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

BGonline.org Forums

Data summary

Posted By: Nack Ballard
Date: Monday, 8 August 2011, at 7:39 a.m.

In Response To: 1k is a donkey - Extended Rollouts - Ummm? (Stick)

I know Nack doesn't trust full 1k rollouts in that they bring with them a lot more variance than people realize but I wonder if he has more confidence in a 1k XG2 4ply/XGR+ rollout than your old GNU 2/2 rollout?

With so few trials, IMO variance must have a greater impact than whatever the quality difference is between Gnu (2-ply) and XG 4-ply (w/roller+ cube). So, while it may be close, I'd take a Gnu 2k over an XG 4-ply 1k. I'd certainly prefer a Gnu 5k over an XG 4-ply 1k.

On to the details:
139


1O '2X ' '5X '4X ' ' '4O

1X '2O1X '5O '2O ' '1O2X

52S-41T-55P-41H-F-65C-F-43S-31U-63139


Thanks for the rollouts. Here is the data we have for the position above:

    [B=P R8 H72] /1 (Kevin's original rollout)
    [P R5 B42] /1 (Stick's same-seed reroll)
    [P R31 B41] /2 (Stick's different-seed rollout)
    [P R25 B37 H62] /10*5 (Nack's 10k rollout shown below)
    [P R22 B37] "^5+ (Stick's XG rollout)

From this data, [P R23 B37 H64] seems like a good estimate of value, and deviance from that is probably statistical noise, chiefly. Note that I haven't (yet) factored in Kevin's original rollout; because you rerolled with the identical seed, I can't help wondering about corruption. Did you check Kevin's version # of Gnu? (And yours?) In any case, you might learn something by adding P to your reroll.

Nack

139


1O '2X ' '5X '4X ' ' '4O

1X '2O1X '5O '2O ' '1O2X

52S-41T-55P-41H-F-65C-F-43S-31U-63139


# Ply Move Equity
• 1 R 11/5 8/5 +0.122
53.1 16.1 0.6- 46.9 12.7 0.4 +0.098 +0.122
0.0 0.1 0.0- 0.0 0.1 0.0 0.001 0.003
Full cubeful rollout with var.redn.
10368 games, Mersenne Twister dice gen. with seed 924133171 and quasi-random dice
Play: 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 12 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
2 R 24/15 +0.097 ( -0.025)
53.3 11.0 0.5- 46.7 13.0 0.5 +0.046 +0.097
0.0 0.1 0.0- 0.0 0.1 0.0 0.001 0.003
Full cubeful rollout with var.redn.
10368 games, Mersenne Twister dice gen. with seed 924133171 and quasi-random dice
Play: 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 12 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
3 R 24/18 11/8 +0.085 ( -0.037)
52.8 9.7 0.5- 47.2 12.2 0.3 +0.034 +0.085
0.1 0.1 0.0- 0.1 0.1 0.0 0.001 0.004
Full cubeful rollout with var.redn.
5184 games, Mersenne Twister dice gen. with seed 924133171 and quasi-random dice
Play: 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 12 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
4 R 13/4* +0.060 ( -0.062)
51.3 16.8 0.7- 48.7 16.3 0.8 +0.030 +0.060
0.1 0.1 0.0- 0.1 0.1 0.0 0.002 0.005
Full cubeful rollout with var.redn.
5184 games, Mersenne Twister dice gen. with seed 924133171 and quasi-random dice
Play: 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 12 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

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

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