Thanks to Stick, sebalotek, Phil, Keene, Casper, ah_clem, Fatboy, Kevin, Mike M., Steven, garyo and Seth, for their responses and quotes, and to Mike Mannon for supplying and/or extending rollouts.
1. | Rollout1 | 8/5 6/5 | eq: +0.402 |
| Player: Opponent: | 59.97% (G:23.15% B:1.66%) 40.03% (G:14.42% B:0.76%) | Conf.: ¡Ó 0.004 (+0.399...+0.406) - [100.0%] Duration: 2 hours 14 minutes |
|
2. | Rollout1 | 24/21 11/10 | eq: +0.386 (-0.016) |
| Player: Opponent: | 58.46% (G:18.86% B:0.58%) 41.54% (G:8.55% B:0.36%) | Conf.: ¡Ó 0.003 (+0.384...+0.389) - [0.0%] Duration: 2 hours 37 minutes |
|
? |
1 31104 Games rolled with Variance Reduction. Moves: 3-ply, cube decisions: XG Roller
|
1. | Rollout1 | 24/21 10/9 | eq: +0.409 |
| Player: Opponent: | 59.17% (G:18.39% B:0.57%) 40.83% (G:8.28% B:0.34%) | Conf.: ¡Ó 0.003 (+0.406...+0.412) - [100.0%] Duration: 2 hours 42 minutes |
|
2. | Rollout1 | 8/5 6/5 | eq: +0.393 (-0.016) |
| Player: Opponent: | 59.45% (G:23.87% B:1.66%) 40.55% (G:14.55% B:0.77%) | Conf.: ¡Ó 0.003 (+0.390...+0.397) - [0.0%] Duration: 2 hours 15 minutes |
|
? |
1 31104 Games rolled with Variance Reduction. Moves: 3-ply, cube decisions: XG Roller
|
?
Above, you can see rollouts of the two positions (side by side, if you sufficiently widen your window). In the first position, making the 5pt is better than anchoring by .016. Exactly the opposite is true in the second position.
The basic idea is that it is inefficient to end up with a builder on the 10pt. If Blue anchors in the first position (moving his builder to the 10pt), his offensive 31 and 63 will be duplicated for the next turn (assuming he is not hit). Conversely, if Blue makes the 5pt in the second position (leaving his builder on the 10pt), his 63 will be duplicated (anchor or 7pt).
You can see a more complete accounting of Blue's follow-up numbers in Mike's post.
Here is a good way to drive the point home: In the first diagrammed position, what would you prefer to roll as Blue: 21, 31 or 32? Rank the three rolls.
The answer is that (of the three) 21 is best, 32 is second best, and 31 is worst.
---------------------------------------------------------------------
.....Roll...Equity...Anchoring play
.....21.....+.434.....(builder on 11pt)
.....32.....+.409.....(builder on 9pt)
.....31.....+.386.....(builder on 10pt) or +.402 (make 5pt)
---------------------------------------------------------------------
I did a single-play 5k rollout for 21 (anchoring with 24/21), generating the +.434 equity listed. The other equities have been extracted from the two rollouts above. Blue would rather anchor if his builder ends up on the 11pt or 9pt. If his choice is for the builder to end up on the 10pt, he should settle for making his 5pt, though that is still worse than both anchor+11pt and anchor+9pt.
It is a similar story at DMP, as you can glean from the side-by-side rollout results below. (Multiply the margins by 2.2 to compare their significance to the money margins from the rollouts above.)
1. | Rollout1 | 8/5 6/5 | eq: +0.191 |
| Player: Opponent: | 59.54% (G:26.38% B:6.26%) 40.46% (G:15.27% B:2.14%) | Conf.: ¡Ó 0.003 (+0.188...+0.194) - [100.0%] Duration: 19 minutes 28 seconds |
|
2. | Rollout1 | 24/21 11/10 | eq: +0.172 (-0.019) |
| Player: Opponent: | 58.58% (G:20.31% B:1.75%) 41.42% (G:9.56% B:0.74%) | Conf.: ¡Ó 0.002 (+0.169...+0.174) - [0.0%] Duration: 19 minutes 12 seconds |
|
? |
1 5184 Games rolled with Variance Reduction. Dice Seed: 48040528 Moves: 3-ply, cube decisions: XG Roller
|
1. | Rollout1 | 24/21 10/9 | eq: +0.186 |
| Player: Opponent: | 59.28% (G:19.76% B:1.93%) 40.72% (G:9.52% B:0.73%) | Conf.: ¡Ó 0.001 (+0.184...+0.187) - [100.0%] Duration: 57 minutes 55 seconds |
|
2. | Rollout1 | 8/5 6/5 | eq: +0.177 (-0.009) |
| Player: Opponent: | 58.83% (G:26.32% B:6.82%) 41.17% (G:15.31% B:2.19%) | Conf.: ¡Ó 0.002 (+0.175...+0.178) - [0.0%] Duration: 59 minutes 00 second |
|
? |
1 15552 Games rolled with Variance Reduction. Dice Seed: 48040528 Moves: 3-ply, cube decisions: XG Roller
|
?
[As you might expect, it is clearly best to make the 5pt at gammon go, or anchor at gammon save, in both positions. To see the effect in both positions when Blue leads ¡V6¡V7 and ¡V5¡V7, we can look forward to viewing the rollouts Steven Carey mentioned here.]
Finally, what would you do in each of the two positions diagrammed below? For answers and analysis, see Backgammon Openings, pages 62¡V63.
?
Nack