1. | XG Roller++ | 8/3 | eq: -0.569 |
| Player: Opponent: | 36.39% (G:10.25% B:0.52%) 63.61% (G:21.60% B:1.54%) | |
|
2. | XG Roller++ | 24/21 8/6 | eq: -0.583 (-0.015) |
| Player: Opponent: | 37.19% (G:11.47% B:0.81%) 62.81% (G:26.79% B:1.36%) | |
|
3. | XG Roller++ | 8/5 6/4 | eq: -0.589 (-0.020) |
| Player: Opponent: | 37.34% (G:10.83% B:0.49%) 62.66% (G:24.91% B:3.01%) | |
|
4. | XG Roller++ | 6/4 6/3 | eq: -0.592 (-0.024) |
| Player: Opponent: | 36.37% (G:10.13% B:0.49%) 63.63% (G:22.92% B:2.07%) | |
|
5. | XG Roller++ | 13/11 13/10 | eq: -0.604 (-0.035) |
| Player: Opponent: | 35.17% (G:9.97% B:0.51%) 64.83% (G:21.27% B:1.99%) | |
|
I've never seen a rollout differ from XGR++ by 0.045 in an [early-game] position...
Without racking my brain for a more extreme early-game case, I recall this one. A comparison of the diagrams above (XGR++) and below (4-ply rollout) reveals a margin swing of .047.
More remarkable is that the 3-ply rollout and 4-ply rollout margins differ by .038, as illustrated here.
Nack
1. | Rollout1 | 8/5 6/4 | eq: -0.527 |
| Player: Opponent: | 37.96% (G:11.45% B:0.56%) 62.04% (G:25.30% B:3.33%) | Conf.: ± 0.007 (-0.533...-0.520) - [100.0%] Duration: 4 hours 03 minutes |
|
2. | Rollout1 | 8/3 | eq: -0.554 (-0.027) |
| Player: Opponent: | 36.50% (G:10.71% B:0.59%) 63.50% (G:21.76% B:1.59%) | Conf.: ± 0.004 (-0.558...-0.549) - [0.0%] Duration: 2 hours 32 minutes |
|
|
1 10368 Games rolled with Variance Reduction. Dice Seed: 16319810 Moves and cube decisions: 4-ply
|