What am I missing here? - Backgammon Forum - Discuss Backgammon Strategy, Clubs and Books

Two Plus Two Forums Sports and Games Chess and Other Board Games Backgammon Forum hosted by Bill Robertie.

What am I missing here?

Post Reply Subscribe

...

12-05-2009 , 07:25 AM

municipalis

centurion

Join Date: Oct 2009 Posts: 191

I'm a fairly new player, so maybe I'm just missing something really obvious, but I played a match and had it analyzed in gnubg, and this move was flagged:

8/4 8/6 seemed to be the obvious move to me, since it didn't leave any blots and got me ready to bear-off.

Apparently though, I'm wrong, and the best move (by a smidgen) is 8/4 7/5. Why? With that move you're fairly likely to get hit next round as you leave two blots, and all you gain is 1 pip in the race... How exactly are you gaining with the more 'risky' 8/4 7/5?

First I ran it through 4-ply (Win/WinG/WinBg/Lose/LoseG/LoseBg)

Code:

    1. Cubeful 4-ply    8/4 7/5                      Eq.:  +1.296
       0.783 0.594 0.219 - 0.217 0.010 0.000
        4-ply cubeful prune
    2. Cubeful 4-ply    8/6 8/4                      Eq.:  +1.293 ( -0.002)
       0.777 0.594 0.229 - 0.223 0.008 0.000
        4-ply cubeful prune

And still not convinced, I ran a full rollout

Code:

    1. Rollout          8/4 7/5                      Eq.:  +1.239
       0.762 0.596 0.251 - 0.238 0.003 0.000 CL  +1.369 CF  +1.239
      [0.010 0.011 0.009 - 0.010 0.001 0.000 CL   0.033 CF   0.038]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 2833507936 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    2. Rollout          8/6 8/4                      Eq.:  +1.202 ( -0.037)
       0.744 0.587 0.266 - 0.256 0.001 0.000 CL  +1.339 CF  +1.202
      [0.009 0.010 0.009 - 0.009 0.000 0.000 CL   0.032 CF   0.037]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 2833507936 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

Quote

12-05-2009 , 11:50 PM

manticore

centurion

Join Date: Dec 2008 Posts: 155

-you wanna clear points now.before white will free some checkers.
-you don`t want the extra checker on the 6point,you prefer bringing your checkers straight to the 4 (5) point.

Quote

12-06-2009 , 10:30 AM

Robertie

old hand

Join Date: May 2004 Posts: 1,954

Your play, 8/4 8/6, is obviously correct. There's no advantage here in leaving a double shot and in addition giving White 4s to start escaping checkers.

The problem is with GNU, not you. A Snowie rollout showed 8/4 8/6 better by a sizeable amount. Not all bots handle these massive backgame situations equally well.

Quote

12-06-2009 , 06:20 PM

municipalis

centurion

Join Date: Oct 2009 Posts: 191

Quote:

Originally Posted by Robertie

Thanks for taking the time to run it through Snowie. Manticore's explanation had crossed my mind as a possibility, but I was still skeptical that the gain in equity from having one less checker on the 6-point counteracted the huge chance of being hit. Good to know that I should sometimes trust my instincts.

Quote

12-06-2009 , 08:22 PM

Mr. Barbegris

old hand

Join Date: May 2009 Posts: 1,628

fwiw, xg rollsout 8-4 6-4 (+1.207) over 8-4 7-5 (+1.163).

Quote

12-07-2009 , 07:06 PM

atrifix

grinder

Join Date: Aug 2004 Posts: 500

GNU, Snowie, XG sometimes (often) do very weird things in these positions. A 0-ply rollout is basically worthless.

Not sure whether GNU thinks you are hurting his timing by getting hit or it is trying to pick up the checker on the deuce point but in either case 8/4 8/6 seems clearly right.

Quote

Post Reply Subscribe

...