Tim's rollout page using gnu 2-ply rollout has hitting on the ace a .025 error. My XG rollout below has it as only half as bad (.013) and after 1296 trials it's still within the margin of error. I'll throw some more cycles at it over the weekend.
I'll also dig out my copy of Robertie and see what he had to say about it. One thing i did miss in my analysis above is that white has a made 5 point, and the better board makes hitting loose more costly.
| | White is Player 2
score: 0 pip: 152 | Unlimited Game Jacoby | pip: 157 score: 0
Blue is Player 1 | |
XGID=-aa---EBBA--cC--ac-cc---AA:0:0:1:53:0:0:1:0:10 |
Blue to play 53 |
1. | Rollout1 | Bar/22 13/8 | eq: -0.488 |
| Player: Opponent: | 39.63% (G:11.15% B:0.52%) 60.37% (G:23.57% B:1.44%) | Conf.: ± 0.016 (-0.504...-0.472) - [87.2%] Duration: 41 minutes 39 seconds |
|
2. | Rollout1 | Bar/22 6/1* | eq: -0.501 (-0.013) |
| Player: Opponent: | 39.09% (G:10.60% B:0.48%) 60.91% (G:23.20% B:1.73%) | Conf.: ± 0.017 (-0.518...-0.485) - [12.8%] Duration: 45 minutes 02 seconds |
|
3. | 3-ply | Bar/22 9/4 | eq: -0.567 (-0.079) |
| Player: Opponent: | 37.95% (G:9.48% B:0.41%) 62.05% (G:24.88% B:1.60%) | |
|
4. | 1-ply | Bar/22 7/2* | eq: -0.975 (-0.487) |
| Player: Opponent: | 32.51% (G:8.27% B:0.43%) 67.49% (G:30.46% B:3.47%) | |
|
5. | 1-ply | Bar/22 8/3 | eq: -1.000 (-0.512) |
| Player: Opponent: | 29.77% (G:7.02% B:0.32%) 70.23% (G:34.95% B:4.02%) | |
|
|
1 1296 Games rolled with Variance Reduction. Moves: 3-ply, cube decisions: XG Roller
|
eXtreme Gammon Version: 2.10