BGonline.org Forums

Rollouts of Opening Moves

Posted By: Leonardo Jerkovic
Date: Friday, 18 December 2009, at 3:32 p.m.

I have XG, what is your opinion on settings for opening moves rollouts? How reliable are those rollouts considering opening moves?

If 3 or 4ply misplays some positions in openings (and yes, as I can see there are alot of close plays 3 or 4ply gets them wrong) can we trust those rollouts finally?

I remember Expert BG rollouts from 1994 reccomended opening 41 24/20 24/23! Expert BG played decent but not perfect, just like me! I don't play as good as XG 3ply entire match, so those rollouts, are they any useful to me as a player playing other not-as-good players? Are 1ply rollouts cubeless enough for me?

I ran 11.000 rollouts on XG 3ply using variance reduction algorithm, and still results are not whithin standard deviation (I started for opening 21 full 3ply rollouts which states splitting is better then slotting)...Are truncated rollouts more reliable? If so, what is the best trunc point? Is variance reduction algorithm OK, does it mess up rollouts results? Is Janowski formula that transforms cubeless into cubeful equities OK, does it mess up rollouts results?

How does XG incorporate cube actions during rollouts? Does it play cubeless to the end (bear-off) no matter when the cube is passed? Is there a way to get seperate cubeless vs cubeful equities?

Pls, your opinions

Messages In This Thread

• Rollouts of Opening Moves
  Leonardo Jerkovic -- Friday, 18 December 2009, at 3:32 p.m.
  • Rollouts of Opening Moves
    Stick -- Friday, 18 December 2009, at 4:00 p.m.
  • 21 in case you're curious
    neilkaz -- Friday, 18 December 2009, at 4:57 p.m.