BGonline.org Forums

Here's what's wrong with the RO's posted so far.

Posted By: neilkaz
Date: Wednesday, 20 January 2010, at 4:16 p.m.

In Response To: Money game cube action (Timothy Chow)

GNU 2 ply grossly underestimates the leader's chances in this position (like many holding games) and XG 3 ply also clearly underestimates chances. XG 4 ply is somewhat better here. As we see after 61 by both sides the position is a HUGE pass, but both GNU and XG 3 ply still take the cube. XG 4 ply passes now. As a result an XG 4 ply cube RO of the initial position should privide a significantly more accurate assessment of the decision to cube this or not. (XG 4 ply also underestimates the leader's chances here, but not by enough to cause nearly as many bad takes later.

When you have a RO result that is showing as TG for a position that certainly cannot be TG, it is because the bot is making bad takes later on. This is common when the bot (like many do) clearly overestimates the trailer's chances in a holding game.

	is Player 2 score: 0 pip: 143 Money session Jacoby Beaver pip: 121 score: 0 is Player 1
XGID=----abEBCC---Baaab-cbb----:0:0:1:00:0:0:3:0:10
on roll, cube action?

Analyzed in Rollout
No Double
Player Winning Chances:	77.98% (G: 7.58% B: 0.19%)
Opponent Winning Chances:	22.02% (G: 1.89% B: 0.06%)
Double/Take
Player Winning Chances:	78.80% (G: 7.42% B: 0.17%)
Opponent Winning Chances:	21.20% (G: 1.77% B: 0.06%)
Cubeless Equities
No Double:	+0.560
Double:	+1.267
Cubeful Equities
No Double:	+1.017
Double/Take:	+1.140 (+0.123)
Double/Drop:	+1.000 (-0.017)
Best Cube action: Too Good to Double / Drop
Rollout details
1296 Games rolled with Variance Reduction. Dice Seed: Random Moves and cube decisions: 3 ply
Confidence No Double:	± 0.012 (+1.005...+1.030)
Confidence Double:	± 0.015 (+1.125...+1.156)
Double Decision confidence:	99.7%
Take Decision confidence:	100.0%
Duration: 10 minutes 55 seconds

eXtreme Gammon Version: 1.12

GNU 2 ply thinks the diagrammed position is a .91 take !!

XG 3 ply says .984 take. XG 4 ply says 1.02 pass.

We see in the above RO that XG 3 ply is making some bad takes next roll as well.

Messages In This Thread

• Money game cube action
  Timothy Chow -- Monday, 18 January 2010, at 7:43 p.m.
  • Money game cube action
    Leonardo Jerkovic -- Monday, 18 January 2010, at 8:16 p.m.
    • Money game cube action
      Timothy Chow -- Monday, 18 January 2010, at 8:32 p.m.
  • Money game cube action
    Stick -- Monday, 18 January 2010, at 9:02 p.m.
    • Money game cube action
      Petter Bengtsson -- Monday, 18 January 2010, at 9:23 p.m.
    • Money game cube action
      Timothy Chow -- Monday, 18 January 2010, at 10:00 p.m.
  • Money game cube action
    neilkaz -- Tuesday, 19 January 2010, at 4:55 p.m.
  • Rollout
    Timothy Chow -- Wednesday, 20 January 2010, at 2:09 a.m.
    • XG Rollout
      Petter Bengtsson -- Wednesday, 20 January 2010, at 2:27 a.m.
    • Rollout -- another
      Daniel Murphy -- Friday, 22 January 2010, at 1:56 a.m.
  • Here's what's wrong with the RO's posted so far.
    neilkaz -- Wednesday, 20 January 2010, at 4:16 p.m.
    • Layered rollouts
      Nack Ballard -- Wednesday, 20 January 2010, at 5:02 p.m.
      • Holding game 54 payoffs
        neilkaz -- Wednesday, 20 January 2010, at 8:31 p.m.
        • Holding game 54 payoffs
          Nack Ballard -- Wednesday, 20 January 2010, at 9:47 p.m.
    • Here's what's wrong with the RO's posted so far.
      Daniel Murphy -- Wednesday, 20 January 2010, at 10:26 p.m.
  • XG RO with 4 ply cube
    neilkaz -- Wednesday, 20 January 2010, at 4:18 p.m.
    • Very instructive, Neil and Nack! Thanks! *NM*
      Timothy Chow -- Wednesday, 20 January 2010, at 10:21 p.m.