BGonline.org Forums

62S,32?

Posted By: Nack Ballard
Date: Tuesday, 24 November 2009, at 4:13 p.m.

In Response To: 62S,32? (Dan Pelton)

I'm not sure how to interpret the results of Stick's rollouts here. One set of summaries prefers moving both back checkers up to 21 and 22, the other prefers slotting the 21 and 4. It looks as though I set of data was generated by gnuBg, the other set by Snowie. What am I missing here?

You are interpreting the posted rollouts correctly. (As Tim says, the bots just have a difference of opinion.) Snowie’s 5k rollout gives [U S12 W20] live cube, or [U S2 W5] cube-adjusted. GNU’s live cube rollout (similarly rounded) gives a noticeably different result of [W S1 U1 H22] – 46k for WSU and 5k for H.

[W = Wild 24/21 6/4, S = Split 24/21 13/11, U = Up 24/22 24/21, and H = Hit 6/1*. Error sizes are given in thousandths. For example “W S1 U1 H22” means that W is best, S is -.001, U is -.001 and H is -.022. Number of trials rolled out are listed in thousands: usually, for example, 5k = 5184, and 46k = 46656.]

For reference, I have a much longer (independent) live cube Snowie rollout that gives [U S11 W13 Z31 $33 T39 539 H40], 31k for the first three plays and 20k for the other five. [Z = reverse split 24/22 13/10, $ = Slot 13/10 6/4, T = sTack/Tower 13/8, and 5 = 13/11 8/5.]

Taking a straight bot average (not weighting for number of trials), Snowie's and GNU’s long live cube results for the top plays yields [U S5.4 W6.4] – all three are “tied” by my definition of being within 10 (i.e., .010) of each other.

I have no XG rollout of 62S-32 yet, but because you’ve shown interest I’ll gladly move it up on the queue.

Nack

Messages In This Thread

• 62S,32?
  Dan Pelton -- Tuesday, 24 November 2009, at 2:28 p.m.
  • 62S,32?
    Timothy Chow -- Tuesday, 24 November 2009, at 3:51 p.m.
  • 62S,32?
    Nack Ballard -- Tuesday, 24 November 2009, at 4:13 p.m.
    • 62S,32?
      Dan Pelton -- Tuesday, 24 November 2009, at 4:47 p.m.
      • 62S,32?
        Nack Ballard -- Tuesday, 24 November 2009, at 5:57 p.m.