| |
BGonline.org Forums
Are full rollouts obsolete?
Posted By: David Rockwell In Response To: Are full rollouts obsolete? (Chuck Bower)
Date: Wednesday, 23 November 2011, at 3:57 p.m.
I haven't compiled any statistics. Nevertheless, it is my impression based on my work with opening game rollouts, that XGR++ is routinely off by .0010 or more. Perhaps this is sufficient for many types of positions. I'm not satisfied when it comes to the opening game.
Here are some data points where XGR++ wasn't sufficiently accurate. They are all GG to save me to trouble of opening multiple files. A serious study would check other scores.
41S-21 g[$ S10] "<31
41S-21 g[S $4] &
51S-51 g[$ S3] "<31
51S-51 g[S $10] &
61P-41 g[$ S7] "<31
61P-41 g[S $8] &
64S-22 g[N H23] "<15
64S-22 g[N H3] &
32D-65 g[S R5] "<31
32D-65 g[R S5] &
I don't know if it is meaningful or not, but I will note that for each of these five plays, the GG play as preferred by the rollout is different than the correct play for money. XGR++ prefers the money play in all cases above other than 64S-22.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.