Post-blitz cube 2

[Tim Chow]

XGID=aB-CaBCA----c--A-d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 2 O:Player 1
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 131 O: 145 X-O: 0-0
Cube: 1
X on roll, cube action

---
Tim Chow

[badgolferman]

Double/Pass

[smcrtorchs]

It is certainly nice to have gaps at the 2 & 4 instead of the 4 &5, this makes a big difference. I approach this position mostly with counting, although this position is quite far away from what I usually count. The features of the position that I count and I believe deserve to be noticed are (in order of importance):

1) The current inner board structure X has. This structure is strong when the blitz is strong.
2) Reason that this structure is strong when the blitz is strong, is because X's 8 point is moved on his ace point. This factor increases in value the stronger the blitz.
3) Numbers of men in X's home
4) The number of builders X has. 2 here and very precious
5) The abscence of X's mid point
6) the semi burned builder on the 3 point.
7) Applying a very small penalty to X because of O's 2 point.
8) The number of O's men in his zone. This is a small factor, but should be considered.

My counting says it is a D/P although as I am not sure how much I should adjust for some factors I am also not sure if the pass is small or big.

[Tim Chow]

The main thing here, I think, is to recognize that it's a huge double, with the take/pass decision being the question. Although the rollout calls this a 0.9 take, which some might consider a very clear take, a change of two pips turns it into a pass (see variant), so I think the take/pass decision is very difficult without some reference positions or something.

Analyzed in Rollout
No double
Player Winning Chances: 65.04% (G:38.02% B:0.49%)
Opponent Winning Chances: 34.96% (G:9.17% B:0.66%)
Double/Take
Player Winning Chances: 65.10% (G:38.62% B:0.45%)
Opponent Winning Chances: 34.90% (G:9.23% B:0.69%)

Cubeful Equities:
No double: +0.667 (-0.239)
Double/Take: +0.906
Double/Pass: +1.000 (+0.094)

Best Cube action: Double / Take

Rollout:
1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: ± 0.011 (+0.656..+0.678)
Confidence Double: ± 0.016 (+0.890..+0.922)

=======
Variant
=======

XGID=aB-CaBCA----cA---d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 2 O:Player 1
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+

Pip count X: 129 O: 145 X-O: 0-0

Cube: 1
X on roll, cube action

Analyzed in Rollout
No double
Player Winning Chances: 65.90% (G:40.21% B:0.52%)
Opponent Winning Chances: 34.10% (G:8.73% B:0.69%)
Double/Take
Player Winning Chances: 66.00% (G:40.50% B:0.52%)
Opponent Winning Chances: 34.00% (G:8.78% B:0.70%)

Cubeful Equities:
No double: +0.713 (-0.287)
Double/Take: +1.002 (+0.002)
Double/Pass: +1.000

Best Cube action: Double / Pass

Rollout:
1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: ± 0.012 (+0.701..+0.725)
Confidence Double: ± 0.017 (+0.985..+1.019)

eXtreme Gammon Version: 2.10

---
Tim Chow

[smcrtorchs]

I will try to extend the rollouts by monday. I do not think that it is a matter of pips, rather than the fact that you can bring a builder around more easily and this affects all 3 winning chances, winning gammon chances and losing gammon chances. When those 3 add up, then you can see small changes like these to create a 0.080 or even 0.100 swings sometimes.

While reference positions are indeed a certain way to tackle everything in backgammon, I would still try to understand this position from the blitz point of view as long as O has not anchored. I certainly do not claim that I understand the best way to approach these positions, so what I write here is just an opinion.

I am not surprised to see my counting system failing this one, especially since I do not have much time for backgammon lately and it is very easy to confuse something with the system or not knowing a small detail that still needs to be learned.

But recreating the position step by step and seeing how the equity changes from the initial position (this is what upon my system is based) is a good way to approach the position, as long as O has not anchored at least. Sure you will not have this 2 pips accuracy that you mention, but then what kind of set of reference positions are you going to use that will allow you to have such an accuracy?

[Tim Chow]

On Saturday, March 1, 2014 6:11:00 AM UTC-5, smcrtorchs wrote:
> Sure you will not have this 2 pips accuracy that you mention, but then what
> kind of set of reference positions are you going to use that will allow you
> to have such an accuracy?

You are right about that. On the other hand, your first comment is a hint about what one might do. Perhaps what one should count is not pips, but hitting rolls for the attacker versus anchoring rolls for the defender?

---
Tim Chow

[Bradley K. Sherman]

Tim Chow <tchow...@yahoo.com> wrote:
>Score is X:0 O:0. Unlimited Game
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X O | | O X O X |
> | O | | O O |
> | O | | O |
> | O | | O |
> | | X | |
> | |BAR| |
> | | O | |
> | | | |
> | O | | X X |
> | O | | X X X X |
> | O X | | X X O X X |
> +12-11-10--9--8--7-------6--5--4--3--2--1-+

> X:131 O:145, Cube:1, X on roll, cube action?

I have forgotten your rules/posting convention, Tim.
If there's Jacoby, then
D/P
otherwise
TG/P

--bks

[Tim Chow]

On Wednesday, February 26, 2014 10:39:05 AM UTC-5, Bradley K. Sherman wrote:
> I have forgotten your rules/posting convention, Tim.

This is my fault to some extent. When I first started getting active on r.g.b., my convention was to have Jacoby turned on. More recently I have switched to Jacoby being turned off by default. This was in part because I identified one of my weaknesses as cashing when I was too good (in a match), so I needed to turn Jacoby off to study that.

---
Tim Chow

[smcrtorchs]

Are you referring about the specific position and its variant or is it a general question? I counted hitting rolls in your original position and at your variant, but they are the same. In the original position X hits with any 2 or 3 and 65, while at the variant he hits with any 2 or 3 and 54.

I extended your rollouts to about 10k. The swing is smaller according to them. 0.070 instead of 0.100. 1 more thing that should definitely be notices is the huge cube vig that O seems to have here. While the cube vig is always higher at blitzes, it looks that here it is even higher than usual, probably because of the late nature of the blitz.

XGID=aB-CaBCA----c--A-d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 1 O:Player 2

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 131 O: 145 X-O: 0-0
Cube: 1
X on roll, cube action

Analyzed in Rollout
No double

Player Winning Chances: 64.99% (G:38.21% B:0.50%)
Opponent Winning Chances: 35.01% (G:9.16% B:0.64%)
Double/Take
Player Winning Chances: 65.16% (G:38.68% B:0.50%)
Opponent Winning Chances: 34.84% (G:9.23% B:0.66%)

Cubeless Equities: No Double=+0.589, Double=+1.192

Cubeful Equities:
No double: +0.674 (-0.245)
Double/Take: +0.920
Double/Pass: +1.000 (+0.080)

Best Cube action: Double / Take

Rollout:

10180 Games rolled with Variance Reduction.

Moves: 3-ply, cube decisions: XG Roller

Confidence No Double: ± 0.004 (+0.670..+0.679)
Confidence Double: ± 0.007 (+0.913..+0.926)

Double Decision confidence: 100.0%
Take Decision confidence: 100.0%


------------ Variant--------------------

XGID=aB-CaBCA----cA---d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 1 O:Player 2

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 129 O: 145 X-O: 0-0
Cube: 1
X on roll, cube action

Analyzed in Rollout
No double

Player Winning Chances: 65.83% (G:40.06% B:0.53%)
Opponent Winning Chances: 34.17% (G:8.85% B:0.65%)
Double/Take
Player Winning Chances: 65.90% (G:40.26% B:0.53%)
Opponent Winning Chances: 34.10% (G:8.85% B:0.66%)

Cubeless Equities: No Double=+0.628, Double=+1.262

Cubeful Equities:
No double: +0.705 (-0.288)
Double/Take: +0.992
Double/Pass: +1.000 (+0.008)

Best Cube action: Double / Take

Rollout:

12161 Games rolled with Variance Reduction.

Moves: 3-ply, cube decisions: XG Roller

Confidence No Double: ± 0.004 (+0.701..+0.709)
Confidence Double: ± 0.006 (+0.987..+0.998)

Double Decision confidence: 100.0%
Take Decision confidence: 99.4%



I also decided to rollout the positions cubeless. However the cubeless rollouts confuse me. I am not sure how XG produces the equity numbers here.

XGID=aB-CaBCA----c--A-d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 1 O:Player 2

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 131 O: 145 X-O: 0-0
Cube: 1
X on roll, cube action

Analyzed in Rollout
No double

Player Winning Chances: 65.19% (G:37.60% B:0.49%)
Opponent Winning Chances: 34.81% (G:8.79% B:0.58%)
Double/Take
Player Winning Chances: 65.40% (G:38.50% B:0.49%)
Opponent Winning Chances: 34.60% (G:8.95% B:0.62%)

Cubeless Equities: No Double=+0.591, Double=+1.204

Cubeful Equities:
No double: +0.937 (-0.057)
Double/Take: +0.993
Double/Pass: +1.000 (+0.007)

Best Cube action: Double / Take

Rollout:

10368 Games rolled with Variance Reduction.
Cubeless
Moves: 3-ply
Confidence No Double: ± 0.008 (+0.928..+0.945)
Confidence Double: ± 0.009 (+0.984..+1.003)

Double Decision confidence: 100.0%
Take Decision confidence: 91.9%

Duration: 53 minutes 06 seconds


----------------Variant-------------------

XGID=aB-CaBCA----cA---d-d-A-bAA:0:0:1:00:0:0:0:0:10

X:Player 1 O:Player 2

Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O X O X |
| O | | O O |
| O | | O |
| O | | O |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | X X |
| O | | X X X X |
| O X | | X X O X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 129 O: 145 X-O: 0-0
Cube: 1
X on roll, cube action

Analyzed in Rollout
No double

Player Winning Chances: 65.91% (G:39.55% B:0.53%)
Opponent Winning Chances: 34.09% (G:8.38% B:0.61%)
Double/Take
Player Winning Chances: 66.10% (G:40.08% B:0.51%)
Opponent Winning Chances: 33.90% (G:8.59% B:0.65%)

Cubeless Equities: No Double=+0.629, Double=+1.271

Cubeful Equities:
No double: +1.000
Double/Take: +1.048 (+0.048)

Double/Pass: +1.000

Best Cube action: Double / Pass

Rollout:

10284 Games rolled with Variance Reduction.
Cubeless
Moves: 3-ply
Confidence No Double: ± 0.008 (+0.992..+1.008)
Confidence Double: ± 0.009 (+1.039..+1.057)

Double Decision confidence: 50.0%
Take Decision confidence: 100.0%

[Walt]

I'll take a shake. This looks like a pass, and seems to remain a pass
with most sequences.



--
//Walt

[Tim Chow]

On Monday, March 3, 2014 12:28:08 PM UTC-5, smcrtorchs wrote:
> Are you referring about the specific position and its variant or is it a
> general question? I counted hitting rolls in your original position and at
> your variant, but they are the same.

Whoops, I miscounted.

> I also decided to rollout the positions cubeless. However the cubeless
> rollouts confuse me. I am not sure how XG produces the equity numbers here.

How are you doing the cubeless rollouts?

---
Tim Chow

[smcrtorchs]

1) From the options menu, choose settings and go to the rollouts tab.
Once you are there, the first drop down box should say (Presets, Move 3-ply, cube decisions XG Roller).
2) Change the aforementioned choice to "custom settings 1", or "custom settings 2" etc.)
Go to the right side of the middle of the box. It should be saying cube decisions and right next to it there is a checkbox saying cubeless.
3)Check this box.
4) Press the ok button.
5) Choose the position that you want to rollout.
Below the analysis, you should spot the rollout button (at the left of the clock button)
6) Right click at that button and choose the custom settings that you chose earlier.

[Tim Chow]

On Wednesday, March 5, 2014 11:39:10 AM UTC-5, smcrtorchs wrote:
> Go to the right side of the middle of the box. It should be saying cube
> decisions and right next to it there is a checkbox saying cubeless.
>
> 3)Check this box.

So I'm not sure if anything has changed in recent versions of XG, but I believe that in addition to this, you need to set the "maximum cube value" to cubeless, in order to get a true cubeless rollout.

---
Tim Chow

[smcrtorchs]

I have not seen so far a maximum cube value setting in XG 2.1, which is the only XG I have ever used. Where would that might be?

[Tim Chow]

On Friday, March 7, 2014 1:05:02 PM UTC-5, smcrtorchs wrote:
> I have not seen so far a maximum cube value setting in XG 2.1, which is the
> only XG I have ever used. Where would that might be?

I believe it's at the bottom of the Setup menu.

---
Tim Chow

[smcrtorchs]

I do not have such a tickbox, so I guess that it is just an XG version issue and such a thing does not exist in XG2.1

[Tim Chow]

On Friday, March 7, 2014 2:46:43 PM UTC-5, smcrtorchs wrote:

> I do not have such a tickbox, so I guess that it is just an XG version issue
> and such a thing does not exist in XG2.1

No, I have XG2.1 as well. I don't have access to my copy at the moment, but I think maybe the problem is that you're not in a setup mode. Tell the program that you want to set up a new position. Then you should see that there is maximum cube value option at the bottom of the Setup menu.

---
Tim Chow

[Tim Chow]

On Friday, March 7, 2014 5:16:43 PM UTC-5, Tim Chow wrote:
> Tell the program that you want to set up a new position. Then you should see
> that there is maximum cube value option at the bottom of the Setup menu.

More precisely (now that I've had a chance to go back and look at XG), go to the File menu, select "Setup -> New Position" and then a Setup menu will appear as one of the options at the top of the window. Max cube value is the bottom option in that menu.

---
Tim Chow

[smcrtorchs]

Aha! thanx.