Using league GPG to adjust is flawed. Most agree that performance vs. peers is the superior way to adjust. I am sure if that is done for Bossy, he would likely win more than one of the Art Rosses awarded since 2005.
I agree, which is why I didn't use the overall league GPG to adjust. I separated EV, PP, and SH scoring. For example, from 1985-86 to 2014-15, EV scoring has dropped 28% (5.61 -> 4.05 GPG), PP scoring has dropped 44% (2.04 -> 1.14 GPG), and SH scoring has dropped 50% (0.28 -> 0.14 GPG). Using overall league GPG would have uniformly dropped all EV, PP, and SH scoring by 33%, which would noticeably underestimate the decline PP and SH scoring (due to the loss of PP opportunities, as you correctly pointed out).
Comparison versus peers is great for determining how much of an outlier a player is, especially for older eras when there's a lot less data around. However, even the performance of high-end players tend to fluctuate, and even a pool of the top ten players is a small sample size. Gretzky's 183-point 1986-87 season is his most dominant year, but I'm hard-pressed to say it's his "best" year. Generally, I'm now more inclined to do multi-season peer comparisons now to average out the fluctuations. And even then, is there as much difference as we think?
Season|1981-82|1982-83|1983-84|1984-85|1985-86|1986-87|Average|2013-14|2014-15
Rank|Adjusted|Adjusted|Adjusted|Adjusted|Adjusted|Adjusted|Adjusted|Actual|Actual
-|-|-|-|-|-|-|-|-|-
Outlier|147|143|144|148|151|138|
145
|
104
|
N/A
Typical 2|102|91|89|97|97|82|
93
|
87
|
87
Typical 3|96|88|87|94|97|81|
91
|
86
|
86
Typical 4|93|84|86|90|90|79|
87
|
84
|
84
Typical 5|90|78|86|86|87|79|
84
|
82
|
81
Typical 6|83|77|81|84|84|78|
81
|
80
|
81
Typical 7|81|77|80|76|83|77|
79
|
80
|
78
Typical 8|80|77|79|76|76|72|
77
|
79
|
77
Typical 9|74|77|76|75|74|72|
75
|
79
|
76
Typical 10|73|75|73|74|71|71|
73
|
79
|
76
Typical 11|73|75|72|73|71|71|
73
|
79
|
73
Typical 12|72|71|70|73|71|69|
71
|
78
|
73
Typical 13|72|70|70|72|70|70|
71
|
76
|
73
Typical 14|68|70|68|72|70|66|
69
|
74
|
73
Typical 15|67|68|67|72|69|66|
68
|
72
|
72
Typical 16|66|67|67|71|69|65|
68
|
70
|
72
Typical 17|65|66|67|70|66|62|
66
|
70
|
71
Typical 18|64|66|66|69|65|64|
66
|
70
|
70
Typical 19|64|65|66|69|63|63|
65
|
69
|
70
Typical 20|64|64|66|69|62|62|
65
|
69
|
70
-|-|-|-|-|-|-|-|-|-
Average 2-10|86|80|82|84|84|77|
82
|
82
|
81
Average 2-20|76|74|75|77|76|71|
75
|
77
|
76
If we're averaging the typical 2-10 highest-scoring players in each season, there's only a one-point variation between the averaged adjusted seasons, 2013-14, and 2014-15 (82, 82, and 81 points respectively).
If we're looking at each adjusted season individually, there's one extraordinarily strong season (1981-82) and one extraordinarily weak season (1986-87). Otherwise, it's a typical 2-10 average of 80-84 points for the adjusted seasons. That's not completely out of line with the last couple of seasons given fluctuations in player performance.
Relevant seasons:
NHL Season
|
EV goals
|
PP goals
|
SH goals
|
Total goals
|
Games in season
|
Assists per goal
1981-82|5.97|1.82|0.23|8.03|80|1.65
1982-83|5.72|1.78|0.23|7.73|80|1.64
1983-84|5.76|1.84|0.29|7.89|80|1.63
1984-85|5.72|1.78|0.28|7.77|80|1.66
1985-86|5.61|2.04|0.28|7.94|80|1.64
1986-87|5.28|1.80|0.26|7.34|80|1.64
2013-14|4.01|1.17|0.16|5.34|82|1.71
2014-15|4.05|1.14|0.14|5.32|82|1.72
EDIT: Also, I'm sorry if it feels like I'm piling on. (I'm honestly not.) It was this discussion that led me into trying a new way of converting player stats. Cheers!