is close to the maximum i would be interested in.

Side note - do you know the roster size for 1970-71? I just adjusted Orr's goals, by creating the league average out of the O6 teams, multiplied the 82/78 games adjustment. I used the same roster adjustment as it has for Howe, and after looking at the Bruins lineup, it seems right, but i have no proof. If it IS a .86 adjustment, it would put Orr from 34 goals to just under 29, which would have Espo losing around 10 goals that year.

tombombadil is offline  
Old
08-11-2012, 09:01 PM
  #9
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Quote:
Originally Posted by tombombadil View Post
great, i agree with all of this. I would like to see your 5 year study. I also agree that 3 is about the minimum for eliminating outliers. The more the better, like you said, but the more you go, the more it hurts guys who were injured. 5 is close to the maximum i would be interested in.

Side note - do you know the roster size for 1970-71? I just adjusted Orr's goals, by creating the league average out of the O6 teams, multiplied the 82/78 games adjustment. I used the same roster adjustment as it has for Howe, and after looking at the Bruins lineup, it seems right, but i have no proof. If it IS a .86 adjustment, it would put Orr from 34 goals to just under 29, which would have Espo losing around 10 goals that year.
I have both simple adjusted (schedule, gpg, assist/goal) and adjusted further using my study (quality of era for top line players). The results from my study would be especially useful, since 1) you are rightly concerned about the 70s being easier for adjusted points, and 2) my study used PPG not points as the basis. The clutch and grab era stars are not helped by the fact that it was so difficult for most of them to stay healthy. The study dings their PPG a little bit, but doesn't factor in that they lost a lot of games due to injury. This is a double whammy of sorts to their doubly adjusted point seasons, which won't be as much of an issue in this context.

Czech Your Math is offline  
Old
08-11-2012, 09:33 PM
  #10
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 2,795
vCash: 500
Quote:
Originally Posted by Czech Your Math View Post
I just think playing most or all of 48 games in at least as reliable as playing 60/82 or something.
I don't think that makes any sense. The reason playing 50 games in a season isn't very reliable for projections is that it's a lower sample size than those who played 80, and so projecting from 50 is less valid; you have less evidence to go on. The same applies to 48 games, regardless of whether 48 was any many games as they were supposed to have played or not.

If you're going to exclude players below a certain number of games played, you have to stick to that, regardless of whether anyone else played more games than that in that season.

Iain Fyffe is offline  
Old
08-11-2012, 09:41 PM
  #11
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
that's how i saw it, Ian. Simple sample size, not 'ability to be healthy'.

I like this list, the more i look at it. I wouldn't mind a 5 year version, or a ummmmm "chronological" period. Like, not cherry-picked seasons, but a 4 year span, or something like that.

There were surprises for me when i did this, but most of it came out the way I would have thought... breaks my heart that Sakic is ahead of Forsberg as I have always thought of Peter as slightly better, peakwise. Interesting for me to see how many guys from really great lines show up... ie. Sedin's, Bert and Naz, The Pizza Line.... Makes it more impressive when someone like Bure or Palffy show up.

tombombadil is offline  
Old
08-11-2012, 09:50 PM
  #12
pdd
Registered User
 
Join Date: Feb 2010
Posts: 5,576
vCash: 500
Quote:
Originally Posted by Czech Your Math View Post
I've looked at O6 team scoring after expansion, but could take another look. After doing the study of fixed group of players from year to year, I can use those results. I still don't think it captures all the expansion effect, but it's the closest I have found. I'll try to post some results here at a later time.

I just think playing most or all of 48 games in at least as reliable as playing 60/82 or something. One player who really needs '95 is Lindros and the issue with him was he couldn't play a full schedule. The other one is Zhamnov, and that may have been a case where he likely doesn't keep up the pace. Jagr and Sakic did just fine the next year. Players like Forsberg, Kariya, Selanne and Gretzky didn't do so well due to being rookies or injuries. Who else put up big numbers that wouldn't be sustainable over 82 games?
Lindros and Coffey aren't outliers in 1995. Look at their performances the few seasons before.

pdd is offline  
Old
08-11-2012, 09:52 PM
  #13
pdd
Registered User
 
Join Date: Feb 2010
Posts: 5,576
vCash: 500
Quote:
Originally Posted by tombombadil View Post
that's how i saw it, Ian. Simple sample size, not 'ability to be healthy'.

I like this list, the more i look at it. I wouldn't mind a 5 year version, or a ummmmm "chronological" period. Like, not cherry-picked seasons, but a 4 year span, or something like that.

There were surprises for me when i did this, but most of it came out the way I would have thought... breaks my heart that Sakic is ahead of Forsberg as I have always thought of Peter as slightly better, peakwise. Interesting for me to see how many guys from really great lines show up... ie. Sedin's, Bert and Naz, The Pizza Line.... Makes it more impressive when someone like Bure or Palffy show up.
Like Steve Yzerman?

pdd is offline  
Old
08-11-2012, 11:37 PM
  #14
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 8,238
vCash: 500
Quote:
Originally Posted by tombombadil View Post
Hi guys. Looking at some of the other threads onhere, this might come across as caveman-like, but I find it interesting, regardless.

I went through hockey-ref's adjusted stats, starting with all the guys who have had at least 1 100 point adjusted season, found the each guy's best 3 years of adjusted points, and made a total.

I made a compromise between healthy players, and the oft-injured ones. Any single season HAD to have involved at least 60 games in order to count, then it was prorated to a full season. So, there are a lot of very big seasons here, on my personal list, that may not have shown up in Art Rossville. This, I feel, is a good compromise. Crosby, Forsberg, Lemieux, Lindros, and MOST of all, Pat Lafontaine, lose some good 'on pace' seasons, and the entire 94-95 season is eliminated - but, I feel 60 games constitutes a season, and anything less is more of an 'on pace' sample size.

I did these numbers with some beer, and got a bit excited, so I flew through it. Many numbers are rounded by season, so several of these totals may be off by a point, or maybe even 2.

Lastly, I removed everyone pre-1980 except Trottier, Bossy, and Dionne. There are a few reasons for this, but mostly I feel that era scoring is so skewed in the first years after expansion, especially in 70-75 when they started playing the 06 teams against the expansion teams an equal amount of games that it wrecks the numbers. League scoring was kinda low, because so many teams were very weak - yet 06 team scoring was similar to 80's scoring, so I feel 06 guys had an enormous adjustment advantage. It was easy to make the 1980 split because only those 3 guys carried their careers strongly into the 80's - the other 70's producers did not. Those 3 also have most of their big years in the 80's.

No further ado:

1. Mario Lemieux - 532
2. Wayne Gretzky - 512

I'm sorry but I don't see the point?

I love Mario but to give an idea how massaging the data precisely to make Lemieux look better affects the outcome:


Best two actual seasons:

1. Gretzky 427
2. Lemieux 367


Best three actual seasons:

1. Gretzky 635
2. Lemieux 528


Best four actual seasons:

1. Gretzky 840
2. Lemieux 688


Best two adjusted seasons (without prorating):

1. Gretzky 336
2. Lemieux 321


Best three adjusted seasons (without prorating):

1. Gretzky 499
2. Lemieux 462


Best four adjusted seasons (without prorating):

1. Gretzky 658
2. Lemieux 591


Best three adjusted seasons - prorated with 61 games played minimum because *I* feel that anything with less than 61 games is probably too small a sample size -

1. Gretzky 512
2. Lemieux 503


Best three adjusted seasons (prorated with 70 minimum games played:

1. Gretzky 512 (notice this doesn't change)
2. Lemieux 503


Best three adjusted seasons (prorated with 75 minimum games played:

1. Gretzky 495
2. Lemieux 458


I mean everyone who was watching hockey in 92-93 was in awe of what Mario did but come on.. it sadly wasn't a 181 point adjusted season in reality.

BraveCanadian is offline  
Old
08-11-2012, 11:42 PM
  #15
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
i think someone is reading way too deeply into making this into a Gretzky/Lemieux thing. I couldn't care less what you, or anyone else thinks about them, and am not partial to either. They were both very good, this is how their numbers turned out in this particular case.

This is not a list of the 49 best forwards, in order, either.


As for the Yzerman comment... ya, i would agree. His 3rd season that made the grade, he had some of the big names around him, but they were young, and he still had a good gap between him and them. Good call. Looking down the list, I would have to say Iginla stands out a lot, too.

tombombadil is offline  
Old
08-12-2012, 12:11 AM
  #16
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Quote:
Originally Posted by Iain Fyffe View Post
I don't think that makes any sense. The reason playing 50 games in a season isn't very reliable for projections is that it's a lower sample size than those who played 80, and so projecting from 50 is less valid; you have less evidence to go on. The same applies to 48 games, regardless of whether 48 was any many games as they were supposed to have played or not.

If you're going to exclude players below a certain number of games played, you have to stick to that, regardless of whether anyone else played more games than that in that season.
I understand that, but I don't see it as clearly defined as you do.

The reason 60/80 games could be less reliable than at least some of the '95 seasons is that some of those players played nearly all of the available games. If a player can pick and choose when he plays (and I'm not saying that's exactly or generally what is happening, but it's always a possibility), then his PPG may be significantly higher and much less reliable. It depends on if the player just missed some games, while having a season at a clearly established level (at least a couple other seasons at that level) or was having a career year in PPG terms. I'd give the benefit of the doubt to the former, but not the latter.

That's why I suggested an alternative, such as harmonic mean approach. Another alternative is to add all of the games and adjusted (but not for schedule) points separately for seasons in which the player's games hit a certain threshold (for instance, 300 total actual games) and then calculate the PPG. The problem with that method is that some players may just miss the threshold and require another season to qualify, which penalizes them.

Increasing it to 5 seasons and including '95 might also solve the problem to a large degree, since the influence of any one season will be much less.

Czech Your Math is offline  
Old
08-12-2012, 12:15 AM
  #17
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
I wouldn't mind seeing the top 20 over 5 years, and 4 years, too..... just to see how it shakes down. I don't really wanna do it, though!

tombombadil is offline  
Old
08-12-2012, 12:16 AM
  #18
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 8,238
vCash: 500
Quote:
Originally Posted by tombombadil View Post
i think someone is reading way too deeply into making this into a Gretzky/Lemieux thing. I couldn't care less what you, or anyone else thinks about them, and am not partial to either. They were both very good, this is how their numbers turned out in this particular case.
It isn't a Gretzky/Lemieux thing it is a reality fantasy thing.

My point is that if the three most productive seasons of any career in hockey history are being decided one way or another by a 1 game difference in an arbitrary cutoff there may be problems afoot.

A full 15+% of Lemieux's "production" in your list is from fantasy land.

But I'll leave you guys to it.

BraveCanadian is offline  
Old
08-12-2012, 12:21 AM
  #19
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Quote:
Originally Posted by eva unit zero View Post
Lindros and Coffey aren't outliers in 1995. Look at their performances the few seasons before.
I didn't even mention Coffey, the OP did. At that point in his career, it was an exceptional performance by Coffey, at least 20% better than his best adjusted PPG in the previous 5 years, and even more so than any since he played with Mario.

In Lindros' case that was his best career adjusted PPG by 10% and it was way better than any season of his to that point. It's at least as much of an exceptional performance for him in % of games played as it is in PPG terms, and that's probably the bigger question in his case, could he have really played more than ~70-75 games at most?

Czech Your Math is offline  
Old
08-12-2012, 12:21 AM
  #20
Iain Fyffe
Hockey fact-checker
 
Iain Fyffe's Avatar
 
Join Date: Feb 2009
Location: Fredericton, NB
Country: Canada
Posts: 2,795
vCash: 500
Quote:
Originally Posted by Czech Your Math View Post
The reason 60/80 games could be less reliable than at least some of the '95 seasons is that some of those players played nearly all of the available games. If a player can pick and choose when he plays (and I'm not saying that's exactly or generally what is happening, but it's always a possibility), then his PPG may be significantly higher and much less reliable.
I'd suggest that is such a remote possibility so as to be easily ignored. It's rare the player that decides when he plays (normally that's the coach's job), and then you have to assume they're choosing to play weaker opponents in order for it to matter.

Quote:
Originally Posted by Czech Your Math View Post
The problem with that method is that some players may just miss the threshold and require another season to qualify, which penalizes them.
That's a problem with any arbitrary number limit on games played to qualify for some statistical thing. A player with 100 points in 59 games is excluded, while someone with 90 in 60 isn't (assuming 60 is your limit).

Iain Fyffe is offline  
Old
08-12-2012, 12:44 AM
  #21
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,665
vCash: 500
Quote:
Originally Posted by tombombadil View Post
I wouldn't mind seeing the top 20 over 5 years, and 4 years, too..... just to see how it shakes down. I don't really wanna do it, though!
I will do it within the next couple days, I have all the data, just need to have the focus to do the rest, which shouldn't take very long. Who knows, it might be sooner rather than later, but no promises.

Quote:
Originally Posted by BraveCanadian View Post
It isn't a Gretzky/Lemieux thing it is a reality fantasy thing.

My point is that if the three most productive seasons of any career in hockey history are being decided one way or another by a 1 game difference in an arbitrary cutoff there may be problems afoot.

A full 15+% of Lemieux's "production" in your list is from fantasy land.

But I'll leave you guys to it.
I don't think OP was trying to promote any specific player, it's not like he put together a top 5 and said "see, I knew Mario was better than Wayne!." This sort of thing happens a lot with the greats, because they always end up at or near the top of such lists. Any list that has them "too low" must be biased against them and any that has them "too high" must be biased for them.

I think OP was trying to find a reasonable compromise and did his best to find one, and I think it's more of a fun exercise than anything. I've seen posters who probably take this sort of thing a lot more seriously put forth efforts that could be questioned just as easily. I don't see any simple solution to recognizing how good a player was when healthy, while also being fair to those who played through injuries. The methodology seems to be somewhat biased for players which were more prone to injury and missing games because of this, but I have no reason to believe it was intentional.

I think you're bristling at the fact that it has Lemieux over Gretzky, and I understand why you would, because it is indeed biased for a player like Mario and against a player like Wayne. I know you are a very knowledgeable and respected poster, so why not help find a solution or politely say your peace and ignore it, rather than imply the OP has an agenda or dismiss it as "fantasy land."

I've proposed a couple solutions:

1) use a harmonic mean which rewards both high PPG and high point total in season, not one or the other

2) use a threshold of X total games over any number of seasons and use actual games and adjusted (for gpg and assist/goal only) points to calculate a PPG

3) increase the number of seasons to 5 or something to yield more reliable results

I think #1 is perhaps the fairest "compromise".

#2 still has the problem of still favoring PPG over points to a large degree, although if the threshold X is high enough, this is mitigated to a large degree, at least as an indication of "talent."

#3 has the same problems as #2, but with similar improvement.

Can someone suggest some thresholds X? If you guys can come to some sort of an agreement, then I don't have to pick the number(s) and receive the brunt of criticism as a "biased" analyst.

Czech Your Math is offline  
Old
08-12-2012, 12:44 AM
  #22
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
That's a problem with any arbitrary number limit on games played to qualify for some statistical thing. A player with 100 points in 59 games is excluded, while someone with 90 in 60 isn't (assuming 60 is your limit).[/QUOTE]

This is true. Arbitrary parameters are... uhh... arbitrary.

I can say, from going through these guys, one at a time, looking for which season was the best - that there were VERY FEW cases where someone really misses out due to the 60 games thing.

Coffey does, and so does Lindros... both from 95.

Lafontaine does, horribly, in 91-92 (i think this is it, without checking) and he has to take a much worse third season of somewhere in the 90 point zone instead.

I can say, from doing the numbers, that it felt like everyone else that this affected had a really close-to-as-strong 4th season that was used, and, without getting specific, probably only cost a couple of points.

On that note - when I made/look at this list - I really tend to ignore the 'rank' number in front of the name. The top 2 stand out. clearly, and Jagr clearly stands in third.... but, I'm not gonna rate Forsberg as better than Selanne based on a point, for instance. I just wanted to see this list, see how many guys hit the threshold... and once i was finished, thought i would share.

Ya, seriously, Lafontaine and Lindros would move up if the paramters were dropped lower... Coffey... and i'm not double checking here, but he probably has something in the 80's that is less than 10 points removed from his 94-95 season.

I'm glad this is getting some attention, and I wouldn't mind seeing a 5 year... the parameters could be dropped to 48 games, even. I just picked 60, roughly 3/4's... no solid argument from me as to why.

tombombadil is offline  
Old
08-12-2012, 12:58 AM
  #23
tombombadil
Registered User
 
Join Date: Jan 2010
Location: West Kelowna, Canada
Country: Canada
Posts: 1,030
vCash: 500
hey, thanks, Czech. good post. you're bang on - it is simply an exercise. I find it less important to pick apart who's ahead of who... I see clusters... and everyone on the list is off the charts great to begin with.

I don't think it's definitive... you know, no list is.. you could break something right down to which arenas give out assists easier (I swear I have seen Daniel get one in Van while he was in the stands ordering a hot dog!)

What, I guess I WAS aiming for, is that I don't think anyone on the list can be accused of getting there due to being a 'compiler', or, on the other side of things, be accused of 'getting given credit for playing less games' Guess I failed at the second one... but, and this seems to be the heart of the matter - we have a guy who wants to include a 48 game year, and a guy who is upset because a 60 is included.

Maybe a poll of what constitutes a 'season' that gets average out would be a good idea?

Looking back, I can see why BC jumped to that conclusion. The fact that Mario has a huge year at 60 games played, coupled with the sad fact that there is a common twisting of stats to create a picture that the statistician wants to see on these boards, probably made it look like that's what i was going for.

The funny truth, in my own bias, is that I don't give a crap about the top 3 - I know where they lie, roughly. Gretz was healthier, Lemieux carried his game into the trap better... I don't even care, they are far ahead of the rest, statistically. The guys I like are the battlers, and the guys that got some Selke recognition. Aside from an exercise, I wanted to see where Trottier, Sakic, Forsberg, Fedorov, Francis, Datsyuk, Messier, Lindros, etc. shook down. And, like I said earlier, it ticked me off when Joe finished ahead of Foppa, who is my favorite player... but i didn't abort the test and twist numbers. I went into this with parameters that I felt were fair, in respect to what I was trying to see. Everyone's best 3 years, adjusted, in years that were, in my mind, full enough to count.

tombombadil is offline  
Old
08-12-2012, 10:03 AM
  #24
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 8,238
vCash: 500
Quote:
Originally Posted by Czech Your Math View Post
I think OP was trying to find a reasonable compromise and did his best to find one, and I think it's more of a fun exercise than anything. I've seen posters who probably take this sort of thing a lot more seriously put forth efforts that could be questioned just as easily. I don't see any simple solution to recognizing how good a player was when healthy, while also being fair to those who played through injuries. The methodology seems to be somewhat biased for players which were more prone to injury and missing games because of this, but I have no reason to believe it was intentional.

I think you're bristling at the fact that it has Lemieux over Gretzky, and I understand why you would, because it is indeed biased for a player like Mario and against a player like Wayne. I know you are a very knowledgeable and respected poster, so why not help find a solution or politely say your peace and ignore it, rather than imply the OP has an agenda or dismiss it as "fantasy land."
As I said I don't see it as a Lemieux vs. Gretzky thing.. I like both of them.

I just think it is a problem any time you start pro-rating with extremely polarized cases like Lemieux (injury prone) and Gretzky (generally healthy) or even a Forsberg or say more recently Crosby's half seasons.

The amount of credit people give Crosby for his last couple spurts of seasons is crazy.

I agree with you though that it is very difficult to give players who are injury prone *some* credit for their talent without crossing the line into giving them too much credit for games they didn't play.

Really I wasn't trying to be so critical as my post comes off as, I was just illustrating that by practically any measure other than giving Lemieux full credit for 24 extra games during his best season reverses the results..

Obviously how much credit is fair will also be a subjective thing so it is pretty tough.

I think you put up some good options to try and minimize the issues while giving some credit for players who have the talent but had injury trouble.

One thing that might be worthwhile exploring is how often players maintain a pace from game 60 to 80 for example. Or in Crosby's case.. how often do players maintain a pace from 40 games to 80? This would give an idea of how fair pro-rating is in the first place and that might impact the weight of it in any method you want to come up with.. so if your cutoff for games played is going to be 60 games, knowing how likely a player is to maintain that pace to 80 should help formulate a fair estimate or possibly both a high and low estimate etc.

Because like you said - picking those thresholds and cutoff numbers is the tough part! Personally I have a hard time calling anything less than 75% of games played a full season but people will definitely vary on that (and everything else!).


Last edited by BraveCanadian: 08-12-2012 at 10:15 AM.
BraveCanadian is offline  
Old
08-12-2012, 11:16 AM
  #25
solidmotion
Registered User
 
Join Date: Jun 2012
Posts: 137
vCash: 500
Quote:
Originally Posted by tombombadil View Post
44. Burning Nipples - 301
Just slipped that in there eh... assuming Bernie Nicholls?

solidmotion is offline  
Closed Thread

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 01:43 PM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. 2014 All Rights Reserved.