|
BGonline.org Forums
Rollout Results Posted
Posted By: Stick In Response To: Rollout Results Posted (Paul Weaver)
Date: Thursday, 30 January 2014, at 6:39 a.m.
There's a reason we should let rollouts go until they reach 100% CI. I almost had this gripe in a few other recent posts but skipped it. People were quoting the difference in plays when there was still a ton of variance left. Variance is reported for a reason. Don't be too confident of your results until you have something significant. I am not confident where the results of my rollout will end up. All I knew is I saw Paul's rollout and disagreed with it so I put on my own, same settings but twice as many trials. Especially if you disagree with a play make sure to roll it out to significance. (and beyond!)
White is Player 2
score: 0
pip: 1659 point match pip: 139
score: 0
Blue is Player 1XGID=-b--a-FAC--AcC---c-cc-A---:0:0:1:53:0:0:0:9:10 Blue to play 53
1. Rollout1 11/6 7/4* eq: -0.312
Player:
Opponent:43.08% (G:13.25% B:1.02%)
56.92% (G:19.44% B:0.84%)Conf.: ± 0.011 (-0.323...-0.301) - [77.1%]
Duration: 35 minutes 19 seconds2. Rollout1 8/3 6/3 eq: -0.318 (-0.006)
Player:
Opponent:42.84% (G:14.05% B:0.91%)
57.16% (G:19.56% B:1.11%)Conf.: ± 0.011 (-0.329...-0.307) - [22.9%]
Duration: 31 minutes 50 seconds1 2592 Games rolled with Variance Reduction.
Dice Seed: 81654656
Moves: 3-ply, cube decisions: XG RollereXtreme Gammon Version: 2.10, MET: Kazaross XG2
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.