| |
BGonline.org Forums
Results-Based PR
Posted By: Rick Janowski In Response To: Results-Based PR (Maik Stiebler)
Date: Sunday, 1 February 2015, at 9:11 a.m.
The use of average opponents' PR serves to normalise the VR win rate whilst also proving a parameter broadly comparable with normal PR. Why is this a problem and what are the alternatives?
By the way - taking the value from "Should have won" seems to be problematic at times (XG bug?). I now think you need to calculate the weighted mean for all match lengths, but XG provides the necessary information. Looking at the data from the Mochy Club matches, I think this method may require at least 3 times the sample size to give similar reliability levels as PR.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.