Not a contrived position, this actually came up today. ND? D/T? D/P? TG?
XGID=-babdaDBB-A--B---aAd-AAA--:0:0:1:00:0:0:0:5:10
X:X O:O
Score is X:0 O:0 5 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O X | | O X X X |
| X | | O |
| | | O |
| | | O |
| | | |
| |BAR| |
| | | |
| | | X O |
| | | X O |
| X X | | X O O O |
| X X X | | X O O O O O |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 174 O: 251 X-O: 0-0/5
Cube: 1
X on roll, cube action
Confidence No Double: ± 0.039 (+0.713..+0.791)
Confidence Double: ± 0.057 (+0.721..+0.834)
On 9/11/2023 3:01 PM, ah....Clem wrote:
Confidence No Double: ± 0.039 (+0.713..+0.791)
Confidence Double: ± 0.057 (+0.721..+0.834)
At minimum, you should roll out enough trials so that the
error bars aren't so large that you can drive a truck through
them.
On 9/12/2023 9:25 AM, Timothy Chow wrote:
On 9/11/2023 3:01 PM, ah....Clem wrote:
Confidence No Double: ± 0.039 (+0.713..+0.791)
Confidence Double: ± 0.057 (+0.721..+0.834)
At minimum, you should roll out enough trials so that the
error bars aren't so large that you can drive a truck through
them.
Do you think throwing more cycles at it will produce meaningful results?
I don't.
My take is that if the evaluation is wrong the analysis wrong and the
rollout is an exercise in crunching meaningless numbers. GIGO.
Not a contrived position, this actually
came up today. ND? D/T? D/P? TG? XGID=-babdaDBB-A--B---aAd-AAA--:0:0:1:00:0:0:0:5:10
XGID=-babdaDBB-A--B---aAd-AAA--:0:0:1:00:0:0:0:5:10
X:Player 1 O:Player 2
Score is X:0 O:0 5 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O X | | O X X X |
| X | | O |
| | | O |
| | | O |
| | | |
| |BAR| |
| | | |
| | | X O |
| | | X O |
| X X | | X O O O |
| X X X | | X O O O O O |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 174 O: 251 X-O: 0-0/5
Cube: 1
X on roll, cube action
As Tim no doubt correctly points out, this is not the kind of position
where the bot verdict can be relied on. But it's the best I've got, so
here it is.
Clearly, O is poised to lose a lot of gammons but also to turn it around
and have an efficient recube. Since I didn't understand which way was
up I held the cube here. XG thinks it's a narrow cube and a double
whopper to pass. Rollout, and the various lower settings basically agree
for what that's worth.
I'm not going to sweat over missing a cube here.
Analyzed in Rollout
No double
Player Winning Chances: 61.84% (G:36.65% B:3.11%)
Opponent Winning Chances: 38.16% (G:6.71% B:0.33%)
Double/Take
Player Winning Chances: 60.77% (G:35.80% B:3.74%)
Opponent Winning Chances: 39.23% (G:6.77% B:0.70%)
Cubeless Equities: No Double=+0.545, Double=+1.202
Cubeful Equities:
No double: +0.752 (-0.026)
Double/Take: +0.777
Double/Pass: +1.000 (+0.223)
Best Cube action: Double / Take
Rollout:
1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: ± 0.039 (+0.713..+0.791)
Confidence Double: ± 0.057 (+0.721..+0.834)
Double Decision confidence: 76.7%
Take Decision confidence: 100.0%
Duration: 34 minutes 03 seconds
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
--
Ah....Clem
The future is fun, the future is fair.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 297 |
Nodes: | 16 (2 / 14) |
Uptime: | 17:07:29 |
Calls: | 6,667 |
Calls today: | 1 |
Files: | 12,216 |
Messages: | 5,336,854 |