[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

21$ 62 XG 4 ply RO so far

Posted By: Nack Ballard
Date: Tuesday, 19 January 2010, at 2:50 a.m.

In Response To: 21$ 62 XG 4 ply RO so far (Timothy Chow)

Tim: Nack, Stick's 46K rollout shows 62N in second place, only 0.0013 behind 62Z. Do you have a different 46K rollout that you're referring to?

Stick: Yes, he does. In short has a rollout where the top 3 plays have the same seed number. (actually he has a separate rollout with the N play rolled out again)

Nack: Elaborating on Stick's (correct) answer (he was the one that sent me the files), I show Gnu's equities for the 21$-62 plays are

Z -.2891 46k
N -.2941 46k, -.2905 46k
S -.2910 46k
$ - . . . . . . . . . . . . . . . . . . . . -.3077 5k
D - . . . . . . . . . -.3160 5k
R - . . . . . . . . . -.3214 5k

The top three plays on the left (ZNS) were rolled out with one seed, the plays in the middle column (NDR) were rolled with another, and the play on the right ($) was rolled with a third seed. To further the bewilderment, all plays are in the same file except the first N (left).

If one feels that same-seed should be honored exclusively, then value is [Z S2 N5] 46k for one set of plays, [N D25 R29] 46k for another, and $ is by itself. If one feels all seed equities should be weight-averaged equally, "value" is [Z S2 N3 $19 D27 R32] 46k 5k$DR.

Stick and I have witnessed higher variance occurring with mismatched seeds, though I think after 46k trials, most of that should be gone. So, my intent was to report an average.

I posted Gnu value as [Z S2 N5 $19 D27 R32] 46k 5k$DR because I was confused about what I had written on my data page, and I applied only the same-seed N margin instead of an average. In short, the third play, N should be 3 (actually 3.2) instead of 5. (Everything else is correct.)

There's quite a difference between

GnuBG [Z S1.9 N3.2] 46k+, and
Snowie [S Z10 N17] 31k.

XG aligns more with Snowie than Gnu:

XG 3-ply [S N10 Z11] 15k
XG 4-ply [S Z9 N15] 12k

There's still a lot of variance left. I think Neil is acknowledging that; he would just rather spend his limited CPU time on identifying best plays, even if it would only turn out by a microbe in some cases, and if one slips through the cracks it won't be by much.

That leaves you, me, David and perhaps others naturally curious about what the margins would become with extensions, but presumably we'll find that out in time when Neil eventually circles back to boosting 21$-62 or other XG 4-ply rollouts emerge for that position.


Messages In This Thread


Post Response

Your Name:
Your E-Mail Address:

If necessary, enter your password below:




[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.