I've done a similar adjustment using the ES and PP data from pick 224 and I've come to a similar conclusion about the draft class. While the placements are different it is overall an incredible top end of the draft, Rossi, Byfield, and Perfetti would be top 3 picks based on their production in any other season. I personally used full point totals for my adjustments as I found in my research at the NHL level that secondary assists were rather noisy, but when you looked at assists as a whole they were more predictive than either A1 or A2.
The top 15 for points per GP adjustments for DY were:
1. Connor Mcdavid - 2.20
2. Marco Rossi - 1.79
3. Mitchell Marner - 1.79
4. Dylan Strome - 1.72
5. Quinton Byfield - 1.69
6. Matthew Tkachuk - 1.60
7. Cole Perfetti - 1.56
8. Taylor Hall - 1.54
9. Ryan Strome - 1.53
10. Tyler Seguin - 1.53
11. John Tavares - 1.47
12. Sam Bennett - 1.47
13. Alex DeBrincat - 1.44
14. Steven Stamkos - 1.43
15. Andrei Svechnikov - 1.43
When you account for estimated toi the rankings change, but there are far less "superstars" in the cohort so I was happy using per game numbers.
For your method did you adjust for league scoring rates or was it just an age adjustment? When I made my age adjustments I used a basic linear model similar to what was outlined
here and the adjustments for scoring were based on the hockey-ref adjustments for
adjusted NHL scoring. You could try and create a more elaborate aging curve, but I'm not sure it's really warranted.