I'll just post the rollouts I have straight off on this because I doubt many people would feel strongly about the play choice.I probably wouldn't have looked at this if the XG analysis settings I was using hadn't sent it to a second pass rollout. The problem is I'm pretty unable to understand how XG is finding such a difference in equity here and how it comes up in a longish race with the candidate play winning the least often being clearly best whilst the one I played, which wins the most, is worst. The GNU rollout is much TCTC to call between any of the plays, as I would expect, and orders them the same way it orders winning chances. Next problem is that I'm wondering why XG and GNU come up with equity values in general that don't look remotely comparable when both rollouts are using the RK MET. I'd appreciate any thoughts on any of this.
# | Ply | Move | Equity |
• | 1 | R | 8/6 8/5 | -0.2741 |
| 0.3905 | 0.0002 | 0.0000 | - | 0.6095 | 0.0003 | 0.0000 | -0.2193 | -0.2741 | 0.0004 | 0.0001 | 0.0000 | - | 0.0004 | 0.0001 | 0.0000 | 0.0009 | 0.0029 | | |
| Full cubeful rollout with var.redn. | |
| 1296 games, Mersenne Twister dice gen. with seed 894974196 and quasi-random dice | |
| Stop when best play is enough JSDs ahead: limit 3.1 (min. 1296 games) | |
| Play: world class 2-ply cubeful prune [world class] | |
| keep the first 0 0-ply moves and up to 8 more moves within equity 0.16 | |
| Skip pruning for 1-ply moves. | |
| Cube: 2-ply cubeful prune [world class] | |
| 2 | R | 13/11 13/10 | -0.2751 ( -0.0010) |
| 0.3899 | 0.0001 | 0.0000 | - | 0.6101 | 0.0003 | 0.0000 | -0.2205 | -0.2751 | 0.0004 | 0.0000 | 0.0000 | - | 0.0004 | 0.0003 | 0.0002 | 0.0009 | 0.0029 | | |
| Full cubeful rollout with var.redn. | |
| 1296 games, Mersenne Twister dice gen. with seed 894974196 and quasi-random dice | |
| Stop when best play is enough JSDs ahead: limit 3.1 (min. 1296 games) | |
| Play: world class 2-ply cubeful prune [world class] | |
| keep the first 0 0-ply moves and up to 8 more moves within equity 0.16 | |
| Skip pruning for 1-ply moves. | |
| Cube: 2-ply cubeful prune [world class] | |
| 3 | R | 13/11 8/5 | -0.2757 ( -0.0016) |
| 0.3899 | 0.0002 | 0.0000 | - | 0.6101 | 0.0004 | 0.0000 | -0.2208 | -0.2757 | 0.0005 | 0.0000 | 0.0000 | - | 0.0005 | 0.0001 | 0.0001 | 0.0009 | 0.0030 | | |
| Full cubeful rollout with var.redn. | |
| 1296 games, Mersenne Twister dice gen. with seed 894974196 and quasi-random dice | |
| Stop when best play is enough JSDs ahead: limit 3.1 (min. 1296 games) | |
| Play: world class 2-ply cubeful prune [world class] | |
| keep the first 0 0-ply moves and up to 8 more moves within equity 0.16 | |
| Skip pruning for 1-ply moves. | |
| Cube: 2-ply cubeful prune [world class] | |
| 4 | R | 13/10 8/6 | -0.2766 ( -0.0026) |
| 0.3891 | 0.0001 | 0.0000 | - | 0.6109 | 0.0008 | 0.0000 | -0.2223 | -0.2766 | 0.0004 | 0.0000 | 0.0000 | - | 0.0004 | 0.0005 | 0.0001 | 0.0009 | 0.0029 | | |
| Full cubeful rollout with var.redn. | |
| 1296 games, Mersenne Twister dice gen. with seed 894974196 and quasi-random dice | |
| Stop when best play is enough JSDs ahead: limit 3.1 (min. 1296 games) | |
| Play: world class 2-ply cubeful prune [world class] | |
| keep the first 0 0-ply moves and up to 8 more moves within equity 0.16 | |
| Skip pruning for 1-ply moves. | |
| Cube: 2-ply cubeful prune [world class] | |