Originally posted by Eric is God
Formula seems sorta simple . . . you should add in standard deviations for squad performace, and correlation coefficients between certain players or squads. Of course then you run into the problem of heteroscedasticy and some multi-collinearity but as long as you could change the formula to a log-form and put in a couple proxy variables it should be fine.
All of the preceding were actual words
Honestly though, looks nice. If Blizzard uses it, it must be good.
Formula seems sorta simple . . . you should add in standard deviations for squad performace, and correlation coefficients between certain players or squads. Of course then you run into the problem of heteroscedasticy and some multi-collinearity but as long as you could change the formula to a log-form and put in a couple proxy variables it should be fine.
All of the preceding were actual words

Honestly though, looks nice. If Blizzard uses it, it must be good.
Comment