When breaking things down like this I like to have rollouts with less variance left in them just because so here's this. So here XG's top play wins 0.14% more games, wins 1.45% fewer gammons (net), and scores 0.5% fewer backgammons (net). So yes, it appears it has to be better due to cube efficiency. To be sure I checked the immediate fanning sequences which XGR gets correct for both plays, easy DP. After that I checked the mediocre entry numbers and they're all easy takes. I didn't check to make sure it was correcting cubing all these situations but it looks good on the dice distribution. Further down this rabbit hole I'm not willing to go.
           
 
         
 | | White is Player 2
score: 0
pip: 153 | Unlimited Game
Jacoby Beaver | pip: 147
score: 0
Blue is Player 1 |
|
| XGID=-b-B--E-C---bC-aabbe----B-:0:0:1:44:0:0:3:0:10 |
| Blue to play 44 |
| 1. | Rollout1 | 24/16* 6/2(2) | eq: +0.642 |
| Player:
Opponent: | 62.17% (G:26.11% B:2.95%)
37.83% (G:8.32% B:0.39%) | Conf.: ± 0.003 (+0.639...+0.644) - [100.0%]
Duration: 20 minutes 24 seconds |
|
| 2. | Rollout1 | 24/16* 8/4(2) | eq: +0.630 (-0.011) |
| Player:
Opponent: | 62.03% (G:28.12% B:3.52%)
37.97% (G:8.88% B:0.46%) | Conf.: ± 0.003 (+0.627...+0.633) - [0.0%]
Duration: 19 minutes 41 seconds |
|
| |
1 46656 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
|