Rollouts and evaluations

[puzzle_t0rico]

Hi there
One more example that why we should rollout a tricky position to be sure of the good play.
In the position below (double match point) black has to play 2-1


I wondered which is the good move: make the 4 point (6/4 5/4) or hit twice (10/9*/7*)

GnuBG says that the good move after analysis 3-ply is to hit

On the board I did not like to hit twice because I am leading the race (24 pips ahead), so if there are more white men behind, better for me but...
So, I launched a rollout (1296 games, without truncation) and then the analyse says that is better to make the 4 point.



Indeed, the difference is not even a 0.5% but the rollout confirmed my feeling on the board.
_________________
If the brain were simple enough to be understood, we would be too stupid to understand it.

[Simborg]

It is not possible to know which play is better without knowing the score of the match, or if it is a money game. If you are playing a match and gammons matter, it is clearly right to hit 2; if gammons don't matter, and you simply want to bring the game home safely, making the 4 point might be better.

For just about any position, the score must be taken into account when you do an evaluation or rollout.

[puzzle_t0rico]

Hi Phil
Thanks for your analyse. Indeed, I put the details of the match at the beginning but they are a bit hidden
It is a match game and it is double match point, so, gammons are completely useless and then building the 4 point comes to be the best move as you confirm with your comment.
Thanks again.
Hope one day I decide to enter in the Live Backgammon Tournaments
_________________
If the brain were simple enough to be understood, we would be too stupid to understand it.