                      |
| ||||
| XGID=----b-E-E---dDaA-b-d-b----:0:0:-1:31:0:0:3:0:10 | |||||
65R-32S-41T-21@-41T-42P-63R-31 | |||||
| 1. | Rollout1 | 13/10* 6/5 | eq: +0.397 | |||
| ||||||
| 2. | Rollout1 | 13/10* 11/10 | eq: +0.366 (-0.030) | |||
| ||||||
| 3. | Rollout1 | 13/10* 10/9 | eq: +0.359 (-0.038) | |||
| ||||||
| 4. | Rollout2 | 11/10* 10/7 | eq: +0.266 (-0.131) | |||
| ||||||
| 5. | Rollout2 | 13/10* 8/7 | eq: +0.262 (-0.135) | |||
| ||||||
| 6. | Rollout2 | 21/20 13/10* | eq: +0.256 (-0.140) | |||
| ||||||
| 1 5184 Games rolled with Variance Reduction. Dice Seed: 90634172 Moves: 3-ply, cube decisions: XG Roller 2 5184 Games rolled with Variance Reduction. Dice Seed: 20067038 Moves: 3-ply, cube decisions: XG Roller | ||||||
 |
| ||||
| XGID=----B-DB--aAdD---e-e-B----:0:0:1:13:0:0:3:0:10 | |||||
42P-63S-31X-21@-63Z-64R-32S-52T-63R-62R-31 | |||||
| 1. | Rollout1 | 13/10* 11/10 | eq: +0.430 | |||
| ||||||
| 2. | Rollout1 | 13/10* 6/5 | eq: +0.423 (-0.007) | |||
| ||||||
| 3. | Rollout2 | 13/10* 10/9 | eq: +0.415 (-0.015) | |||
| ||||||
| 1 62208 Games rolled with Variance Reduction. Dice Seed: 92924208 Moves: 3-ply, cube decisions: XG Roller 2 20736 Games rolled with Variance Reduction. Dice Seed: 92924208 Moves: 3-ply, cube decisions: XG Roller | ||||||
If you widen your window enough, you can view the above two positions side by side. As a pair, they originally appeared here. Before that, the first position was also posed here, with most of the responses under that subthread. Henceforward, I'll refer to them as the "8pt position" and the "7pt position" (by the different point that Blue owns), respectively.
The results of the top three moves are:
........8pt [$ P30 938] "<=5
........7pt [P $7 915] "<=62*20
In the 8pt position, $ (Slot, 13/10* 6/5) beats P (Point, 13/10* 11/10) by .030. In the 7pt position, P beats $ by .007, a swing of .037. Wow.
Much of the reason that $ is such a strong candidate is that White's 53 and 51 (which nobody mentioned) will jar a second blot loose. It is important to be able to (and remember to) spot such rolls.
The 53 effect exists only in the 8pt position. In the 7pt position, White is able to play R (Run, bar/20*/17) for an absolute equity of -.007 (according to XGR++ eval), instead of having to settle for S (Split-and-down, bar/20* 13/10) for an equity of -.467. Multiplying the .460 difference by 1/18 (the chances of rolling 53) gives us .026. (This is not a rigorous measurement of the 53 effect, but it's close enough.) That one roll explains the lion's share of the $/P swing of .037 (for Blue's 31) that we're seeing.
[As to the question posed of whether the decision in the 8pt position is close, it's right on the .03 cusp, according to the BW error scale! If anyone wants to extend the 5k rollout, let me know. :)]
With all this talk of QF, I get the impression that some respondents think I'm trickier than I really am. I'm just an innocuous position poster like anyone else, innocent as a lamb.
Lest you start hitting and slotting willy-nilly after seeing the result of the first rollout above, it is good to be aware that factors need to align to support such a move. Not only should you look for and assess the impact of root/swing number(s), but be sure to consider the race (and/or timing).
Below, White's pipcount has been altered -- her third spare is back on the midpoint instead of on the 8pt, so that Blue, after playing his move, will be up 2 pips instead of down 3 pips, as compared to the "8pt" (first) position. The result of this 5-pip adjustment is that $ loses its edge of .030 and instead P wins by .015, a swing of .045.
Nack
  |
| ||||
| XGID=----b-E-D---dEaA-b-d-b----:0:0:-1:31:0:0:3:0:10 | |||||
32S-65R-21@-41T-42P-63R-31 | |||||
| 1. | Rollout1 | 13/10* 11/10 | eq: +0.436 | |||
| ||||||
| 2. | Rollout1 | 13/10* 6/5 | eq: +0.422 (-0.015) | |||
| ||||||
| 3. | Rollout1 | 13/10* 10/9 | eq: +0.417 (-0.019) | |||
| ||||||
| 1 15552 Games rolled with Variance Reduction. Dice Seed: 92861368 Moves: 3-ply, cube decisions: XG Roller | ||||||