FIBS Board backgammon forum

 

1. Cubeful 2-ply 24/17      Eq.: +0.362
  68.1% 17.4% 0.7% - 31.9% 7.4% 0.2%  2-ply cubeful prune [world class]
2. Cubeful 2-ply 24/23 17/11*    Eq.: +0.343 ( -0.019)
  67.1% 24.8% 1.5% - 32.9% 11.5% 0.4%
  2-ply cubeful prune [world class]
3. Cubeful 2-ply 17/11*/10*    Eq.: +0.314 ( -0.048)
  65.7% 26.6% 1.6% - 34.3% 10.8% 0.4%
  2-ply cubeful prune [world class]
4. Cubeful 0-ply 17/11* 13/12    Eq.: +0.161 ( -0.200)
  58.1% 22.9% 1.2% - 41.9% 14.8% 0.8%
  0-ply cubeful prune [expert]
5. Cubeful 0-ply 17/11* 8/7    Eq.: +0.140 ( -0.222)
  57.0% 24.0% 1.3% - 43.0% 17.0% 0.9%
  0-ply cubeful prune [expert]

A nice position with a range of strategies to consider. Red can hit twice, hit once and move to the edge of the prime or even eschew the hit entirely and make good his escape. Any other plays are blunders. What to do? It's a one point match so winning or losing a gammon doesn't need to be considered, only what will win the game.
The "escape now" plan has instant appeal, smartly dealing with a major hurdle on Red's long march to victory. It also has the drawback that Red still trails in the race by 14 pips after the roll and the race may yet be a factor. This isn't crucial yet but not hitting does mean that White has a lot of freedom to play. If White can enter quickly, Red may find himself having to hit loose, as Red's checkers after this play are not well placed to complete the vital 4pt.
Hitting twice deals with the race problem for now but leaves Red needing a 1 and then a 6 at some future point. Even that may not be enough if White can turn his 5 prime into a sixer. I think we can definitely reject the double hit, but the single hit and move up is very attractive. Two men back for White may give Red the time to hit a third checker and arrange his builders to complete a prime without having to hit loose. Of course Red will still have to escape, but he should get time to throw a six before the end. That's the play that gets my vote.
Unfortunately pressing the spoiler button reveals.............gibberish! A little editing needed there one feels but careful examination reveals that the gnu printout, for that is what it is, prefers 24/17. As we can also deduce that it isn't a rollout, it is evident that yet again we have been served evaluation as analysis. Bot evaluation is always interesting and of course often correct, but gospel it 'aint. Rollouts show us that in fact 24/23, 17/11* is best, winning this game about 68% of the time. 24/17 is second winning about 66.5% and the double hit is a distant third.

Mine is the vote for 17/11*/10*, which is wrong. I somehow didn't see the white checker on the bar. Without it the double-tap is the correct play.

With the position viewed correctly, my choice is 23/22 17/11* as that results in two on the bar and buys time to get the 6 to complete the backrunner's escape. Doing the double-tap, 17/11*/10*, is an unnecessary extravagence given the blot already on the bar. Running, 24/17, without an further hit is too dangerous given the weakness of the home table which gives White too much opportunity to either get a hit or regroup.

A rollout (2-ply for 12 plies and 0-ply thereafter, 2-ply cube throughtout) gives:

    1. Rollout          24/23 17/11*      +0.361
    2. Rollout          24/17                 +0.342 (-0.019)
    3. Rollout          17/11*/10*         +0.316 (-0.045)


Details:
    1. Rollout          24/23 17/11*                 Eq.:  +0.361
        68.0%  31.2%   4.8% -  32.0%  13.0%   1.7% CL  +0.361 CF  +0.361

    2. Rollout          24/17                        Eq.:  +0.342 ( -0.019)
        67.1%  22.0%   2.8% -  32.9%  10.3%   0.5% CL  +0.342 CF  +0.342

    3. Rollout          17/11*/10*                   Eq.:  +0.316 ( -0.045)
        65.8%  35.7%   6.3% -  34.2%  11.3%   2.0% CL  +0.316 CF  +0.316

      [  0.2%   0.7%   0.4% -   0.2%   0.4%   0.2% CL   0.004 CF   0.004]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 813063424 and quasi-random dice
        Play: world class 2-ply cubeful prune [world class]
        keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
        Skip pruning for 1-ply moves.
        Cube: 2-ply cubeful prune [world class]
        Different evaluations after 12 plies:
        Play: 0-ply cubeful prune [expert]
        Cube: 2-ply cubeful prune [world class]