Thanks, Neil, and also Peter. I guess there is a tradeoff between time and accuracy in rollouts, especially with a less than optimal computer. Do you agree that the "roll for both no double and dbl" box can be unchecked?
I did another rollout with 3 ply, but with the instruction to stop after one play is 98%+. (That's why there are only 340 games.) But the "conf." is +/- .027. If that is the same as margin of error, it's pretty high if we're talking about a difference of .044.
        
    
          
 |  is HardToGet
score: 2
pip: 175 | | 7 point match | pip: 151
score: 0
 is blotto |
|
| XGID=--bbaCBBBB-----cA-AcbBb---:1:-1:1:61:0:2:0:7:10 |
| to play 61 |
| 1. | Rollout1 | 18/12 5/4* | eq: +0.955 |
| Player:
Opponent: | 82.63% (G:32.33% B:1.15%)
17.37% (G:3.35% B:0.19%) | Conf: ± 0.027 (+0.928...+0.982)
Duration: 13 minutes 22 seconds |
|
| 2. | Rollout1 | 16/10 5/4* | eq: +0.911 (-0.044) |
| Player:
Opponent: | 80.67% (G:32.00% B:1.33%)
19.33% (G:3.88% B:0.26%) | Conf: ± 0.029 (+0.882...+0.940)
Duration: 12 minutes 13 seconds |
|
| 3. | XG Roller+ | 18/11 | eq: +0.834 (-0.121) |
| Player:
Opponent: | 79.75% (G:26.83% B:1.28%)
20.25% (G:3.97% B:0.15%) | |
|
| 4. | 2 ply | 18/17 16/10 | eq: +0.752 (-0.203) |
| Player:
Opponent: | 76.91% (G:25.69% B:1.13%)
23.09% (G:4.09% B:0.09%) | |
|
| 5. | 2 ply | 16/9 | eq: +0.730 (-0.225) |
| Player:
Opponent: | 76.02% (G:25.56% B:1.07%)
23.98% (G:3.75% B:0.07%) | |
|
| |
1 340 Games rolled with Variance Reduction.
First 1 moves and cube decisions: 3 ply
Remaining moves and cube decisions: 3 ply
|
eXtreme Gammon Version: 1.20, MET: eXtremeGammon