BGonline.org Forums

Snowie’s fake AtS data

Posted By: Taper_Mike
Date: Monday, 22 April 2013, at 11:00 a.m.

In Response To: Snowie’s fake AtS data (Maik Stiebler)

Timothy: Differences in ranking of plays between a Snowie AtS rollout and a Snowie money rollout when we know that Snowie plays the checkers exactly the same way in both cases seem implausible. I am ignorant of the details of Snowie’s AtS rollout procedure. If you and Maik are as well, then truly this is a case of the blind leading the blind. This began with my speculations regarding the book Understanding Backgammon. I guess there is no reason to stop hypothesizing now. Here is my surmise explaining why Snowie can switch the ranking of plays when it applies a formula to the equities of one of its money rollouts in order to create “fake” AtS data. It is worth every nickel you paid for it. It is really quite simple. Snowie uses the percentages is generates for single wins, gammon wins, backgammon wins, gammon losses, and backgammon losses, in conjunction with a MET, to determine the MWC for each play in a rollout. The ranking of plays according to MWC need not be the same as the ranking of plays in unlimited games. That means, for instance, that a play that nets a high percentate of gammon wins should rise in equity against a play that does not at any match score where the gammon value is elevated. Conversely, at DMP, where gammons are useless, the rate of gammons wins is irrelevant. In unlimited games, of course, gammon wins are highly relevant. The difference between Snowie, on the one hand, and GnuBg and XG, on the other, is that Snowie performs its calculations of MWC only once, at the the end of a money rollout. GnuBg and XG do theirs for every move considered during a rollout. Here are a couple of examples where Snowie does reorder the plays at score:

43D-44

43D-44	[b R1 B2] ~31	Nack Ballard
43D-44	d[R b17 B18] ~31	Nack Ballard

Key: B (Both) = 24/16*, 13/9(2), b (both) = 24/20(2), 13/9(2), R (Run) = 24/16*(2)

43S-61

43S-61	[X P5 H14] ~5 — Cubeful; not Live Cube	BgOnline
43S-61	[X P9 H23] ~31	Nack Ballard
43S-61	s[P X25 H32] ~31	Nack Ballard

Key: H (Hit) = 13/7, 6/5*, P (Point) = 13/7, 8/7, X (Hit and split) = 24/18, 6/5*

Explanation of nacbracs

Mike

Messages In This Thread

• A familiar problem?
  David Levy -- Tuesday, 16 April 2013, at 8:49 p.m.
  • A familiar problem?
    Taper_Mike -- Tuesday, 16 April 2013, at 10:10 p.m.
    • A familiar problem?
      Timothy Chow -- Wednesday, 17 April 2013, at 12:19 a.m.
      • A familiar problem?
        Dmitriy Obukhov -- Wednesday, 17 April 2013, at 2:35 a.m.
        • Woolsey-Jones, Problem 80
          Taper_Mike -- Wednesday, 17 April 2013, at 4:06 a.m.
          • Woolsey-Jones, Problem 80
            Timothy Chow -- Wednesday, 17 April 2013, at 3:21 p.m.
          • Woolsey-Jones, Problem 80
            Dmitriy Obukhov -- Wednesday, 17 April 2013, at 3:34 p.m.
          • Woolsey-Jones, Problem 80
            Stick -- Wednesday, 17 April 2013, at 5:20 p.m.
          • Some variants (bot content)
            Timothy Chow -- Thursday, 18 April 2013, at 2:40 p.m.
          • Woolsey-Jones, Problem 80 — 3-ply XG2 Rollout
            Taper_Mike -- Friday, 19 April 2013, at 2:45 a.m.
            • Woolsey-Jones, Problem 80 — 3-ply XG2 Rollout
              Igor -- Friday, 19 April 2013, at 3:00 p.m.
              • Woolsey-Jones, Problem 80 — 3-ply XG2 Rollout
                Timothy Chow -- Saturday, 20 April 2013, at 1:47 a.m.
                • Woolsey-Jones, Problem 80 — 3-ply XG2 Rollout
                  Igor -- Saturday, 20 April 2013, at 4:36 a.m.
            • Woolsey-Jones, Problem 80 — 3-ply XG2 Rollout
              Timothy Chow -- Saturday, 20 April 2013, at 1:53 a.m.
              • Snowie’s fake AtS data
                Taper_Mike -- Saturday, 20 April 2013, at 6:22 a.m.
                • Snowie’s fake AtS data
                  Timothy Chow -- Sunday, 21 April 2013, at 7:33 p.m.
                  • Snowie’s fake AtS data
                    Maik Stiebler -- Monday, 22 April 2013, at 9:10 a.m.
                    • Snowie’s fake AtS data
                      Taper_Mike -- Monday, 22 April 2013, at 11:00 a.m.
                      • Snowie’s fake AtS data
                        Timothy Chow -- Wednesday, 24 April 2013, at 1:50 a.m.
                        • Snowie’s fake AtS data
                          Taper_Mike -- Wednesday, 24 April 2013, at 3:19 a.m.
                          • Snowie’s fake AtS data
                            Maik Stiebler -- Wednesday, 24 April 2013, at 8:19 a.m.
        • A familiar problem?
          Timothy Chow -- Wednesday, 17 April 2013, at 3:18 p.m.
  • A familiar problem?
    ah_clem -- Wednesday, 17 April 2013, at 7:25 p.m.
  • A familiar problem? — XGR++ Eval
    Taper_Mike -- Tuesday, 23 April 2013, at 6:12 p.m.