Spoiler
With our three point board vs his anemic one-point board we're going to go for an exchange of hits if we can. If we can send a second checker back just about any roll that doesn't allow him to anchor is a market loser. I count 22/36 hitting rolls, and only 14/36 rolls for him that anchor. That's a lot of market losing sequences.
But are we good enough? We're behind in the race, but we have a pretty good blockade and an advanced anchor. Overall, our position is quite superior. With the 61% chance of hitting we have a clear advantage in threats. That's an advantage in both position and threats, and using the PRAT (http://www.bkgm.com/rgb/rgb.cgi?view+702) criteria that should be enough to double.
Moreover,since we're behind in the match score, we should be more willing to push the cube than usual, so I say ship it.
double.
im gonna out to the movies. If this continues this way ill move this along.
roadkillbooks
Ok its 5 am fibsboard time. Not a lot of votes but 3-1 is 75% which is most definately a cubefriendly percentage so....
Blitzxz do you take or drop?
Quote from: diane on February 05, 2010, 08:16:05 AM
I didnt get a chance to vote again :(
Right. We're only allowed to vote once. Nobody is allowed to vote
again once they've voted.
For what it takes , after 11 hours after the poll was opened and I found it locked even after my foreposters noticed and complained, I'd say cube.
Please don't ask me to elaborate why. Trust me on this :thumbsup2:
Quote from: ah_clem on February 05, 2010, 05:48:08 PM
Right. We're only allowed to vote once. Nobody is allowed to vote again once they've voted.
:tongue2: :tongue2: :tongue2: :tongue2:
:lol:
There's some difficulty here as I see lots of outcomes where I lose a gammon and gammons have extra value here. However, I lead the race and the game is far from over which gives me very good winning chances. Forum's 4 checkers behind might also cause some troubles to close me out as there might not be enough ammo. I'm not going to give any more free points for Forum so after some thought I take.
gnubg rollout
Spoiler
It's a marginal double. And a huge take.
Cube analysis
Rollout cubeless equity +0.372 (Money: +0.287)
Cubeful equities:
1. Double, take +0.625
2. Double, pass +1.000 ( +0.375)
3. No double +0.593 ( -0.032)
Proper cube action: Double, take
Rollout details:
Centered 1-cube:
0.581 0.211 0.021 - 0.419 0.100 0.008 CL +0.372 CF +0.593
[0.002 0.002 0.002 - 0.002 0.002 0.001 CL 0.007 CF 0.017]
Player blitzxz owns 2-cube:
0.585 0.207 0.028 - 0.415 0.115 0.013 CL +0.868 CF +0.625
[0.002 0.002 0.003 - 0.002 0.003 0.002 CL 0.021 CF 0.023]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 843463219 and quasi-random
dice
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]
Quote from: roadkillbooks on February 10, 2010, 08:10:04 PM
The only real explanation is that the ME tables are prob different but I think that is a large difference
I don't know what is extremeGammon's match equity table, but I'm pretty sure that match equity table can't make any kind of difference, unless you use asymmetric tables. All new match equity tables are so close to each other. The explanation might be that gnu 0-ply has problems with the cube. 0/2 gnu rollout agrees more with your numbers but there' still a difference. Although the standard errors are quite big, so 1296 rollout is not enough to judge what are the exact numbers. So variance in rollouts and different cube handling in gnu and XG are propably the reasons for difference. Gnu also manages to win little bit more gammons, which I think is odd.
Rollout details:
Centered 1-cube:
0,583 0,214 0,017 - 0,417 0,111 0,009 CL +0,366 CF +0,631
[0,002 0,002 0,002 - 0,002 0,002 0,001 CL 0,007 CF 0,014]
Player blitzxz owns 2-cube:
0,586 0,210 0,023 - 0,414 0,124 0,012 CL +0,918 CF +0,672
[0,002 0,002 0,002 - 0,002 0,003 0,002 CL 0,018 CF 0,020]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 623956183 and quasi-random dice
Play: 0-ply cubeful [expert]
Cube: 2-ply cubeful prune [world class]