Game 2 Move 4 Zorba Cube decision

[ah_clem]

Dr. Z, would you like to double here?

Position ID: hteMAyDYnoUBUA
Match ID: MAGgAAAACAAA

[Zorba]

13 rolls could hit that blot, if I counted correctly. If the Herd enters, it looks like a clear take and maybe not even a double. The Herd dances 25% of the time and it might be a pass after that, although I'd guess not a big pass. So there could be some small marketlosers here, 13/36 * 9/36 = 117/1296 ~= 9% of the time.

If I don't hit, I don't have much of an advantage, if at all. I also dance ~11% of the time myself after which I'll be a clear underdog.

So, no cube here. If things go well I'll have a strong cube next turn.

[NIHILIST]

I think you may have undervalued your position. You say that 13 numbers hit THAT blot, by which I assume you mean the blot on our bar point.

What about 5-4, 5-5 and 4-4 that bear on the OTHER blot ? While they miss the blot on the bar point they seem equally, if not more ( 5-5 ) devastating on the other side of the board. That gets the hitting numbers to 17, near even money for a position where gammon possibilities favor you pretty strongly.

I think if I were in your chair I'd whip it.


Bob

[NIHILIST]

OOPS ! Add 1-5 and 3-5 to the hitting side. That brings the hitting numbers to 21.

Bob

[ah_clem]

I think if I were in your chair I'd whip it.

A visual cue might be more persuasive.

[Zorba]

I'd love to play backgammon for money against these people!

[ah_clem]

gnubg results

Spoiler

2-ply analysis says no double (.06).  0-ply rollout says marginal double (.005).  2-ply rollout says no double (.02). Massive take regardless.

This is a close one. 





Cube analysis
2-ply cubeless equity  +0.249 (Money:  +0.283)
  0.589 0.235 0.011 - 0.411 0.134 0.006
Cubeful equities:
1. No double            +0.420
2. Double, pass         +1.000  ( +0.580)
3. Double, take         +0.359  ( -0.061)
Proper cube action: No double, take (9.5%)



Rollout details:
Centered 1-cube:
  0.590 0.240 0.012 - 0.410 0.137 0.015 CL  +0.299 CF  +0.427
[0.002 0.002 0.001 - 0.002 0.002 0.002 CL   0.006 CF   0.013]
Player The Herd owns 2-cube:
  0.595 0.244 0.013 - 0.405 0.147 0.021 CL  +0.694 CF  +0.433
[0.002 0.003 0.001 - 0.002 0.003 0.003 CL   0.016 CF   0.019]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 904334444 and quasi-random

dice
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]

1. No double            +0.427  ( -0.005)
2. Double, pass         +1.000  ( +0.567)
3. Double, take         +0.433 
Proper cube action: double, take



Rollout details:
Centered 1-cube:
  0.593 0.242 0.012 - 0.407 0.135 0.019 CL  +0.305 CF  +0.452
[0.001 0.002 0.001 - 0.001 0.002 0.002 CL   0.005 CF   0.011]
Player You owns 2-cube:
  0.600 0.249 0.010 - 0.400 0.142 0.027 CL  +0.701 CF  +0.433
[0.002 0.003 0.001 - 0.002 0.003 0.003 CL   0.013 CF   0.016]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 898960946 and quasi-random

dice
Play:  2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.1
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]


1. No double            +0.452 
2. Double, pass         +1.000  ( +0.548)
3. Double, take         +0.433  ( -0.020)
Proper cube action: no double, take

[close]

[Zorba]

@NIHILIST

I didn't mention the loose hits on the acepoint blot, as they are not nearly as strong as hitting the barpoint blot IMO. Sending a second checker back gains me a lot more than just putting that last runner on the bar, and not just in the race: it will also be much harder to escape my blockade with two men instead of one. Besides, hitting loose on the acepoint means leaving a direct shot there that will be hit ~30% of the time, so it really is a mixed blessing. The blot on the acepoint can also be a liability later on.

5-5 is a great roll though of course. Also, the loose hits have good potential when the Herd dances (25%), because of the other blots. But the bottom line is that with the loose hit, I get return hit more often than I get a dance.

XGID=-a--BBD-BA--cB-b-bad---bAA:0:0:1:00:0:1:0:5:10

XG results:

Spoiler

It looks like a pretty clear No Double with XG, I'll do a better rollout later to get a good figure on the margin here.

Analyzed in 3 ply
Player Winning Chances:    58,60% (G: 23,19% B:  1,20%)
Opponent Winning Chances:  41,40% (G: 12,63% B:  0,65%)

Cubeless Equities: No Double=+0,240, Double=+0,559

Cubeful Equities:
       No Double:   +0,407
       Double/Take: +0,312 (-0,094)
       Double/Drop: +1,000 (+0,593)

Best Cube action: No Double / Take

Analyzed in 4 ply
Player Winning Chances:    59,61% (G: 23,52% B:  1,14%)
Opponent Winning Chances:  40,39% (G: 12,22% B:  0,61%)

Cubeless Equities: No Double=+0,268, Double=+0,625

Cubeful Equities:
       No Double:   +0,431
       Double/Take: +0,387 (-0,044)
       Double/Drop: +1,000 (+0,569)

Best Cube action: No Double / Take

Analyzed in XG Roller+
Player Winning Chances:    59,19% (G: 22,47% B:  1,05%)
Opponent Winning Chances:  40,81% (G: 11,85% B:  0,67%)

Cubeless Equities: No Double=+0,253, Double=+0,551

Cubeful Equities:
       No Double:   +0,422
       Double/Take: +0,333 (-0,090)
       Double/Drop: +1,000 (+0,578)

Best Cube action: No Double / Take

Analyzed in Rollout
No Double
  Player Winning Chances:    59,24% (G: 23,46% B:  1,23%)
  Opponent Winning Chances:  40,76% (G: 12,68% B:  1,26%)
Double/Take
  Player Winning Chances:    59,43% (G: 24,00% B:  1,01%)
  Opponent Winning Chances:  40,57% (G: 12,63% B:  1,88%)

Cubeless Equities: No Double=+0,247, Double=+0,612

Cubeful Equities:
       No Double:   +0,443
       Double/Take: +0,386 (-0,057)
       Double/Drop: +1,000 (+0,557)

Best Cube action: No Double / Take

Rollout
  2592 Games rolled with Variance Reduction.
  Moves: 1 ply, cube decisions: 3 ply
  Confidence No Double: ± 0,020 (+0,423<E<+0,463)
  Confidence Double:    ± 0,027 (+0,359<E<+0,414)
 
  Double Decision confidence: 99,9%
  Take Decision confidence: 100,0%
 
  Duration: 19 minutes 13 seconds

eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross

[close]

[Zorba]

Long XG 3-ply rollout:

Spoiler

Pretty close to a double, but not quite there (0.04 error to cube). Winning chances are not even 60%. The high gammon rate (~40% relative) plus very high volatility still make it almost a double though.

XGID=-a--BBD-BA--cB-b-bad---bAA:0:0:1:00:0:1:0:5:10
Analyzed in Rollout
No Double
  Player Winning Chances:    59,12% (G: 23,95% B:  1,19%)
  Opponent Winning Chances:  40,88% (G: 12,89% B:  1,39%)
Double/Take
  Player Winning Chances:    59,70% (G: 24,90% B:  1,05%)
  Opponent Winning Chances:  40,30% (G: 13,65% B:  2,17%)

Cubeless Equities: No Double=+0,244, Double=+0,618

Cubeful Equities:
       No Double:   +0,436
       Double/Take: +0,396 (-0,040)
       Double/Drop: +1,000 (+0,564)

Best Cube action: No Double / Take

Rollout
  10368 Games rolled with Variance Reduction.
  Moves and cube decisions: 3 ply
  Search interval: Large
  Confidence No Double: ± 0,008 (+0,428<E<+0,444)
  Confidence Double:    ± 0,012 (+0,384<E<+0,408)
 
  Double Decision confidence: 100,0%
  Take Decision confidence: 100,0%
 
  Duration: 13 hours 14 minutes

eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross

[close]