BGonline.org Forums

Rollout strength - A mathematical model

Posted By: Maik Stiebler
Date: Sunday, 19 February 2012, at 10:10 a.m.

In Response To: Rollout strength - A mathematical model (eXtreme Gammon)

Thanks again. You have convinced me that XGR++ makes 1-ply rollouts obsolete.

I think with 2-ply rollouts it's a different story, as your table shows that their systematic error is much smaller than that of XGR++. I just have to defend 2-ply rollouts, because they are my standard setting (with 2592 trials). In practice, I think the noise is not that much of a problem, as I can easily reject the result from the first 2592 trials, when the bot tells me that there might be a noise issue.

I will continue to use 2-ply rollouts, though your table - showing the great performance of XGR++ once more - reminds me of experimenting with stepped settings with higher ply for the first few moves.

Messages In This Thread

• Rollout strength - Your thoughts on a mathematical model
  eXtreme Gammon -- Friday, 10 February 2012, at 1:41 p.m.
  • Rollout strength - Your thoughts on a mathematical model
    Bob Koca -- Friday, 10 February 2012, at 2:28 p.m.
    • Rollout strength - Your thoughts on a mathematical model
      eXtreme Gammon -- Friday, 10 February 2012, at 2:55 p.m.
  • Rollout strength - A mathematical model
    eXtreme Gammon -- Friday, 17 February 2012, at 3:33 p.m.
    • Rollout strength - A mathematical model
      Maik Stiebler -- Saturday, 18 February 2012, at 10:03 a.m.
      • Rollout strength - A mathematical model
        eXtreme Gammon -- Saturday, 18 February 2012, at 4:53 p.m.
        • Rollout strength - A mathematical model
          Maik Stiebler -- Sunday, 19 February 2012, at 10:10 a.m.