
BGonline.org Forums
PR differencies and winning ratios
Posted By: Rick Janowski
Date: Saturday, 5 May 2012, at 9:05 a.m.
In Response To: PR differencies and winning ratios (Maik Stiebler)
I have a large amount of data from IIBGF matches which shows very good correlation with linear relationship of 33 ELO difference = 1 pr difference (assumed in eXtremeGammon), from ELO ratings from about 1550 upwards.
Whereas plugging into the normal ELO equation should give a reasonable estimate, I think a more refined approach would be to use unequal opponent match equity charts as included in the excellent book "Can a Fish Taste Twice as Good?" by Jake Jacobs and the late lamented Walter Trice. This has the benefit of potentially modelling situations where the pr/ELO edge is fundamentally checker play based or cube based (addressing some of Stick's comments). Moreover, the methodology is theoreticaly sounder than the more simplistic ELO formula approach particularly for shorter matches. Essentially, the idea is to consult the equal scores (eg, 5 away, 5 away, etc) in the match equity tables appropriate to the estimated ELO difference to estimate the match winning chances at a particular score. If the ELO difference falls between two tables, then the match equity can be estimated by interpolation.

BGonline.org Forums is maintained by Stick with WebBBS 5.12.