Here is a great thread to read through with regards to OPR.
http://www.chiefdelphi.com/forums/sh... +power+rating
Bongle and others did some fantastic work throughout that thread.
I highlighted page 11 because the DPR topic comes up. One problem with the DPR (for last year's game) was too much room for error. While OPR has some areas for error to influence the overall outcome, DPR was based off of OPR results and thus compounded some of the error.
Also last year was fairly unique with the extreme amount of penalties. Unless you had good data on how many G22s, and who did them, the matrix method might just as easily assume you were being defended well.
Now that was last year's game. This year could be totally different.