Re: [Bug-gnubg] Incorrect equities calculated

[Jim Segrave]

On Thu 07 Aug 2003 (15:16 -0400), Jim Blue wrote:
> In the saved position below, the equities at 2-ply and roll-out are 
> reasonable, in the neigherborhood of -.5, but at 3-ply they all come out to 
> be -1. (I set up this position by editing it, not in the course of a game, in 
> case that matters.)
> 
> My settings for both Analysis and Evaluation are World Class, 2 plies, no 
> reduction, cubeful, 0 noise, move filter normal.

That's a cubeless rollout and it says you might win 26% of the games
*if* they were played to completion. But the cubeful estimates seem to
be saying that you'll be doubled, you'll have to pass and you will
lose this game.

> ----------
> 
> GnuBG (O, 0 pts) vs. Jim (X, 0 pts) (Match to 3)
> 
> Game number 1
> 
> Match Information:
> 
> GnuBG's rating: n/a
> Jim's rating: n/a
> Date: August 07, 2003
> Event: n/a
> Round: n/a
> Place: n/a
> Annotator: n/a
> Comments: n/a
> Move number 4: X to play 15
> 
>     GNU Backgammon  Position ID: 7LwDBwC03YQBAw
>                     Match ID   : cIl0AAAAAAAA
>     +24-23-22-21-20-19------18-17-16-15-14-13-+  O: GnuBG
>     |       O  O  X  O |   | O              X |  0 points
>     |       O  O  X  O |   | O              X |  
>     |          O     O |   | O                |  
>     |                O |   |                  |  
>     |                  |   |                  |  
>     |                  |BAR|                  |v 3 point match (Cube: 1)
>     |                  |   |                  |  
>     |                  |   |                  |  
>     |                X |   |                O |  
>     |          X  X  X |   | X              O |  Rolled 15
>     |       X  X  X  X |   | X     X        O |  0 points
>     +-1--2--3--4--5--6-------7--8--9-10-11-12-+  X: Jim
> Pip counts: O 102, X 128
> 
> 
> 
> Cube analysis
> 2-ply cubeless equity  -0.488 (Money:  -0.480)
>   0.266 0.013 0.000 - 0.734 0.025 0.001
> Cubeful equities:
> 1. No double            -0.933
> 2. Double, pass         +1.000  ( +1.933)
> 3. Double, take         -2.372  ( -1.439)
> Proper cube action: No double, take (42.7%)
> 
> Rolled 15:     1. Rollout          7/6 7/2                      Eq.:  -0.532
>        0.240 0.011 0.000 - 0.760 0.018 0.001 CL  -0.532
>       [0.002 0.001 0.000 - 0.002 0.001 0.000 CL   0.005]
>         Truncated cubeless rollout (depth 11) (trunc. at one-sided bearoff) 
> with var.redn.
>         108 games, Mersenne Twister dice gen. with seed 1176041872 and 
> quasi-random dice
>         Play: 0-ply cubeful
>         Cube: 0-ply cubeful
>      2. Cubeful 3-ply    9/8 7/2                      Eq.:  -1.000 ( -0.468)
>        0.254 0.013 0.000 - 0.746 0.027 0.001
>      3. Cubeful 3-ply    7/2 6/5                      Eq.:  -1.000 ( -0.468)
>        0.254 0.014 0.000 - 0.746 0.027 0.001
>      4. Cubeful 3-ply    7/2 4/3                      Eq.:  -1.000 ( -0.468)
>        0.254 0.013 0.000 - 0.746 0.027 0.000
>      5. Cubeful 3-ply    7/2 3/2                      Eq.:  -1.000 ( -0.468)
>        0.254 0.013 0.000 - 0.746 0.027 0.001
> 
> 
> Output generated Thu Aug 07 15:06:42 2003
> by GNU Backgammon 0.14-devel 1.1245 030805 (Text Export version 1.50)
> 

> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-gnubg


-- 
Jim Segrave           address@hidden