View Single Post
Old 05-10-2008, 12:11 PM   #5 (permalink)
MrX
SBR MVP
 
MrX's Avatar
 
Join Date: 01-10-06
Location: Kakapoopoopeepeeshire
Posts: 1,249
MrX is online now
Default

Quote:
Originally Posted by St. Andrew View Post
3. Then, take the average of the runs given up (#1) and the opponent's mean runs scored (#2), round to the nearest run, and get a projected game score.
This step is fatally flawed. Say league average runs scored is 4.5/game. Now say you have team A that scores 6 runs/game playing team B that allows 5.5 runs/game. You're method predicts 5.75 runs.

This is obviously not right. Team A scores 6 versus average pitching. Now they are facing sub-par pitching and they will score less?

What you want to do is something more like this: Expected runs = RS*(RA/LA) = 6*(5.5/4.5) = 7.33

I will bite my tongue on the sample size issues as per your request.
__________________
I'm completely in favor of the separation of Church and State. My idea is that these two institutions screw us up enough on their own, so both of them together is certain death. --George Carlin
Reply With Quote