|
BGonline.org Forums
XGR++ data
Posted By: eXtreme Gammon In Response To: Are full rollouts obsolete? (Chuck Bower)
Date: Thursday, 24 November 2011, at 1:33 a.m.
Here are a few raw numbers that may help with the discussion:
This is based on the work made for the Depreli revised study. So it's based on 500 money games (both sides so 1000 games). Positions were rolled using 3-ply/XGR settings until CI reached 0.005 equity.
Errors made (cumulative numbers, i.e. the number of big errors include also the blunders). The overall number of decisions is 18098
big blunders (greater than 0.120): None blunders (greater than 0.080): 1 Big errors (greater than 0.040): 2 errors (greater than 0.020): 28 small errors (greater than 0.010): 113 More importantly here is how close the equity from XGR++ is from the RO equity. keep in mind that this is only on the 5122 positions analyzed which does not include positions were the best move was in agreement by all programs/level.
between 0.000 and 0.005: 31.17% between 0.005 and 0.015: 31.73% between 0.015 and 0.025: 16.21% between 0.025 and 0.035: 9.31% between 0.035 and 0.045: 4.68% between 0.045 and 0.055: 2.62% between 0.055 and 0.065: 1.23% between 0.065 and 0.075: 0.67% between 0.075 and 0.085: 0.56% between 0.085 and 0.095: 0.49% between 0.095 and 0.105: 0.32% between 0.105 and 0.115: 0.33% between 0.115 and 0.125: 0.29% more than 0.125: 0.38%
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.