In a recent game against XG, I found myself in the following position:
| | is XG Roller
score: 0 pip: 222 |
| | | | | | | | | | | | | | |
| | | | | | 3 point match |
| | | | | | | | | | | | | | | | pip: 97 score: 0
is Matt |
| XGID=-bdBBBCAAABAc---ab-c------:0:0:1:31:0:0:0:3:10 |
to play 31 |
Being a fundamentally timid carbon-based lifeform, I booked up the prime with 10/7 9/8. I was somewhat startled when XG Roller dinged this innocuous looking play as a blunder, preferring the broken 10/9 8/5:
1. | XG Roller | 10/9 8/5 | eq: +0.835 |
| Player: Opponent: | 58.78% (G:39.12% B:12.62%) 41.22% (G:2.32% B:0.08%) | |
|
2. | XG Roller | 9/5 | eq: +0.821 (-0.014) |
| Player: Opponent: | 58.82% (G:37.99% B:12.90%) 41.18% (G:2.83% B:0.10%) | |
|
3. | XG Roller | 11/8 6/5 | eq: +0.802 (-0.033) |
| Player: Opponent: | 58.42% (G:38.50% B:13.23%) 41.58% (G:2.31% B:0.08%) | |
|
4. | XG Roller | 8/5 6/5 | eq: +0.796 (-0.039) |
| Player: Opponent: | 58.29% (G:38.20% B:12.54%) 41.71% (G:2.50% B:0.08%) | |
|
5. | XG Roller | 10/7 9/8 | eq: +0.727 (-0.108) |
| Player: Opponent: | 56.39% (G:36.41% B:15.20%) 43.61% (G:2.09% B:0.08%) | |
|
eXtreme Gammon Version: 1.11, MET: eXtremeGammon
So I bumped up my settings to XG Roller+:
1. | XG Roller+ | 9/5 | eq: +0.708 |
| Player: Opponent: | 57.06% (G:35.38% B:12.18%) 42.94% (G:2.67% B:0.11%) | |
|
2. | XG Roller+ | 11/8 6/5 | eq: +0.706 (-0.002) |
| Player: Opponent: | 56.40% (G:35.70% B:12.81%) 43.60% (G:2.06% B:0.08%) | |
|
3. | XG Roller+ | 10/7 9/8 | eq: +0.702 (-0.007) |
| Player: Opponent: | 56.50% (G:35.46% B:12.23%) 43.50% (G:1.69% B:0.06%) | |
|
4. | XG Roller+ | 10/9 8/5 | eq: +0.675 (-0.033) |
| Player: Opponent: | 56.00% (G:35.34% B:11.59%) 44.00% (G:2.08% B:0.08%) | |
|
5. | XG Roller+ | 8/5 6/5 | eq: +0.662 (-0.047) |
| Player: Opponent: | 55.45% (G:34.98% B:12.76%) 44.55% (G:2.07% B:0.08%) | |
|
eXtreme Gammon Version: 1.11, MET: eXtremeGammon
This looks promising. The bot has now decided my play isn't so bad. But it still prefers the very radical 9/5. I'd better do a rollout:
1. | Rollout1 | 10/9 8/5 | eq: +0.722 |
| Player: Opponent: | 56.32% (G:36.99% B:15.23%) 43.68% (G:1.21% B:0.12%) | Conf: ± 0.048 (+0.674...+0.770) Duration: 3 hours 10 minutes 01 second |
|
2. | Rollout1 | 10/7 9/8 | eq: +0.697 (-0.025) |
| Player: Opponent: | 56.79% (G:36.07% B:14.73%) 43.21% (G:1.57% B:0.13%) | Conf: ± 0.046 (+0.651...+0.743) Duration: 2 hours 29 minutes 53 seconds |
|
3. | Rollout1 | 9/5 | eq: +0.652 (-0.070) |
| Player: Opponent: | 56.02% (G:35.45% B:14.91%) 43.98% (G:2.13% B:0.17%) | Conf: ± 0.048 (+0.604...+0.700) Duration: 3 hours 11 minutes 15 seconds |
|
4. | XG Roller+ | 11/8 6/5 | eq: +0.706 (-0.016) |
| Player: Opponent: | 56.40% (G:35.70% B:12.81%) 43.60% (G:2.06% B:0.08%) | |
|
5. | XG Roller+ | 8/5 6/5 | eq: +0.662 (-0.060) |
| Player: Opponent: | 55.45% (G:34.98% B:12.76%) 44.55% (G:2.07% B:0.08%) | |
|
|
1 1296 Games rolled with Variance Reduction. Moves and cube decisions: 3 ply
|
eXtreme Gammon Version: 1.11, MET: eXtremeGammon
Okay, now I'm officially perplexed. My PC is from the ark, so it takes ages to extend the rollouts. Perhaps if they were longer they'd settle on something that less peculiar? Or is my neat looking prime somehow the wrong idea? Does anybody buy what XG is selling here? If so, can you explain it to me?