BGonline.org Forums

GnuBG rollouts' slot bias

Posted By: Nack Ballard
Date: Monday, 3 May 2010, at 7:33 a.m.

In Response To: 65R-41 4 ply rollout (Bob Koca)

The long gnu RO's here have $ better by .0084 for 65R-41 and better by .0060 for 65R - 51. Is it reasonable to assume that the XG rollout will also make splitting relatively better by about .085 putting S in the lead for 51 reply? How about for 21 reply, will split also catch up by about .0085 (leaving $ still best by a decent amount)?

I have no XG rollout results for (the non-close) 65R-21, and none for 65R-51 other than those posted by Stick (3-ply) and Neil (4-ply). As for 65R-41, I recently posted all I have here.

I've combined the data for those three positions into this single post, and streamlined it to focus on just the S and $ plays for money:

65R-21:
Snowie .. [$ S22] 15k
GnuBG .. [$ 30.4] 5k

65R-51:
Snowie .. [S $3] 31k
GnuBG .. [$ S6] 31k
XG 3-ply [S $1] 15k
XG 4-ply [$=S] 46k

65R-41:
Snowie .. [$ S1] 20k
GnuBG .. [$ S8.4] 31k
XG 3-ply [S $9] 15k
XG 4-ply [S $1] 46k

Comparing to the rollout results of Snowie, Gnu’s results are slottier by 8.4, 9.0 and 9.4 (thousandths) for 65R-21, 65R-51 and 65R-41, respectively.

There is less data for XG/Gnu comparison. Gnu is slottier by 7.0 and 17.4 than XG 3-ply (relative margin inconsistency here could be due in part to variance in only 15k trials), and Gnu is slottier by 6.0 and 7.4 than 4-ply, for 65R-51 and 65R-41, respectively.

I’ve been harping on Gnu’s slot bias (in rollout results) for a long time. David Rockwell has done analysis supporting our theories; for example, he has demonstrated that Gnu misplays its replies to opening 21 by an average of .006+ against 21$ but by only .001 against 21S (whereas I believe Snowie’s errors were relatively small and balanced). There is probably more going on as well, given that Gnu is [$ S19] 92k; compare to Snowie [$ S8] 80k, XG 3-ply [$ S5.5] 140k, and XG 4-ply [$ S4] 24k.

Nack

Messages In This Thread

• 65R-41 4 ply rollout
  neilkaz -- Monday, 3 May 2010, at 2:58 a.m.
  • 65R-41 data
    Nack Ballard -- Monday, 3 May 2010, at 4:08 a.m.
    • 65R-41 — Gammon Save — 2-ply GnuBg Rollout
      Taper_Mike -- Friday, 14 September 2012, at 12:34 a.m.
  • 65R-41 4 ply rollout
    Bob Koca -- Monday, 3 May 2010, at 5:08 a.m.
    • 65R-41 4 ply rollout
      Stick -- Monday, 3 May 2010, at 5:19 a.m.
    • GnuBG rollouts' slot bias
      Nack Ballard -- Monday, 3 May 2010, at 7:33 a.m.
  • 65R-41$-41
    Paul Weaver -- Monday, 3 May 2010, at 3:36 p.m.
    • 65R-41$-41
      David Rockwell -- Monday, 3 May 2010, at 4:04 p.m.
      • 65R-41$-41
        Casper van der Tak -- Monday, 3 May 2010, at 4:14 p.m.
      • 65R-41$-41
        neilkaz -- Monday, 3 May 2010, at 7:40 p.m.
    • 65R-41$-41
      Nack Ballard -- Monday, 3 May 2010, at 7:18 p.m.
  • 65R-41 4 ply rollout
    Nack Ballard -- Monday, 3 May 2010, at 7:34 p.m.