BGonline.org Forums

Preliminary Rollouts

Posted By: Matt Cohn-Geier
Date: Friday, 14 March 2008, at 5:23 p.m.

In Response To: Position from Madison 2007 (Matt Cohn-Geier)

So my thoughts on the initial position were this: maybe it's very slightly TG (if I escape, I have a clear but very marginal play-on), or maybe it's very slightly D/P (44 hurts me; after everything else it's still a giant D/P). Against an equal player, I would probably double. However, I have no idea what my opponent will do. I'm pretty sure he'd drop it as is, but if I roll a big number like 65 or something, maybe he'll incorrectly take. So I roll on. I'm somewhat relieved to hear that I have support on BGO for taking this line.

But eventually I got to thinking. When I was on the bar and holding an 8 cube, I wasn't very phased outwardly, just rolling the dice and playing backgammon. Them's the breaks. But my internal thoughts, if they had been in English syntax, would have been "DEAR GOD WHY DIDN'T I JUST F--- CASH THIS" or something to that effect.

He's substantially weaker than me--200 ELO sounds about right, based on GamesGrid ratings. And the truth is, I don't even know my 9-away METs. I could estimate them, sure. And though I've read Jacobs & Trice's book, there's no way I could possibly accurately adjust them for the ELO difference. So what if I have "fooled" him into a correct take? The truth is that I have NO IDEA what the doubling window or the takepoint is anymore, and probably no backgammon player would. Memorizing ELO-adjusted METs is a level of fanaticism one would only be able to find amongst chess players.

Of course Keene is correct that if I had been playing in the Open this would never have come up. But if I played in the Open, how would I eat? On the other hand, I ended up winning this match and still not cashing in the Advanced, for whatever that's worth.

After reconsidering, I'd cash every day of the week and twice on Sunday. Winning a couple points from -3, -9 is a lot easier than chalking most of the match up to one 8-cube, no matter how bad his take may have been in theory.

The following rollouts confirm the consensus, but these are all based on g11. JT 100 point ELO rollouts to come, since I don't have JT 200 tables.

The score (after 0 games) is: White 4, Blue 0 (match to 9 points)

Move number 8: White to play 32

White	88
Blue	81
Position ID: c3cHAEDbdgNAEA Match ID: AYEpAUAAAAAA

• White doubles

Cube decision
Rollout cubeless equity		+0.8967	(Money: +0.8996)
Cubeful equities:
1.	Double, pass	+1.0000
2.	Double, take	+1.6459	+0.6459
3.	No double	+0.9928	-0.0072
Proper cube action:		Redouble, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player White owns 2-cube 0.9368 0.0283 0.0004 - 0.0632 0.0027 0.0000 +0.8967 +0.9928 Standard error 0.0005 0.0009 0.0000 - 0.0005 0.0001 0.0000 0.0012 0.0019 Player Blue owns 4-cube 0.9499 0.0264 0.0008 - 0.0501 0.0011 0.0000 +1.7546 +1.6459 Standard error 0.0009 0.0021 0.0004 - 0.0009 0.0002 0.0000 0.0037 0.0063
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 1056795567 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

The score (after 0 games) is: White 4, Blue 0 (match to 9 points)

Move number 17: White on roll, cube decision?

White	76
Blue	81
Position ID: c3cHAEDbdgcAEA Match ID: AQEgAUAAAAAA

• White doubles

Cube decision
Rollout cubeless equity		+0.7572	(Money: +0.7635)
Cubeful equities:
1.	Double, pass	+1.0000
2.	Double, take	+1.1639	+0.1639
3.	No double	+0.8305	-0.1695
Proper cube action:		Redouble, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player White owns 2-cube 0.8689 0.0296 0.0003 - 0.1311 0.0041 0.0001 +0.7572 +0.8305 Standard error 0.0012 0.0016 0.0000 - 0.0012 0.0003 0.0000 0.0025 0.0054 Player Blue owns 4-cube 0.8795 0.0307 0.0002 - 0.1205 0.0035 0.0003 +1.4574 +1.1639 Standard error 0.0013 0.0036 0.0000 - 0.0013 0.0004 0.0002 0.0053 0.0100
Full cubeful rollout with var.redn.
997 games, Mersenne Twister dice gen. with seed 1056795567 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

The score (after 0 games) is: White 4, Blue 0 (match to 9 points)

Move number 21: Blue on roll, cube decision?

White	22
Blue	81
Position ID: 2wsAAMzdHQAAAQ Match ID: UgkgAUAAAAAA

• Blue doubles

Alert: wrong double ( -0.1123)! [bad]

Cube decision
Rollout cubeless equity		-0.6592	(Money: -0.7166)
Cubeful equities:
1.	No double	-0.4415
2.	Double, pass	+1.0000	+1.4415
3.	Double, take	-0.5538	-0.1123
Proper cube action:		No redouble, take (7.2%)
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player Blue owns 4-cube 0.1787 0.0000 0.0000 - 0.8213 0.0721 0.0018 -0.6592 -0.4415 Standard error 0.0004 0.0000 0.0000 - 0.0004 0.0010 0.0005 0.0007 0.0012 Player White owns 8-cube 0.1818 0.0000 0.0000 - 0.8182 0.3035 0.0029 -0.5538 -0.5538 Standard error 0.0005 0.0000 0.0000 - 0.0005 0.0105 0.0005 0.0019 0.0019
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 1056795567 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

Messages In This Thread

• Position from Madison 2007
  Matt Cohn-Geier -- Thursday, 13 March 2008, at 9:54 p.m.
  • Position from Madison 2007
    Stick -- Thursday, 13 March 2008, at 10:40 p.m.
    • Position from Madison 2007
      Matt Cohn-Geier -- Thursday, 13 March 2008, at 10:57 p.m.
      • Position from Madison 2007
        Stick -- Thursday, 13 March 2008, at 11:07 p.m.
  • Position from Madison 2007
    neilkaz -- Thursday, 13 March 2008, at 11:51 p.m.
  • Position from Madison 2007
    Daniel Murphy -- Friday, 14 March 2008, at 2:09 a.m.
  • Position from Madison 2007
    Keene -- Friday, 14 March 2008, at 5:58 a.m.
  • Preliminary Rollouts
    Matt Cohn-Geier -- Friday, 14 March 2008, at 5:23 p.m.
    • Preliminary Rollouts
      Daniel Murphy -- Friday, 14 March 2008, at 5:50 p.m.
  • ROLLOUTS: Nine more
    Matt Cohn-Geier -- Monday, 17 March 2008, at 1:23 a.m.