I got you the first time.
Although i dont know how to improve it, that formula often gets it completely wrong on the day.
Im more interested in how you justify your faith in it, particularly in the face of the points I made that you ignored.
An improvement would be to use the 12th percentile score on the day.....
While i'm at it there is no way an algorithm can acurately determine an ACR. This is the obvious place GA should be applying their millions of historical rounds.