Whileee
Registered User
- May 29, 2010
- 46,075
- 33,132
Thanks for the important clarification. Any way to stratify the data to see if what looks like improvement in Lemieux' effectiveness is bearing out in the analytics?As even the developer of that model would tell you:
1) It's descriptive, not predictive (it uses on-ice GF/60 RAPM without any sh% regression as a large source of offense)
2) It's not relative to ice time (Ex: Petan has a higher GAR/60 than Lemieux)
3) RAPM would be the model you would use for that type of analysis to test your hypothesis.
RAPM:
I like your idea of trying Petan up the lineup, or perhaps like this...
Ehlers-Scheifele-Wheeler
Perreault-Little-Laine
Petan-Copp-Roslovic
Tanev-Lowry-Appleton