Post-consolidation VsX Benchmarks

[seventieslord]

This is a good start. But we should get these numbers going back to 1927 (and yes, eventually work on something for the pre-merger era as well).
*
Also, it’s important that we see this on an annual basis. As you said and as we all know, the league depth has changed gradually over the years. We need to see if there are big swings in the numbers of players meeting these thresholds from year to year, in order to understand which seasons have a benchmark that may be set too easy or too difficult.

 

[bluesfan94]

Yeah, I haven't had time to finish it yet, but I should be able to soon.

 

[Sturminator]

The ugly ducklings from this period, going by the >50% standard are:

1970-71: 55
1971-72: 33
1972-73: 66

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

1985-86: 59
1986-87: 92
1987-88: 65

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

1997-98: 89
1998-99: 63
1999-00: 97

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

2001-02: 116
2002-03: 83
2003-04: 115

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

Not sure what to make of the above. Interestingly, all of these are seasons where the standard Vs2 benchmark is used - meaning that the top of the scoring table was quite normal. None of the seasons with the Vs3 or the averaged benchmark show up as strange, which speaks for the sensibility of the system.

- in 1971-72, there was a big gap between the #7 and #8 scorers. A better normalization might average the top 14 (ack!) scorers in that season. I dunno.

- I can't find anything strange about 1986-87.

- 1998-99 looks quite normal, as well.

- Same for 2002-03

Other than 1971-72, the only strange data here looks like the result of normal fluctuations, rather than a problem with the benchmarking. I may need to revisit 1971-72, though.

 

[Sturminator]

Cross-posting this from the main draft thread so BBS's work doesn't get lost.

Here's the weighted VsX scores for best 7 years

1. Gretzky 155.9
2. Esposito 125.0
3. Lemieux 121.8
4. Beliveau 109.8
5. Mikita 109.6
6. Morenz 106.8
7. Cowley 105.3
8. Dionne 104.1
9. Sakic 99.1
10. Lach 97.0

11. Bentley 96.4
12. Boucher 95.9
13. Trottier 94.9
14. Thornton 94.6
15. Yzerman 94.4
16. Apps 94.2
17. Forsberg 91.7
18. Stewart 91.1
19. Oates 90.8
20. Barry 90.6

21. Messier 90.2
22. Ullman 89.5
23. Schmidt 89.3
24. Stastny 89.0
25. Ratelle 88.8
26. Clarke 88.5
27. Francis 88.3
28. Richard 87.5
29. Hawerchuk 86.6
30. Savard 86.4

31. Lindros 86.3
32. Delvecchio 85.5
33. Sittler 85.2
34. Perreault 85.2
35. Crosby 84.0
36. undrafted 83.6
37. Gilmour 83.2
38. Sedin 82.9
39. Sundin 82.7
40. Turgeon 82.6

41. Kennedy 82.4
42. Modano 82.1
42. Roenick 82.0
44. Fedorov 82.0
45. Nicholls 81.7
46. Weiland 81.3
47. H. Smith 79.9
48. Lafontaine 79.8
49. Datsyuk 79.7
50. Weight 79.5

51. Watson 79.1
52. B. Richards 79.1
53. Lecavalier 78.2
54. Federko 77.8
55. McKenney 76.5
56. Lemaire 76.5
57. Goyette 76.0
58. Malkin 75.9
59. Keon 73.4
60. Brind'Amour 73.4

61. Macleish 73.2
62. P. Mahovlich 71.7
63. Niewendyk 70.9
64. B. Smith 69.6
65. Muller 65.4

I used the percentages Dreakmur gave me for Morenz, but haven't looked at any of the other pre-merger guys. Hooley Smith, for example, could pick up another score higher than his best 7 from after 1927. Cowley's best season (1941, not a war year) gives him a score of 140.9.

This is what the VsX benchmarks generate for the best scoring centers of all-time after using seventieslord's peak weighting system and my own favored seven year cutoff (which I think is most fair to older players). Just to summarize the peak weighting system:

The 1st, 2nd, 3rd, 4th, 5th, 6th and 7th best seasons are multiplied by 20, 18, 17, 16, 15, 14 and 13, respectively, and then the whole thing is divided by 113 (if I'm doing the math correctly) to arrive at a final number.

I'm quite happy with the results thus far.

 

[Rob Scuderi]

I had some time on my hands with a long break between picks, here are 10 best weighted seasons according to VsX. I haven't looked around a roster thread so some of these guys may not be top sixers. Some of the undrafteds will go in the ATD, but I wouldn't be surprised if some don't - maybe they should though. Either way I'll edit them in if and when they're selected.

1. G. Howe 122.9
2. Jagr 108.9
3. Richard 103.0
4. Bo. Hull 102.8
5. Lindsay 100.4
6. Bathgate 99.9
7. Lafleur 95.0
8. Bossy 91.1
9. Selanne 90.3
10. Blake 89.3
11. Bill Cook 89.0
12. C. Conacher 87.9
13. Bentley 87.8
14. Geoffrion 86.8
15. Recchi 86.0
16. Br. Hull 85.8
17. Schriner 85.2
18. Bucyk 84.1
19. S. Howe 83.9
20. Jackson 83.7
21. Kurri 83.6
22. Iginla 83.3
23. Mahovlich 82.9
24. Robitaille 82.3
25. St. Louis 81.6
26. Moore 81.3
27. R. Conacher 81.3
28. Kariya 81.0
29. Abel 80.4
30. Kovachulchuk 79.7
31. Fleury 79.6
32. Alfredsson 79.5
33. Gilbert 78.8
34. Joliat 78.5
35. Hossa 78.5
36. Bure 77.3
37. Mosienko 77.1
38. Mar. Naslund 77.0
39. Shanahan 76.1
40. (undrafted) 75.8
41. Tkachuk 75.6
42. Heatley 75.2
43. Elias 75.0
44. Mogilny 75.0
45. LeClair 74.9
46. Hodge 74.7
47. Goulet 74.6
48. Thompson 73.7
49. L. Patrick 73.6
50. Palffy 73.2
51. Olmstead 72.9
52. Hextall 72.7
53. Damphousse 72.2
54. Ovechkin 71.9
55. Bun Cook 71.4
56. Rousseau 71.3
57. Dumart 71.0
58. Dillon 70.9
59. Sedin 70.7
60. Demita 70.7
61. Gottselig 70.6
62. Cournoyer 70.3
63. McDonald 70.2
64. Larmer 69.8
65. Amonte 69.8
66. Hejduk 69.7
67. H. Lewis 69.6
68. Drillon 69.4
69. Middleton 69.3
70. Bondra 69.0
71. Ciccarelli 68.5
72. Mullen 68.2
73. G. Anderson 67.7
74. Andreychuk 67.5
75. Barber 66.8
76. Gartner 66.8
77. Prentice 66.3
78. Taylor 66.3
79. Aurie 66.2
80. (undrafted) 66.0
81. Propp 65.5
82. Wharram 65.1
83. Shutt 64.8
84. Gaborik 64.6
85. Nolan 64.4
86. R. Martin 64.2
87. Doan 64.0
88. J. Ward 62.8
89. Robert 62.2
90. Stevens 61.8
91. Verbeek 61.8
92. G. Stewart 61.7
93. Cashman 61.0
94. Litzenberger 60.9
95. Guerin 60.9
96. Northcott 60.2
97. Bauer 60.1
98. Hadfield 60.1
99. Nevin 60.0
100. Pronovost 59.9
101. K. Nilsson 59.8
102. Armstrong 59.5
103. Roberts 59.3
104. Tocchet 58.7
105. S. Smith 58.7
106. Simmer 58.6
107. Stasiuk 58.4
108. (undrafted) 57.9
109. H. Watson 57.4
110. Neely 57.0
111. Nash 56.4
112. Bailey 55.7
113. Duff 55.6
114. Gare 55.3
115. Tonelli 55.2
116. Mats Naslund 54.8
117. Kerr 54.7
118. Gillies 54.4
119. C. Lemieux 52.3
120. Tikkanen 50.8

Mosienko is the first name that stuck out to me, and having your top three scores as '44, '45, '46 will do that. I understand he's not drafted for his scoring, but I don't see why Tikkanen gets a pass with Gillies an Duff being maligned. Bathgate so high is also interesting. Even if you cut it down to best 7 he still beats Lafleur 104.7 to 99.4

 

[Sturminator]

Mosienko is the first name that stuck out to me, and having your top three scores as '44, '45, '46 will do that. I understand he's not drafted for his scoring, but I don't see why Tikkanen gets a pass with Gillies an Duff being maligned. Bathgate so high is also interesting. Even if you cut it down to best 7 he still beats Lafleur 104.7 to 99.4

Tikkanen shouldn't really get a pass. He's not fit for an ATD scoringline, in my opinion, and has even the above stats inflated by playing with Gretzky - Kurri for a good portion of his prime.

Bathgate's standing here is quite interesting. I don't have a problem with it. Maybe the biggest distortion that the VsX project revealed was the craziness of Bathgate's era. That he consistently managed to land in the "outlier tier" (and he was the only player outside of Montreal or Detroit to do so) in spite of receiving very little offensive help in New York suggests strongly that the rankings system has been underrating him, and he looks very strong even in that system. Bathgate was a huge talent trapped on a bad team. Is he Guy Lafleur without the playoff career? It wouldn't surprise me, at all.

edit: I believe Daniel Alfredsson is missing. Beyond that, excellent work, BBS.

 

[Sturminator]

Syd Howe's score is missing.

Frank Mahovlich is lower than I had expected, though his value as a scorer is obviously elevated by his goalscoring bias (goals being generally more valuable than assists) and by the fact that he played in a highly defensive system in Toronto for about half of his prime. Very interesting results. If you've got the time, I'd be interested to see the results based on a seven season standard, as well.

 

[TheDevilMadeMe]

Tikkanen shouldn't really get a pass. He's not fit for an ATD scoringline, in my opinion, and has even the above stats inflated by playing with Gretzky - Kurri for a good portion of his prime.

Bathgate's standing here is quite interesting. I don't have a problem with it. Maybe the biggest distortion that the VsX project revealed was the craziness of Bathgate's era. That he consistently managed to land in the "outlier tier" (and he was the only player outside of Montreal or Detroit to do so) in spite of receiving very little offensive help in New York suggests strongly that the rankings system has been underrating him, and he looks very strong even in that system. Bathgate was a huge talent trapped on a bad team. Is he Guy Lafleur without the playoff career? It wouldn't surprise me, at all.

edit: I believe Daniel Alfredsson is missing. Beyond that, excellent work, BBS.

Maybe Bathgate's stats properly rate him, and it just so happened that other than Bathgate, all the all-time great talent in the league was on either Montreal or Detroit. This was the decade that Montreal went to the finals 10 years in a row, after all.

 

[Sturminator]

Maybe Bathgate's stats properly rate him, and it just so happened that other than Bathgate, all the all-time great talent in the league was on either Montreal or Detroit. This was the decade that Montreal went to the finals 10 years in a row, after all.

Maybe. There are three years where softer benchmarking due to averaging affects Bathgate's VsX score:

1956-57:
1. Howe - 89
2. Lindsay - 85
3. Beliveau - 84
4. Bathgate - 77
5. Litzenberger - 64
6. M. Richard - 62
7. McKenney - 60
8. Moore - 58
...average as benchmark: 72

1957-58:
1. Moore - 84
2. H. Richard - 80
3. Bathgate - 78
4. Howe - 77
5. Horvath - 66
6. Litzenberger - 62
7. MacKell - 60
8. Delvecchio - 59
...average as benchmark: 71

1958-59:
1. Moore - 96
2. Beliveau - 91
3. Bathgate - 88
4. Howe - 78
5. Litzenberger - 77
6. Geoffrion - 66
7. Sullivan - 63
8. Hebenton - 62
8. Sloan - 62
8. McKenney - 62
...average as benchmark: 75

...and Bathgate is far from the only player affected by the softer benchmarking in this period. The standard seems to handle all of the other players involved in an appropriate way, wouldn't you say? Raising the benchmarks back up to the old standard (so the #2 scorer in all cases) would savage Dickie Moore and Ed Litzenberger quite specifically, and I'm not sure if that passes the smell test. They look about right where they are right now. Maybe Bathgate has simply been underrated? It's kind of ridiculous the point totals he put up in the relative wasteland of New York. I believe he led the entire league in scoring (narrowly over Howe) over his eight year peak in an era where the elite scorers were of a very high quality.

 

[Rob Scuderi]

Fixed Syd. I added Roy Conacher and I believe that should be everyone.

 

[overpass]

Fixed Syd. I added Roy Conacher and I believe that should be everyone.

Toe Blake?

 

[Rob Scuderi]

Toe Blake?

Sorry getting a pulse on your third and fourth line wingers was more important

I will edit him in, one of the big names missing in the first posting somehow.

 

[overpass]

Sorry getting a pulse on your third and fourth line wingers was more important

I will edit him in, one of the big names missing in the first posting somehow.

Thanks. I checked the top of the list and knew he was missing.

Gotta remind the ATD that Blake wasn't just a coach.

 

[Sturminator]

Mosienko is the first name that stuck out to me, and having your top three scores as '44, '45, '46 will do that.

Yeah, the numbers seem to be consistently overrating wartime players, but then again, so did the old rankings system. The 1940's as a whole don't look so bad, but the war...yeah. I still haven't come up with any universal, systematic way to address this problem, and I somehow doubt that I ever will. The entire percentages system assumes a level of competition that I just don't think was present. Asteriks make me a sad bunny, but I'm honestly at a loss.

 

[Sturminator]

Maybe Bathgate's stats properly rate him, and it just so happened that other than Bathgate, all the all-time great talent in the league was on either Montreal or Detroit. This was the decade that Montreal went to the finals 10 years in a row, after all.

I've looked at this some more:

1954-55:
2. M. Richard - 74

1955-56:
1. Beliveau - 88
2. Howe - 79
3. M. Richard - 71
...benchmark = M. Richard: 71

1956-57:
1. Howe - 89
2. Lindsay - 85
3. Beliveau - 84
4. Bathgate - 77
5. Litzenberger - 64
6. M. Richard - 62
7. McKenney - 60
8. Moore - 58
...average as benchmark: 72

1957-58:
1. Moore - 84
2. H. Richard - 80
3. Bathgate - 78
4. Howe - 77
5. Horvath - 66
6. Litzenberger - 62
7. MacKell - 60
8. Delvecchio - 59
...average as benchmark: 71

1958-59:
1. Moore - 96
2. Beliveau - 91
3. Bathgate - 88
4. Howe - 78
5. Litzenberger - 77
6. Geoffrion - 66
7. Sullivan - 63
8. Hebenton - 62
8. Sloan - 62
8. McKenney - 62
...average as benchmark: 75

1959-60:
2. Horvath - 80

The averaged benchmarks seem to follow a smoother progression of scoring than using the #2 scorers. The #2 scorers in the two years prior to the run of averaged benchmarks scored 74 and 71 points, respectively. The averaged scores more-or-less remain at this level, and tick up slightly in 1958-59, blending relatively smoothly into the #2 score for 1959-60.

Throwing out the averaged benchmarks and going with the #2 scorers for this period leaves us with a huge jump of 14 points from 1955-56 to 1956-57, and then a large fall of 10 points from 1958-59 to 1959-60. This is quite uneven, and suggests that the #2 scorers over this three year period were, in fact, legitimate outliers. I'm happy with the averaged benchmarks over this period. They are likely imperfect, but they seem better than just going with the #2 scores. I think Bathgate has simply been historically underrated because we didn't realize before just how often during his prime he landed in a tier of scorers that was far ahead of "the pack".

 

[Sturminator]

I have generated a little graph showing the differences between the handling of benchmarks in the VsX system and the old Vs2 system. Hopefully, it is relatively readable:

 

[Sturminator]

Looking through the data some more, I believe one way of identifying when the league was going through upheavals is to look at when the scoring changes drastically from one season to the next. I've always struggled a bit to identify the beginning and end of the war years, for example, but looking at scoring fluctuations seems to make the matter fairly clear.

- Beginning of war years: 1941-42

- End of war years: 1945-46

Why do I say that? Well, look at the scoring trends. Before the 1941-42 season, our scoring benchmarks are remarkably flat - fluctuating over a two point range for the preceding five seasons: 45, 44, 44, 43, 44, which suggests that through this period, the league was relatively stable in terms of overall scoring levels, and the "star" talent against which our benchmarks are set.

Then we have the war years, where league scoring jumps crazily for three years in a row, before crashing back down over the next two seasons. Benchmarks during the war years from 1941-42 to 1945-46: 54, 66, 77, 63, 52

And then, starting in 1946-47, we see relative stability again, punctuated by the tainted 1948-49 season (which I'm still not sure how to handle), before the next scoring leap in 1954-55. Benchmarks over this period: 63, 60, 54, 69, 66, 69, 61, 61.

This is all rough stuff, but the fluctuations in our benchmarks also coincide with wild swings in league scoring over this period, as well. I think 1941-42 to 1945-46 is our period of questionable data, but I think maybe we should leave the rest of the 1940's alone.

edit: ehhh...I'd better control for games played before going any further.

 

[Sturminator]

Ok, I've got myself an SSRS report with backing data set up, and I have been playing around with some numbers today, generating different line charts. My data at this point is: year, VsX, Vs2, League Goals, League Games, League Teams. From those six data points, I have been able to generate quite a lot of useful data. Here is maybe the best thing I've done so far:

I deleted the legend because SSRS legends suck, and it was just taking up space. So here is the Legend:

- the jagged blue line is the VsX benchmark / (League Goals/League Games). Basically, I divide the VsX benchmark by Goals/Game as a way of determining how big a "slice of the pie" the benchmark scorers were getting in each season.

- The tiered line with circles is the number of teams in the league

- the straight lines are something I just threw in there - basically they denote certain tier averages during different eras of NHL history.

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

Have a look. I know the graphic is a bit hard to read because it's a lot of information crammed into a small space, but the results are fascinating. The NHL has twice had a major tier jump, where benchmark scorers receive a good chunk more of the league scoring, and never give those gains back.

- the first time is the 1949-50 season. I dug into this a bit more, and I noticed that not only were the books being cooked in Chicago, but league assist/goals numbers were way up all across the board. I believe this was the season in which the NHL began counting second assists. The introduction of the second assist into a points calculation would give the illusion that the benchmark scorers were getting a bigger slice of the scoring pie, when in fact nothing had changed. That is, I believe what happens at the 1949-50 point in the graph.

- the second time is expansion. The benchmark / goals-per-game goes from about 13 to about 18, and it never goes back. This suggests that, as TDMM had suggested, O6 "star" scorers (who are the targets of our benchmark) got a smaller slice of the pie than post-expansion stars. Why should this be? I would guess it because the league was deeper in scorers, so the "stars" didn't stand out as much from the 3rd liners and such.

Anyway, fascinating results thus far. The VsX methodology does seem to handle the pre-expansion superstars quite well, but the above should show us that we have to be careful about applying it to anybody below that lofty standard. Some star scorers may have had considerably more chances than others, depending on team dynamics, and without very specific information, VsX data on second tier scorers is likely to be pretty polluted.

 

[seventieslord]

There were goals with as many as three assists awarded in the 1930s, IIRC, so while there may be something unique about 1950, I don't think the introduction of the 2nd assist is it.

Sturm, can I have your sheet so that I can do some analysis? You can send it to me at the same address I emailed you from.

Wish I had more time to digest all this. busiest week of the year at work.

 

[seventieslord]

I just had a bit of a revelation. This all started when Nalyd proposed throwing out a player's best season entirely. While I couldn't back it unconditionally, I saw bits of merit in it.

If you were to rank all the best scorers' best seasons in order (as in, here is Denis Savard, and here are his ten best seasons, in order), which of those 10 seasons seems to correlate most closely with how good of an all-time player (or at least scorer) the player is?

Is it is his best season? I'd say no. A ton of players had a best season that is in no way indicative of their proven sustained ability (Yzerman, Rob Brown, Nicholls, in 1989 for starters, then there's Cashman, Bucyk, Kevin Stevens, LeCavalier, etc.).

What about their 10th-best? I'd say no as well. Some guys like Pierre Turgeon and Ron Francis have much better 10th-best seasons than a fair number of players typically considered better than they are.

My gut is telling me this: once could be an accident, twice could be a coincidence, three times is a habit. I think that if you were trying to make a catch-all offensive prime metric, there is a solid case for weighing a players' 3rd-best season most heavily. Something like this:

1st: 11
2nd: 12
3rd: 13
4th: 12
5th: 11
6th: 10
7th: 9
8th: 8
9th: 7
10th: 6

A weighting system like this would achieve what we are trying to achieve, which is weigh peak, prime and longevity to reasonable degrees, but it also avoids over-rewarding one and two-year wonders.

Thoughts?

 

[Sturminator]

I just had a bit of a revelation. This all started when Nalyd proposed throwing out a player's best season entirely. While I couldn't back it unconditionally, I saw bits of merit in it.

If you were to rank all the best scorers' best seasons in order (as in, here is Denis Savard, and here are his ten best seasons, in order), which of those 10 seasons seems to correlate most closely with how good of an all-time player (or at least scorer) the player is?

Is it is his best season? I'd say no. A ton of players had a best season that is in no way indicative of their proven sustained ability (Yzerman, Rob Brown, Nicholls, in 1989 for starters, then there's Cashman, Bucyk, Kevin Stevens, LeCavalier, etc.).

What about their 10th-best? I'd say no as well. Some guys like Pierre Turgeon and Ron Francis have much better 10th-best seasons than a fair number of players typically considered better than they are.

My gut is telling me this: once could be an accident, twice could be a coincidence, three times is a habit. I think that if you were trying to make a catch-all offensive prime metric, there is a solid case for weighing a players' 3rd-best season most heavily. Something like this:

1st: 11
2nd: 12
3rd: 13
4th: 12
5th: 11
6th: 10
7th: 9
8th: 8
9th: 7
10th: 6

A weighting system like this would achieve what we are trying to achieve, which is weigh peak, prime and longevity to reasonable degrees, but it also avoids over-rewarding one and two-year wonders.

Thoughts?

That's actually a really good idea. Essentially, we're applying the central insight behind the VsX methodology to our benchmarking across a different axis - that of individual careers. Yes, I like it. A couple more comments:

- I would still break off the analysis at seven seasons when comparing across all post-consolidation eras, for reasons which have been discussed extensively. Going out to ten seasons when comparing post-expansion players seems reasonable, as the length of a typical productive prime has definitely increased.

- I would also introduce a further methodology: in any season (up to 7 or 10 or X) where a player falls below the 50% threshhold, simply award him a score of 50. This avoids punishing players with a 0 score in any season, which I believe can have a severely distoriting effect on the values of guys with short careers. Simply set 50 as the lower bound of our benchmarking values. It seems fair, especially in combination with a methodology which weighs outwards from the 3rd best season (so doesn't overrate guys with very short peaks). 50 is a crappy score (something like replacement value?) and will hurt any player in a comparison if he's got more than a couple of them (cough...Malkin...cough), but it is not as distorting as simply having a 0.

I'm a huge nerd, so this is exciting stuff to me. Hopefully, you can make something out of the report I've sent you. I feel like we're making good progress in coming up with a much more sensible system for evaluating scoring across eras on a year-to-year and career-to-career basis. I appreciate your help, and it's also great to get assistance from newer GMs like BBS and bluesfan. Keep up the good work, guys.

 

[TheDevilMadeMe]

I just had a bit of a revelation. This all started when Nalyd proposed throwing out a player's best season entirely. While I couldn't back it unconditionally, I saw bits of merit in it.

If you were to rank all the best scorers' best seasons in order (as in, here is Denis Savard, and here are his ten best seasons, in order), which of those 10 seasons seems to correlate most closely with how good of an all-time player (or at least scorer) the player is?

Is it is his best season? I'd say no. A ton of players had a best season that is in no way indicative of their proven sustained ability (Yzerman, Rob Brown, Nicholls, in 1989 for starters, then there's Cashman, Bucyk, Kevin Stevens, LeCavalier, etc.).

What about their 10th-best? I'd say no as well. Some guys like Pierre Turgeon and Ron Francis have much better 10th-best seasons than a fair number of players typically considered better than they are.

My gut is telling me this: once could be an accident, twice could be a coincidence, three times is a habit. I think that if you were trying to make a catch-all offensive prime metric, there is a solid case for weighing a players' 3rd-best season most heavily. Something like this:

1st: 11
2nd: 12
3rd: 13
4th: 12
5th: 11
6th: 10
7th: 9
8th: 8
9th: 7
10th: 6

A weighting system like this would achieve what we are trying to achieve, which is weigh peak, prime and longevity to reasonable degrees, but it also avoids over-rewarding one and two-year wonders.

Thoughts?

Hmmm.... why not just weight the top 5 seasons equally, then start a scale that slides down from there? There is something historically significant about having one really great season and I don't think it should be downgraded too much. But you're right in that it probably shouldn't be weighed higher than other top 5 seasons.

- I would also introduce a further methodology: in any season (up to 7 or 10 or X) where a player falls below the 50% threshhold, simply award him a score of 50. This avoids punishing players with a 0 score in any season, which I believe can have a severely distoriting effect on the values of guys with short careers. Simply set 50 as the lower bound of our benchmarking values. It seems fair, especially in combination with a methodology which weighs outwards from the 3rd best season (so doesn't overrate guys with very short peaks). 50 is a crappy score (something like replacement value?) and will hurt any player in a comparison if he's got more than a couple of them (cough...Malkin...cough), but it is not as distorting as simply having a 0.

Great idea.

 

[bluesfan94]

Sorry this took so long. Between midterms and spring break and fantasy baseball, this got pushed to the backburner.

 

[TheDevilMadeMe]

I was trying to figure out what to do with Clint Smith's two good seasons during the two worst war years, and I just have to go back to what I think seventieslord did - for the two years we know are really affected by the war - 1944 and 1945 - just compare to the #1 highest scorer. You're basically assuming that in the war years, #1 is not an outlier by definition and if there are big gaps between the top scorers, it's because of a lack of depth of talent, rather than the guys at the top being outliers. And I think that's more accurate.

It doesn't really help 1942, 1943, and 1946 which are partial war years to various extents, but I think it provides a decent shorthand for the two worst years.

 

[Rob Scuderi]

I was trying to figure out what to do with Clint Smith's two good seasons during the two worst war years, and I just have to go back to what I think seventieslord did - for the two years we know are really affected by the war - 1944 and 1945 - just compare to the #1 highest scorer. You're basically assuming that in the war years, #1 is not an outlier by definition and if there are big gaps between the top scorers, it's because of a lack of depth of talent, rather than the guys at the top being outliers. And I think that's more accurate.

It doesn't really help 1942, 1943, and 1946 which are partial war years to various extents, but I think it provides a decent shorthand for the two worst years.

I honestly think you just have to asterisk it. No amount of fudging can account for the crappy level of competition the scores were racked up against. You go from giving Smith a 94 to an 88 in 1944 and 86 to a 74 in 1945. I agree its an improvement, but the bottom line is those scores don't deserve to be treated the same way as other scores in the 70s or 80s earned in a non-depleted league with much different competition at the top and throughout the league.

I think no matter what you do, you're still going to have Mosienko too high and undrafteds that look much better than they were.