BGonline.org Forums

32S-43 data (by rollout)

Posted By: Nack Ballard
Date: Tuesday, 21 December 2010, at 1:48 a.m.

In Response To: 32S-43 RO, 17k results (Alexandre Le Siege)

Here is a summary of 32S-43 data by bot:

[H X9 K30] ~31*5
[H X15 K53] /41*5
[H X8] <62
[H X12 K30] ^17*5

Only one XG 3-ply rollout (third line) and one XG 4-ply rollout (fourth line) of this position has been done.

Isolating just the H and X plays, it may be worth noting that the Snowie composite (first line of the four) is composed of these two rollout results:

[X H1] 7776 trials
[H X12] 23328 trials

...and the GnuBG composite (second line of the four) is composed of these two results:

[H X27] 10368 trials
[H X11] 31104 trials

Lesson: Variance is higher than is widely believed! When in doubt, keep rolling it out (add more trials).

Nack

Messages In This Thread

• 32S-43 RO, 17k results
  Alexandre Le Siege -- Monday, 20 December 2010, at 7:33 p.m.
  • 32S-43, 32S-32, 43Z-43 and 43Z-32
    Nack Ballard -- Tuesday, 21 December 2010, at 12:54 a.m.
  • 41S-52 next, noted, thx *NM*
    Nack Ballard -- Tuesday, 21 December 2010, at 12:56 a.m.
  • 32S-43 data (by rollout)
    Nack Ballard -- Tuesday, 21 December 2010, at 1:48 a.m.