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.

Vs#2 Thread

Reply
 
Thread Tools
Old
02-25-2012, 09:09 PM
  #26
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 24,986
vCash: 500
Quote:
Originally Posted by TheDevilMadeMe View Post
Looking at this chart - I'm going to say it now. VS5 or VS10 is a superior method to VS2. The data at 5 and 10 follows a much more gentle curve and doesn't appear as prone to random fluxuations as the second place scorer.

Actually, looking at it again, there are some years when even the 5th place scorer fluxuates. VS10 is a nice smooth curve through the entire time period - where the only peaks are years like 1993, when there should be one.
Problem with this is, pre-expansion, you are going to see some unholy scores for the top3 scorers a few times. Particularly from 49-55, the scoring lists were awful beyond the top few (this is the inverse of the problem you perceived last time when mackell had crap scores in two years he was a top-10 scorer)

seventieslord is online now   Reply With Quote
Old
02-25-2012, 09:12 PM
  #27
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
Quote:
Originally Posted by seventieslord View Post
Problem with this is, pre-expansion, you are going to see some unholy scores for the top3 scorers a few times. Particularly from 49-55, the scoring lists were awful beyond the top few (this is the inverse of the problem you perceived last time when mackell had crap scores in two years he was a top-10 scorer)
I wasn't thinking of pre-expansion, because I think vs2 is of limited use for the Original 6 era. vs5 and vs10 would be worse though.

The wild fluxuation of the #2 scorer make its use questionable, however.

TheDevilMadeMe is online now   Reply With Quote
Old
02-26-2012, 09:24 AM
  #28
BM67
Registered User
 
BM67's Avatar
 
Join Date: Mar 2002
Location: In "The System"
Country: Canada
Posts: 4,579
vCash: 500
Are wild fluctuations in the 70s really a problem with the system?

The league had expansion every couple of years. They had players moving back and forth with the WHA. The schedule wasn't balanced, and they realigned from 2 divisions to 4. Throw in the folding of the Seals/Barons and the first wave of European players for added spice.

Fluctuation is to be expected.

The #10-15-20-25 scorer all jumped by 11 points from 1971-72 to 1972-73. At the same time the top 6 scoring spots all had a decline in points.

The 72-73 NHL is a very different place than the 71-72 NHL. Adding in the Flames and Islanders as well as the WHA completely changes the landscape, and we haven't even looked at possible rule changes.

Point totals for selected spots in the top 25 68-80

Year TeamsGP #1 #2 #3 #4 #5 #10 #15 #20 #25
1968 1274 87 84 82 78 77 69 64 57 54
1969 1276 126 107 103 97 90 78 70 65 62
1970 1276 120 99 86 78 77 70 67 63 59
1971 1478 152 139 116 105 96 76 72 66 63
1972 1478 133 117 109 106 97 81 74 70 66
1973 1678 130 104 101 100 95 92 85 81 77
1974 1678 145 122 105 89 87 82 77 75 72
1975 1880 135 127 121 119 117 95 86 79 77
1976 1880 125 119 113 112 111 99 91 83 82
1977 1880 136 122 105 97 95 90 81 78 74
1978 1880 132 123 117 97 94 87 81 77 73
1979 1780 134 130 129 126 108 91 85 80 77
1980 2180 137 137 125 106 105 94 92 89 80

BM67 is offline   Reply With Quote
Old
02-26-2012, 12:38 PM
  #29
Hockey Outsider
Registered User
 
Hockey Outsider's Avatar
 
Join Date: Jan 2005
Country: Canada
Posts: 3,388
vCash: 500
My take on the "Vs. #2" is here: http://hfboards.hockeysfuture.com/sh...911&highlight=

I used the average of 2nd through 4th place. My rationale is that if you base a ranking on a single position (whether it's 2nd or 5th or 10th), there will always be outliers. By taking an average of a few positions, the numbers will be less likely to be skewed by a particularly good/bad performance. It also mitigates the need for you to arbitrarily identify and exclude "outlier" seasons.

If you use a lower position (say average of 5th to 10th), the numbers become seriously skewed pre-expansion (in many years, there may not have been a lot of elite talent beyond the top few spots).

All of the results (current as of 2008 or whenever I last updated this) are in the link. I think they look intuitively reasonable.

Hockey Outsider is offline   Reply With Quote
Old
02-26-2012, 02:53 PM
  #30
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
Quote:
Originally Posted by Hockey Outsider View Post
My take on the "Vs. #2" is here: http://hfboards.hockeysfuture.com/sh...911&highlight=

I used the average of 2nd through 4th place. My rationale is that if you base a ranking on a single position (whether it's 2nd or 5th or 10th), there will always be outliers. By taking an average of a few positions, the numbers will be less likely to be skewed by a particularly good/bad performance. It also mitigates the need for you to arbitrarily identify and exclude "outlier" seasons.

If you use a lower position (say average of 5th to 10th), the numbers become seriously skewed pre-expansion (in many years, there may not have been a lot of elite talent beyond the top few spots).

All of the results (current as of 2008 or whenever I last updated this) are in the link. I think they look intuitively reasonable.
Good work. This system still doesn't work for me in 1989 though - when the top 4 were all outliers. It still also compares everyone to a bunch of Bruins in a couple of seasons in the 1970s too.

TheDevilMadeMe is online now   Reply With Quote
Old
02-27-2012, 11:36 AM
  #31
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 24,986
vCash: 500
I don't see a problem with removing those players and them comparing to a 2-4 or 2-5 group.

seventieslord is online now   Reply With Quote
Old
02-27-2012, 01:02 PM
  #32
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
Quote:
Originally Posted by seventieslord View Post
I don't see a problem with removing those players and them comparing to a 2-4 or 2-5 group.
Arbitrarily removing players because they scored too much in a certain year makes the method less scientific. I think there's merit to creating a system with basic rules as to when to remove a player, like I tried above.

Maybe you're right though; maybe we should just give up trying to find a formula.

I do think I'm going to start using VS5 when comparing post-expansion players to each other.

TheDevilMadeMe is online now   Reply With Quote
Old
03-16-2012, 08:04 AM
  #33
BM67
Registered User
 
BM67's Avatar
 
Join Date: Mar 2002
Location: In "The System"
Country: Canada
Posts: 4,579
vCash: 500
Using the number of teams to set the mark. We start at either #1 or #2 through to T, the number of teams, and use the average as the "vs#2" number.

The table shows how many points that would be, and where that would sit in the scoring race.

Year Teams #1-#T Rank #2-#T Rank
1918 3 44.67 3 43 3
1919 3 29.67 2 28 2
1920 4 43 3 41 3
1921 4 39.75 3 38.67 4
1922 4 39.5 2 37.33 4
1923 4 32.5 3 31 3
1924 4 22.75 3 22.33 3
1925 6 38.83 5 37.4 5
1926 7 31.14 3 29.33 5
1927 10 29.7 6 28.89 6
1928 10 32.7 6 30.67 6
1929 10 25.5 7 24.78 7
1930 10 56.6 6 54.78 7
1931 10 41.8 6 40.78 7
1932 8 46.88 6 46 6
1933 9 40.89 6 39.75 6
1934 9 41.22 4 39.88 4
1935 9 45 4 43.5 6
1936 8 39.38 4 38.57 4
1937 8 40.62 5 39.86 6
1938 8 42.75 4 41.43 5
1939 7 42.14 3 41.33 4
1940 7 42.14 4 40.5 4
1941 7 46.29 2 43.67 7
1942 7 50.57 5 49.67 5
1943 6 66 4 64.6 4
1944 6 75 3 73.6 4
1945 6 65.5 4 62.6 5
1946 6 51.5 3 49.6 5
1947 6 63.83 3 62.2 4
1948 6 57 3 56.2 4
1949 6 56.67 3 54.4 3
1950 6 65.83 4 63.4 5
1951 6 66.17 2 62.2 3
1952 6 65.33 3 61.2 4
1953 6 66.5 3 60.8 4
1954 6 60.83 4 56.8 4
1955 6 68.5 4 67.2 4
1956 6 73.33 3 70.4 4
1957 6 76.83 5 74.4 5
1958 6 74.5 5 72.6 5
1959 6 82.67 4 80 4
1960 6 75.83 3 74.8 3
1961 6 81.33 4 78.6 4
1962 6 77 3 75.6 5
1963 6 76 3 74 4
1964 6 79.17 3 77.2 4
1965 6 74.67 4 72.2 4
1966 6 79.5 2 76 5
1967 6 73.33 3 68.6 4
1968 12 74.75 7 73.64 7
1969 12 90.83 5 87.64 6
1970 12 79.92 4 76.27 6
1971 14 92.43 6 87.85 6
1972 14 93.57 7 90.54 8
1973 16 95.12 5 92.8 10
1974 16 90.06 4 86.4 6
1975 18 100.83 7 98.82 8
1976 18 101.28 9 99.88 10
1977 18 93 8 90.47 8
1978 18 92.83 6 90.53 7
1979 17 100.88 8 98.81 8
1980 21 100.14 8 98.3 9
1981 21 106.05 7 103.15 10
1982 21 113.62 9 108.7 9
1983 21 105.33 8 100.8 12
1984 21 109.05 9 104.25 11
1985 21 111.38 7 106.55 7
1986 21 112.52 8 107.4 9
1987 21 97.48 8 93.2 11
1988 21 107.24 8 104.2 11
1989 21 109.86 8 105.4 8
1990 21 105.81 8 104 9
1991 21 102.43 10 99.4 11
1992 22 99.95 10 98.48 11
1993 24 117.12 11 115.26 11
1994 26 98.42 11 97.16 11
1995 26 53.19 10 52.52 12
1996 26 104.5 11 102.24 12
1997 26 88.08 10 86.72 11
1998 26 78.15 11 77.2 13
1999 27 82.56 13 80.85 14
2000 28 77.79 13 77.11 13
2001 30 86.43 13 85.24 13
2002 30 74.9 16 74.17 16
2003 30 82.57 13 81.76 13
2004 30 74.67 15 74 15
2006 30 91.57 12 90.41 13
2007 30 91 14 90 15
2008 30 84.23 11 83.28 12
2009 30 84.2 12 83.21 13
2010 30 83.4 14 82.41 15
2011 30 77.47 10 76.55 11
2012 30 76.83 15 75.72 18
2013 30 48.07 13 47.66 17

One mod I am thinking of is using T-1 for split league years, PCHA and WHA, and the "war years".

What do you think of this?


Last edited by BM67: 12-20-2013 at 11:07 AM.
BM67 is offline   Reply With Quote
Old
03-21-2012, 11:40 AM
  #34
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 24,986
vCash: 500
Quote:
Originally Posted by BM67 View Post
Using the number of teams to set the mark. We start at either #1 or #2 through to T, the number of teams, and use the average as the "vs#2" number.

The table shows how many points that would be, and where that would sit in the scoring race.

Year Teams #1-#T Rank #2-#T Rank
1918 3 44.67 3 43 3
1919 3 29.67 2 28 2
1920 4 43 3 41 3
1921 4 39.75 3 38.67 4
1922 4 39.5 2 37.33 4
1923 4 32.5 3 31 3
1924 4 22.75 3 22.33 3
1925 6 38.83 5 37.4 5
1926 7 31.14 3 29.33 5
1927 10 29.7 6 28.89 6
1928 10 32.7 6 30.67 6
1929 10 25.5 7 24.78 7
1930 10 56.6 6 54.78 7
1931 10 41.8 6 40.78 7
1932 8 46.88 6 46 6
1933 9 40.89 6 39.75 6
1934 9 41.22 4 39.88 4
1935 9 45 4 43.5 6
1936 8 39.38 4 38.57 4
1937 8 40.62 5 39.86 6
1938 8 42.75 4 41.43 5
1939 7 42.14 3 41.33 4
1940 7 42.14 4 40.5 4
1941 7 46.29 2 43.67 7
1942 7 50.57 5 49.67 5
1943 6 66 4 64.6 4
1944 6 75 3 73.6 4
1945 6 65.5 4 62.6 5
1946 6 51.5 3 49.6 5
1947 6 63.83 3 62.2 4
1948 6 57 3 56.2 4
1949 6 56.67 3 54.4 3
1950 6 65.83 4 63.4 5
1951 6 66.17 2 62.2 3
1952 6 65.33 3 61.2 4
1953 6 66.5 3 60.8 4
1954 6 60.83 4 56.8 4
1955 6 68.5 4 67.2 4
1956 6 73.33 3 70.4 4
1957 6 76.83 5 74.4 5
1958 6 74.5 5 72.6 5
1959 6 82.67 4 80 4
1960 6 75.83 3 74.8 3
1961 6 81.33 4 78.6 4
1962 6 77 3 75.6 5
1963 6 76 3 74 4
1964 6 79.17 3 77.2 4
1965 6 74.67 4 72.2 4
1966 6 79.5 2 76 5
1967 6 73.33 3 68.6 4
1968 12 74.75 7 73.64 7
1969 12 90.83 5 87.64 6
1970 12 79.92 4 76.27 6
1971 14 92.43 6 87.85 6
1972 14 93.57 7 90.54 8
1973 16 95.12 5 92.8 10
1974 16 90.06 4 86.4 6
1975 18 100.83 7 98.82 8
1976 18 101.28 9 99.88 10
1977 18 93 8 90.47 8
1978 18 92.83 6 90.53 7
1979 17 100.88 8 98.81 8
1980 21 100.14 8 98.3 9
1981 21 106.05 7 103.15 10
1982 21 113.62 9 108.7 9
1983 21 105.33 8 100.8 12
1984 21 109.05 9 104.25 11
1985 21 111.38 7 106.55 7
1986 21 112.52 8 107.4 9
1987 21 97.48 8 93.2 11
1988 21 107.24 8 104.2 11
1989 21 109.86 8 105.4 8
1990 21 105.81 8 104 9
1991 21 102.43 10 99.4 11
1992 22 99.95 10 98.48 11
1993 24 117.12 11 115.26 11
1994 26 98.42 11 97.16 11
1995 26 53.19 10 52.52 12
1996 26 104.5 11 102.24 12
1997 26 88.08 10 86.72 11
1998 26 78.15 11 77.2 13
1999 27 82.56 13 80.85 14
2000 28 77.79 13 77.11 13
2001 30 86.43 13 85.24 13
2002 30 74.9 16 74.17 16
2003 30 82.57 13 81.76 13
2004 30 74.67 15 74 15
2006 30 91.57 12 90.41 13
2007 30 91 14 90 15
2008 30 84.23 11 83.28 12
2009 30 84.2 12 83.21 13
2010 30 83.4 14 82.41 15
2011 30 77.47 10 76.55 11

One mod I am thinking of is using T-1 for split league years, PCHA and WHA, and the "war years".

What do you think of this?
This is what I call not being a slave.

Sorry I didn't respond to this earlier.

I am a fan of a system that recognizes that the relative quality of a 3rd place finish 50 years ago might only be about as good as a 6th today. With comparables in the 10th-16 range the last 20 seasons, I suspect this may have taken it too far. I think that the size of the talent pool has expanded at a rate at least somewhat similar the same rate as the league, but not exactly, so doesn't this end up falling into the "10th place in the O6? Ha, that's like 50th in today's 30 team league!" trap? We all hate that argument.

On first glance it looks like seasons like 1970, 1971, 1974 and 1989 suffer from a lot of outliers and outlier beneficiaries at the top, as that "rank" number spikes downwards compared to its progression.

It is a good start. I think that the ranking number we use should go up throughout time; maybe just not this drastically. I know it's a case of trying to devise a system and sticking to it rather than fudging it. That's fine. It might even be an overall improvement. To say for sure, I'd need to see numbers such as how many players meet the 50, 60, 70, 80 benchmarks per season.

seventieslord is online now   Reply With Quote
Old
04-09-2012, 12:00 AM
  #35
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
Seventieslord,

I was shocked to find out that Eric Lindros' 6th place finish somehow scores 100 in a system that is supposed to compare players against the #2 scorer.

I was equally shocked that Luc Robitaille (who never finished above 5th) scores 100 in a season.

Even if you're going by the "remove players who look like outliers," I can't see any season where "2" is really #6 and 1989 is the only one where I can see justification in using the 5th place scorer.

I get removing Lemieux and Jagr in 1996 but why would you not then use Sakic as the standard?

TheDevilMadeMe is online now   Reply With Quote
Old
04-09-2012, 12:24 AM
  #36
Rob Scuderi
Registered User
 
Rob Scuderi's Avatar
 
Join Date: Sep 2009
Country: United States
Posts: 2,788
vCash: 500
Quote:
Originally Posted by TheDevilMadeMe View Post
Seventieslord,

I was shocked to find out that Eric Lindros' 6th place finish somehow scores 100 in a system that is supposed to compare players against the #2 scorer.

I was equally shocked that Luc Robitaille (who never finished above 5th) scores 100 in a season.

Even if you're going by the "remove players who look like outliers," I can't see any season where "2" is really #6 and 1989 is the only one where I can see justification in using the 5th place scorer.

I get removing Lemieux and Jagr in 1996 but why would you not then use Sakic as the standard?
Looking at the specific seasons you mention, you've definitely got me stumped. Like I said, I'm fairly ignorant on this topic but Lindros's 100% in '96 seems strange.

Question though, after removing outliers do we still use the second best left? Or is the first non-outlier given the #2 designation?

Robitaille's 100% is in '92. Here's the top scorers.

1. Mario Lemieux*-PIT 131
2. Kevin Stevens-PIT 123
3. Wayne Gretzky*-LAK 121
4. Brett Hull*-STL 109
5. Luc Robitaille*-LAK 107
Mark Messier*-NYR 107
7. Jeremy Roenick-CHI 103
Steve Yzerman*-DET 103
9. Brian Leetch*-NYR 102
10. Adam Oates-TOT 99

If Hull becomes #1 with outliers then I'd say Robitaille earned that 100% for sure, but if Hull is also an outlier then I'm back to being confused. I'm sure this is 70s favorite topic by now but hopefully he can explain a bit better.

Rob Scuderi is offline   Reply With Quote
Old
04-09-2012, 12:37 AM
  #37
BenchBrawl
joueur de hockey
 
BenchBrawl's Avatar
 
Join Date: Jul 2010
Posts: 9,362
vCash: 500
I hate the fact of completely taking out outliers , in a way the fact they are outliers in your era advantage you against second-tier players from no-outlier's eras because their superstars aren't destroying them enough.

BenchBrawl is offline   Reply With Quote
Old
04-09-2012, 08:02 AM
  #38
BM67
Registered User
 
BM67's Avatar
 
Join Date: Mar 2002
Location: In "The System"
Country: Canada
Posts: 4,579
vCash: 500
Quote:
Originally Posted by Bring Back Scuderi View Post
Looking at the specific seasons you mention, you've definitely got me stumped. Like I said, I'm fairly ignorant on this topic but Lindros's 100% in '96 seems strange.

Question though, after removing outliers do we still use the second best left? Or is the first non-outlier given the #2 designation?

Robitaille's 100% is in '92. Here's the top scorers.

1. Mario Lemieux*-PIT 131
2. Kevin Stevens-PIT 123
3. Wayne Gretzky*-LAK 121
4. Brett Hull*-STL 109
5. Luc Robitaille*-LAK 107
Mark Messier*-NYR 107
7. Jeremy Roenick-CHI 103
Steve Yzerman*-DET 103
9. Brian Leetch*-NYR 102
10. Adam Oates-TOT 99

If Hull becomes #1 with outliers then I'd say Robitaille earned that 100% for sure, but if Hull is also an outlier then I'm back to being confused. I'm sure this is 70s favorite topic by now but hopefully he can explain a bit better.
70s treats Mario and Gretzky as outliers by name, so they are out automatically. Stevens played with Mario, so he's out. That leaves Hull as the #1. Then Robitaille is also removed as an outlier, and #2 would be Messier.

If you replaced Mario and Wayne with Trottier and Dionne, the 70s method would give a #2 of Stevens, as 131 is not an outlier score other than by who did it, and Stevens is only an outlier by who he played with.

BM67 is offline   Reply With Quote
Old
04-09-2012, 08:54 AM
  #39
Sturminator
I voted for Kodos
 
Sturminator's Avatar
 
Join Date: Feb 2002
Location: West Egg, New York
Country: Ukraine
Posts: 7,386
vCash: 500
Quote:
Originally Posted by BM67 View Post
70s treats Mario and Gretzky as outliers by name, so they are out automatically. Stevens played with Mario, so he's out. That leaves Hull as the #1. Then Robitaille is also removed as an outlier, and #2 would be Messier.

If you replaced Mario and Wayne with Trottier and Dionne, the 70s method would give a #2 of Stevens, as 131 is not an outlier score other than by who did it, and Stevens is only an outlier by who he played with.
Yes, I think Stevens is the #2 this season, and that's it. I see no sensible reason to adjust this season in a Vs2 analysis. Ignoring the actual performance of Gretzky/Lemieux in any given season and ignoring everyone with whom they played simply removes too much top offensive talent from the analysis, and ends up leaving the remaining players looking better than they really were.

Sturminator is offline   Reply With Quote
Old
04-09-2012, 02:13 PM
  #40
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
I have mixed feelings about removing Mario/Stevens from 1994 - they were a step up from the pack but not a big step. Either way, I see no reason to also remove Brett Hull. This controversy over outliers is why I think the VS5 method could be better - you never have to worry about outliers in VS5.

TheDevilMadeMe is online now   Reply With Quote
Old
04-09-2012, 02:22 PM
  #41
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
Here is 1996:

1. Lemieux 161
2. Jagr 149
3. Sakic 120
4. Francis 119
5. Forsberg 116
6. Lindros 115
7. Selanne 108
7. Kariya 108

I would imagine 70s methodology is to remove Mario and his teammate Jagr. Sakic is removed as the #1 remaining. Francis and Forsberg are removed as teammates of guys in front of them. So Lindros becomes the standard

I strongly disagree with this methodology. I think if you're using VS2, you need to compare to the second place scorer except in extreme circumstances. Jagr in 1996 actually is extreme, so then why not just go down to the next name and compare to Sakic? It sure doesn't look like Sakic, Francis, and Forsberg are outliers to me.

TheDevilMadeMe is online now   Reply With Quote
Old
04-09-2012, 04:08 PM
  #42
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 24,986
vCash: 500
Quote:
Originally Posted by TheDevilMadeMe View Post
Here is 1996:

1. Lemieux 161
2. Jagr 149
3. Sakic 120
4. Francis 119
5. Forsberg 116
6. Lindros 115
7. Selanne 108
7. Kariya 108

I would imagine 70s methodology is to remove Mario and his teammate Jagr. Sakic is removed as the #1 remaining. Francis and Forsberg are removed as teammates of guys in front of them. So Lindros becomes the standard

I strongly disagree with this methodology. I think if you're using VS2, you need to compare to the second place scorer except in extreme circumstances. Jagr in 1996 actually is extreme, so then why not just go down to the next name and compare to Sakic? It sure doesn't look like Sakic, Francis, and Forsberg are outliers to me.
BM and sturm answered your first inquiry.

Re: Forsberg, I can't believe I did that, but apparently I did. I should have left him in as the #2 after outliers.

I think the idea that a Mario Lemieux-fueled Kevin Stevens should be the player that you judge others' offensive totals by, is ridiculous.

Quote:
you never have to worry about outliers in VS5.
sure you do. I can get behind a #5 system in theory (aside from my concerns raised in the Turgeon thread in the HOH section) but that doesn't make the outliers any less damaging to the system. I would personally still want to remove them. The goal is to not punish players for playing in a league with players who break the mold. If Gretzky and Lemieux did not exist, the most reasonable outcome would be Hull, Messier, Roenick, Yzerman, Leetch.

seventieslord is online now   Reply With Quote
Old
04-09-2012, 05:10 PM
  #43
TheDevilMadeMe
Global Moderator
 
TheDevilMadeMe's Avatar
 
Join Date: Aug 2006
Location: Brooklyn
Country: United States
Posts: 41,030
vCash: 500
I just don't understand why you don't use Joe Sakic as the standard in 1996. And if you count Mario Lemieux and Kevin Stevens as outliers in 1992, surely Brett Hull was not also one?

Quote:
sure you do. I can get behind a #5 system in theory (aside from my concerns raised in the Turgeon thread in the HOH section) but that doesn't make the outliers any less damaging to the system. I would personally still want to remove them. The goal is to not punish players for playing in a league with players who break the mold. If Gretzky and Lemieux did not exist, the most reasonable outcome would be Hull, Messier, Roenick, Yzerman, Leetch.
Why? Name a single season in history where #5 is an outlier. 1989 is the only season I can think of where even #4 is an outlier


Last edited by TheDevilMadeMe: 04-09-2012 at 06:13 PM.
TheDevilMadeMe is online now   Reply With Quote
Old
04-09-2012, 06:06 PM
  #44
BraveCanadian
Registered User
 
BraveCanadian's Avatar
 
Join Date: Jun 2010
Country: Canada
Posts: 8,419
vCash: 500
Quote:
Originally Posted by TheDevilMadeMe View Post
Here is 1996:

1. Lemieux 161
2. Jagr 149
3. Sakic 120
4. Francis 119
5. Forsberg 116
6. Lindros 115
7. Selanne 108
7. Kariya 108

I would imagine 70s methodology is to remove Mario and his teammate Jagr. Sakic is removed as the #1 remaining. Francis and Forsberg are removed as teammates of guys in front of them. So Lindros becomes the standard

I strongly disagree with this methodology. I think if you're using VS2, you need to compare to the second place scorer except in extreme circumstances. Jagr in 1996 actually is extreme, so then why not just go down to the next name and compare to Sakic? It sure doesn't look like Sakic, Francis, and Forsberg are outliers to me.
That would be because they aren't.

BraveCanadian is offline   Reply With Quote
Old
04-09-2012, 06:08 PM
  #45
vecens24
Registered User
 
vecens24's Avatar
 
Join Date: Jun 2009
Country: United States
Posts: 5,002
vCash: 500
I agree with TDMM that Sakic should be the standard that season. That makes more sense to me.

vecens24 is offline   Reply With Quote
Old
04-09-2012, 08:16 PM
  #46
seventieslord
Moderator
 
seventieslord's Avatar
 
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 24,986
vCash: 500
Quote:
Originally Posted by TheDevilMadeMe View Post
I just don't understand why you don't use Joe Sakic as the standard in 1996.
Because Sakic is the #1 non-outlier, not #2.

Quote:
And if you count Mario Lemieux and Kevin Stevens as outliers in 1992, surely Brett Hull was not also one?
sorry, but why?

Quote:
Why? Name a single season in history where #5 is an outlier. 1989 is the only season I can think of where even #4 is an outlier
If #5 is behind nothing but outliers, then that could very well be as impressive as #1 in a season without outliers. Therefore, selecting them as the standard instead of the "#2" would likely be disadvantaging that year, to the extent that the player you want me to use is ahead of the player I want to use.

the more I think about all this, and how we are trying to make this "fit" what we think it should look like (as in, number of players by year who are 60+, 70+, 80+, etc) we almost need to go to an "adjusted" rankings system. Maybe that would please everyone...

All we'd need to do is agree on what kinds of multipliers should be used from era to era. i.e. if you're 8th in 1930, that's like __ by 1960, __ by 1980, and ___ in today's league.

seventieslord is online now   Reply With Quote
Old
05-09-2012, 09:10 AM
  #47
BM67
Registered User
 
BM67's Avatar
 
Join Date: Mar 2002
Location: In "The System"
Country: Canada
Posts: 4,579
vCash: 500
This illustrates how much moving the "decade" by one season can change things.

Player GP G A Pts Yrs G1 G2 G3 G4 A1 A2 A3 A4 P1 P2 P3 P4
Number 1 2001-2011 820 532 757 1,103 10 1000% 1106% 1138% 1204% 1000% 1123% 1175% 1225% 1000% 1042% 1114% 1157%
Number 2 2001-2011 820 482 673 1,060 10 911% 1000% 1029% 1090% 898% 1000% 1047% 1092% 960% 1000% 1069% 1110%
Number 3 2001-2011 820 468 642 991 10 885% 972% 1000% 1060% 858% 956% 1000% 1043% 901% 939% 1000% 1039%
Number 4 2001-2011 820 442 617 954 10 836% 919% 946% 1000% 824% 919% 960% 1000% 867% 903% 963% 1000%
Number 1 2002-2012 820 533 755 1,091 10 1000% 1117% 1171% 1223% 1000% 1126% 1187% 1233% 1000% 1052% 1105% 1159%
Number 2 2002-2012 820 478 669 1,039 10 903% 1000% 1047% 1095% 895% 1000% 1056% 1096% 951% 1000% 1051% 1102%
Number 3 2002-2012 820 457 633 988 10 866% 958% 1000% 1047% 848% 948% 1000% 1038% 907% 953% 1000% 1049%
Number 4 2002-2012 820 437 611 943 10 827% 916% 957% 1000% 818% 915% 965% 1000% 865% 909% 954% 1000%

BM67 is offline   Reply With Quote
Old
05-10-2012, 06:05 PM
  #48
BM67
Registered User
 
BM67's Avatar
 
Join Date: Mar 2002
Location: In "The System"
Country: Canada
Posts: 4,579
vCash: 500
Here are the top 30 scorers for the post-lockout seasons.

Year#1#2#3#4#5#6#7#8#9#10#11#12#13#14#15#16#17#18#19#20#21#22#23#24#25#26#27#28#29#30Avg 1-30
2006125123106103103102100989793929190908989878786858585848180807979797991.57
2007120114108105102100100969695949492918787878584848383838281818080787891
20081121069897969292928987848383838282828180787878777675747472727284.23
200911311010397949191898888868482828280807979787877777675747473737384.2
2010112109109101959491898886858584838282817977767675717171707070707083.4
2011104999894918685808077767676767373737373737171717069686867676677.47
2012109979384838281818178787777777676767473717069696969676767676776.83

With the exception of a tie for 2nd lowest at #3 between 2008 and 2011, the lowest two scores at all spots come in the last two seasons. Indeed the #30 from 2006 & 2007 would be in the top 10 in 2011 & 2012.

13 of the lowest totals come in 2011, and 13 come in 2012, with 4 ties. The biggest gap between the lowest and 2nd lowest comes at #4 where the 84 in 2012 is a full 10 points back of the 94 put up in 2011.

The points average from 1-30 has dropped every year since 2006.

BM67 is offline   Reply With Quote
Old
02-07-2013, 05:51 AM
  #49
Sturminator
I voted for Kodos
 
Sturminator's Avatar
 
Join Date: Feb 2002
Location: West Egg, New York
Country: Ukraine
Posts: 7,386
vCash: 500
I'm reviving this thread because I think we need another round of discussion on this subject. The Vs2 method is starting to become pretty comfortable to us in the ATD and its use among the GMs is becoming relatively widespread. We might want to discuss some kind of "official ATD standard" for Vs2 statistical analysis so as to avoid misunderstandings when discussing player statistics. I will continue the previous argument.

Quote:
Originally Posted by seventieslord View Post
Quote:
Here is 1996:

1. Lemieux 161
2. Jagr 149
3. Sakic 120
4. Francis 119
5. Forsberg 116
6. Lindros 115
7. Selanne 108
7. Kariya 108

I would imagine 70s methodology is to remove Mario and his teammate Jagr. Sakic is removed as the #1 remaining. Francis and Forsberg are removed as teammates of guys in front of them. So Lindros becomes the standard

I strongly disagree with this methodology. I think if you're using VS2, you need to compare to the second place scorer except in extreme circumstances. Jagr in 1996 actually is extreme, so then why not just go down to the next name and compare to Sakic? It sure doesn't look like Sakic, Francis, and Forsberg are outliers to me.

I just don't understand why you don't use Joe Sakic as the standard in 1996.
Because Sakic is the #1 non-outlier, not #2.
I think seventies is making a very basic rational mistake here. The point of the Vs2 system is to remove outliers and then compare performance against the highest non-outlier. In this case, that is very obviously Joe Sakic. Completely removing Mario and Jagr and then treating Sakic as the #1 and removing him as well is just goofy. The whole point is to compare to the first non-outlier. Sakic is that guy.

Sturminator is offline   Reply With Quote
Old
02-07-2013, 06:55 AM
  #50
Dreakmur
Registered User
 
Dreakmur's Avatar
 
Join Date: Mar 2008
Location: Orillia, Ontario
Country: Canada
Posts: 7,993
vCash: 500
Quote:
Originally Posted by Sturminator View Post
I'm reviving this thread because I think we need another round of discussion on this subject. The Vs2 method is starting to become pretty comfortable to us in the ATD and its use among the GMs is becoming relatively widespread. We might want to discuss some kind of "official ATD standard" for Vs2 statistical analysis so as to avoid misunderstandings when discussing player statistics. I will continue the previous argument.



I think seventies is making a very basic rational mistake here. The point of the Vs2 system is to remove outliers and then compare performance against the highest non-outlier. In this case, that is very obviously Joe Sakic. Completely removing Mario and Jagr and then treating Sakic as the #1 and removing him as well is just goofy. The whole point is to compare to the first non-outlier. Sakic is that guy.
Agreed. The point of using #2 is based off the assumption that #1 is often an outlier.

I always use #2, or the first non-outlier.

Dreakmur is offline   Reply With Quote
Reply

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 05:33 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.