                        |
| ||||
| XGID=bB-B--C-B---dD---c-d-b-AA-:0:0:1:00:0:0:3:0:10 | |||||
42P-51S-52S-55A-F-C [D 1000–212] "<=164 | |||||
| Analyzed in Rollout | No double | Double/Take |
| Player Winning Chances: | 65.045% (G:39.768% B:0.440%) | 65.555% (G:42.074% B:0.474%) |
| Opponent Winning Chances: | 34.955% (G:9.066% B:0.521%) | 34.445% (G:9.550% B:0.600%) |
| Cubeless Equities | +0.6071 | +1.2701 |
| Cubeful Equities | ||
| No double: | +0.7882 (-0.2118) | ±0.0011 (+0.7871..+0.7894) |
| Double/Take: | +1.0000 | ±0.0016 (+0.9984..+1.0016) |
| Double/Pass: | +1.0000 (+0.0000) | |
| Best Cube action: Double / Take | ||
| Rollout details | ||
| 164600 Games rolled with Variance Reduction. Dice Seed: 85570326 Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 100.0% | |
| Take Decision confidence: | 50.0% | |
| Duration: 1 day 07 hours 37 minutes | ||
The nearest multiple of 1296 that is less than 164,600 is 164,592, a difference of only 8. A suspicious reader such as myself might suppose that Nack began with the exact multiple, and then added just enough trials to turn the take/pass decision into a perfect tie.
I should have known you'd be paying attention. :)
Actually, once I got close to four zeros, I added two trials at a time, in order to get an even number (making it less likely to be instantly busted as a non-1296-multiple).
I may well have missed something earlier, but from the time that I started watching closely, the first quad-zero that was also 50.0% (i.e., not 50.1 or 50.2) came at 161786, and was also there at 161794, 161796 and 161798. It came back to quad-zero 50.0% after a few hundred and I almost settled for the repeater-number 162222, but subsequently, by trying only multiples of 100 and 1296, eventually found 164600. (The 1296-multiple of 164592 was a quad-zero but at the imperfect 50.1%, as I recall.)
I never did find a 1296-multiple that coincided with quad-zero 50.0%. Then again, I never expected to get as lucky as I did.
There's always different seed numbers, if you want to try. I'm sure that ultimately you could even find a workable multiple of 12960. But I suspect you can make better use of your CPU time. :)
Nack