BGonline.org Forums

Which play is better based upon rollouts

Posted By: Daniel Murphy
Date: Monday, 26 April 2010, at 5:14 p.m.

In Response To: Which play is better based upon rollouts (Seth)

With an opening roll of 4-1 (Slot) 4-1$..opponent rolls 1-1 Which play is correct based upon these 2 rollouts?

[snip text copied from http://www.bgonline.org/ 2nd roll rollouts]

Based on the rollouts, the best play is pick'em between 24/20* and 8/7(2) 6/5(2).

The rollouts don't quite agree.

Gnubg gives an insignificant nod to making two points. The 95% confidence intervals are 0.004 and 0.005, relatively large in comparison with the difference between plays of 0.0035. In other words, pick'em.

Snowie has a more pronounced preference for 24/20* with a difference of 0.029 (live cube equity) and 95% confidence intervals of 0.014 and 0.013. If Snowie were the only bot, this result would indicate that 24/20* is best, according to Snowie. But the confidence intervals could be smaller, and it would be nice if the rollout were extended, especially in light of the differing result from the Gnubg rollout.

I think one frequent commentator has observed that Gnubg tends to overvalue splits, i.e, undervalues slots. (Neil? Anyone?) If that's so, then I would conclude that the rollouts together say there's a good chance that 24/20* is slightly better than 8/7(2) 6/5(2).

Rollouts summarized showing 95% confidence intervals:

	24/20*	8/7(2) 6/5(2)
Snowie	0.154 - 0.182	0.126 - 0.152
Gnubg	0.132 - 0.140	0.127 - 0.137

---

Stick, the 95% confidence intervals in the Snowie rollouts aren't displayed properly. I'm guessing that f.x. "live cube: 0.168 10.014" should read "live cube: 0.168 ±0.014".

Messages In This Thread

• Which play is better based upon rollouts
  Seth -- Monday, 26 April 2010, at 1:24 p.m.
  • Which play is better based upon rollouts
    Daniel Murphy -- Monday, 26 April 2010, at 5:14 p.m.
  • 41$-11
    neilkaz -- Monday, 26 April 2010, at 5:30 p.m.
  • Which play is better based upon rollouts
    Stick -- Monday, 26 April 2010, at 6:38 p.m.
    • Which play is better based upon rollouts
      David Rockwell -- Monday, 26 April 2010, at 7:40 p.m.
      • Which play is better based upon rollouts
        Michael Depreli -- Monday, 26 April 2010, at 8:53 p.m.
        • Which play is better based upon rollouts
          Timothy Chow -- Monday, 26 April 2010, at 9:41 p.m.
          • Which play is better based upon rollouts
            Matt Cohn-Geier -- Monday, 26 April 2010, at 9:47 p.m.
            • Which play is better based upon rollouts
              Michael Depreli -- Tuesday, 27 April 2010, at 12:39 p.m.
              • Which play is better based upon rollouts
                David Rockwell -- Tuesday, 27 April 2010, at 1:17 p.m.
                • Which play is better based upon rollouts
                  Timothy Chow -- Tuesday, 27 April 2010, at 3:14 p.m.
          • Which play is better based upon rollouts
            Daniel Murphy -- Monday, 26 April 2010, at 10:26 p.m.
            • Which play is better based upon rollouts
              Timothy Chow -- Monday, 26 April 2010, at 11:08 p.m.
              • Which play is better based upon rollouts
                Stick -- Monday, 26 April 2010, at 11:18 p.m.
                • Which play is better based upon rollouts
                  David Rockwell -- Tuesday, 27 April 2010, at 2:19 a.m.
                  • Which play is better based upon rollouts
                    Timothy Chow -- Tuesday, 27 April 2010, at 2:41 a.m.
        • Which play is better based upon rollouts
          David Rockwell -- Tuesday, 27 April 2010, at 2:22 a.m.
          • Which play is better based upon rollouts
            Michael Depreli -- Tuesday, 27 April 2010, at 8:47 a.m.
            • Which play is better based upon rollouts
              David Rockwell -- Tuesday, 27 April 2010, at 12:47 p.m.
  • 41$-11 data (to date)
    Nack Ballard -- Monday, 26 April 2010, at 7:41 p.m.
  • Thanks *NM*
    Seth -- Monday, 26 April 2010, at 8:07 p.m.