BGonline.org Forums

Rollout: On the bar but anchored on the ace point

Posted By: Timothy Chow
Date: Thursday, 20 May 2010, at 1:11 a.m.

In Response To: On the bar but anchored on the ace point (Timothy Chow)

Over the board, I had to decide whether to take, and the best reference position I knew was the following borderline take/drop position. Scroll down to keep reading.

---

tchow on roll: Cube action?

gnubg	165
tchow	136
Position ID: mHfgAFjMneADIA Match ID: cAkAAAAAAAAA

Cube decision
Rollout cubeless equity		+0.619
Cubeful equities:
1.	Double, pass	+1.000
2.	Double, take	+1.004	+0.004
3.	No double	+0.877	-0.123
Proper cube action:		Double, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Centered 1-cube 0.699 0.266 0.023 - 0.301 0.065 0.003 +0.619 +0.877 Standard error 0.001 0.002 0.001 - 0.001 0.001 0.000 0.003 0.009 Player gnubg owns 2-cube 0.704 0.260 0.024 - 0.296 0.064 0.003 +1.249 +1.004 Standard error 0.002 0.002 0.001 - 0.002 0.001 0.000 0.009 0.015
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 867348751 and quasi-random dice
Play: supremo 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

---

There are a few important adjustments that have to be made between the two positions. In the new position, the attacker has (partially) escaped both back checkers to the bar and has a four-point board. The defender has a stronger board with both the 5 point and the 4 point made. I decided that the attacker's improvements were greater than the defender's improvements and dropped. Evidently it's the other way around. In particular the defender's extra home-board point makes quite a difference; if we unmake the 4 point and put the checkers as spares on the 6 point and 8 point, then I believe it's a drop. Also if the attacker's home-board gap is on the 3 point instead of the 4 point then again I believe it's a drop. It seems that the defender's threat of hitting back holds the attack in check enough to allow a take.

The doubling decision is more in question. I didn't roll it out long enough to be sure but it looks like a borderline double.

I've toyed with the idea of developing a point-count system for blitzes, assigning a certain number of positive or negative points to various features of the position to arrive at a recommendation. Unfortunately this seems rather complicated, but maybe I just haven't tried hard enough. Has anybody else attempted to develop such a system?

---

gnubg on roll: Cube action?

gnubg	125
tchow	164
Position ID: 2M7gAFg2N+DBAA Match ID: MAEAAKAAEAAA

Cube decision
Rollout cubeless equity		+0.565
Cubeful equities:
1.	Double, take	+0.862
2.	Double, pass	+1.000	+0.138
3.	No double	+0.849	-0.013
Proper cube action:		Double, take
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Centered 1-cube 0.672 0.271 0.023 - 0.328 0.071 0.002 +0.565 +0.849 Standard error 0.001 0.001 0.001 - 0.001 0.001 0.000 0.003 0.009 Player tchow owns 2-cube 0.674 0.267 0.025 - 0.326 0.066 0.003 +1.142 +0.862 Standard error 0.002 0.002 0.001 - 0.002 0.001 0.000 0.009 0.014
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 853051919 and quasi-random dice
Play: supremo 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

Messages In This Thread

• On the bar but anchored on the ace point
  Timothy Chow -- Tuesday, 18 May 2010, at 11:12 p.m.
  • On the bar but anchored on the ace point
    John O'Hagan -- Wednesday, 19 May 2010, at 12:56 a.m.
  • On the bar but anchored on the ace point
    Daniel Murphy -- Wednesday, 19 May 2010, at 1:44 a.m.
  • Rollout: On the bar but anchored on the ace point
    Timothy Chow -- Thursday, 20 May 2010, at 1:11 a.m.