Open Side Menu Go to the Top
Register
the big thread about cube decisions the big thread about cube decisions

07-17-2015 , 11:32 AM
I can add that 54 and 63 are additional winning numbers, and 11% of 1 is 0,11 equity point. If white is doubled this number goes times two equals 0,22.

So likely the clue is the decisiveness of a few numbers.

How can a 0,1 error be called a blunder in the case of cube decisions. Bad for our self-esteem.

Last edited by yogiman; 07-17-2015 at 11:50 AM.
the big thread about cube decisions Quote
07-18-2015 , 09:51 AM
Position ID: 78ZBwABr24ABAw
Match ID: MAEAAAAAAAAE


White - Pips 106

Black - Pips 81
Black on roll. Cube action?
Created with www.BGdiagram.com

Last edited by yogiman; 07-18-2015 at 10:11 AM.
the big thread about cube decisions Quote
07-18-2015 , 10:27 AM
So it is white who doubles.
the big thread about cube decisions Quote
07-21-2015 , 05:42 AM
I don't think white has a double. If he fails to hit, there are few rolls that don't leave him open to being hit back. Of course, being hit as white doesn't lead to an immediate disaster, so perhaps he should...

Bah - my doubling play seems to have massively deteriorated lately and I'm not sure why. It is driving me mad. I'll go with no double/take.
the big thread about cube decisions Quote
07-21-2015 , 01:58 PM
White - Pips 106

Black - Pips 81
White on roll. Cube action?
Created with www.BGdiagram.com
Code:
Cube analysis
Rollout cubeless equity  +0,461

Cubeful equities:
1. Double, take         +0,699
2. Double, pass         +1,000  ( +0,301)
3. No double            +0,524  ( -0,176)
Proper cube action: Double, take
It's not so bad if white misses. If he throws less than 6, he can play in his homeboard. More than 6, and he will play from his 13-point. In case he is hit, he is favorite to re-enter. And that 21-point will not be easy to make for black.
Black has a take, because there is no immediate coverage in case he is on the bar.

However, when the 21-point is plugged, his no-double is as strong as the double in the previous position:
White - Pips 106

Black - Pips 87
White on roll. Cube action?
Created with www.BGdiagram.com
Code:
Cube analysis
Rollout cubeless equity  +0,136

Cubeful equities:
1. No double            +0,219
2. Double, pass         +1,000  ( +0,781)
3. Double, take         +0,037  ( -0,182)
Proper cube action: No double, take
the big thread about cube decisions Quote
07-27-2015 , 02:33 AM
Position ID: u7hjAEDcthDABg
Match ID: cAkAAAAAAAAE

White - Pips 95

Black - Pips 139
Black on roll. Cube action?
Created with www.BGdiagram.com
the big thread about cube decisions Quote
07-27-2015 , 07:38 AM
Being so far ahead in the race and with so much play left, I think the opp have an easy take here.

Our blockade is still not completed and vulnerable to some enter-and-run/hit sequences. Opp doesn't have any exposing numbers yet. We don't have many GREAT rolls, that change the position much on the next exchange.

On that basis I wouldn't double yet, although doubling can't be much of an error here. I think we should wait for a more volatile state of this game.
the big thread about cube decisions Quote
07-27-2015 , 09:41 AM
I find these volatile situations very tricky to judge, and yogi's later analysis that shows a single pip in the outfield tips the decision one way or another makes me even more terrified to make a call.

Here I would go with double/take. Our trap is forming, and a couple of rolls see it improve next turn. Conversely, white may be able to get out and run, and he is well ahead in the race.
the big thread about cube decisions Quote
07-28-2015 , 12:35 PM


Code:
Cube analysis
Rollout cubeless equity  +0,465

Cubeful equities:
1. Double, take         +0,741
2. Double, pass         +1,000  ( +0,259)
3. No double            +0,725  ( -0,016)
Proper cube action: Double, take
Quote:
Originally Posted by Karol Szczerek
Being so far ahead in the race and with so much play left, I think the opp have an easy take here.

Our blockade is still not completed and vulnerable to some enter-and-run/hit sequences. Opp doesn't have any exposing numbers yet. We don't have many GREAT rolls, that change the position much on the next exchange.

On that basis I wouldn't double yet, although doubling can't be much of an error here. I think we should wait for a more volatile state of this game.
Everyone is right. It is a borderline decision, though for the inexperienced it looks like a scary take.

Because it costs me nothing I present my experimentations with the initial position. The relevant decisions are in red, the other colors correspond with those in the conclusion.

–-----------

3 to 20:


Code:
Cube analysis
Rollout cubeless equity  +0,434

Cubeful equities:
1. No double            +0,703
2. Double, pass         +1,000  ( +0,297)
3. Double, take         +0,695  ( -0,008)
Proper cube action: No double, take
Like the original position this one is borderline.

Same position, but with 22-point covered:


Code:
Cube analysis
Rollout cubeless equity  +0,365

Cubeful equities:
1. No double            +0,587
2. Double, pass         +1,000  ( +0,413)
3. Double, take         +0,523  ( -0,064)
Proper cube action: No double, take
15 to 18,19:


Code:
Cube analysis
Rollout cubeless equity  +0,476

Cubeful equities:
1. Double, take         +0,782
2. Double, pass         +1,000  ( +0,218)
3. No double            +0,742  ( -0,040)
Proper cube action: Double, take
Same position, but with 22-point covered:
White - Pips 85

Code:
Cube analysis
Rollout cubeless equity  +0,388

Cubeful equities:
1. No double            +0,651
2. Double, pass         +1,000  ( +0,349)
3. Double, take         +0,594  ( -0,057)
Proper cube action: No double, take


---------------

21 to 13:


Code:
Cube analysis
Rollout cubeless equity  +0,621

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,116  ( +0,116)
3. No double            +0,870  ( -0,130)
Proper cube action: Double, pass
Same position, but with 22-point covered:



Code:
Cube analysis
Rollout cubeless equity  +0,546

Cubeful equities:
1. Double, take         +0,947
2. Double, pass         +1,000  ( +0,053)
3. No double            +0,817  ( -0,131)
Proper cube action: Double, take

3 to 13:


Code:
Cube analysis
Rollout cubeless equity  +0,586

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,087  ( +0,087)
3. No double            +0,840  ( -0,160)
Proper cube action: Double, pass
The pipcount doesn't matter. Black is likely going to make a 5-prime. Period.

Same position, but with 22-point covered:



Code:
Cube analysis
Rollout cubeless equity  +0,547

Cubeful equities:
1. Double, take         +0,952
2. Double, pass         +1,000  ( +0,048)
3. No double            +0,808  ( -0,144)
Proper cube action: Double, take
----------

18,19 to 16:


Code:
Cube analysis
Rollout cubeless equity  +0,547

Cubeful equities:
1. Double, take         +0,918
2. Double, pass         +1,000  ( +0,082)
3. No double            +0,764  ( -0,154)
Proper cube action: Double, take

To see what is the role of the killed checker, I move it to the 16-point:



Code:
Cube analysis
Rollout cubeless equity  +0,354

Cubeful equities:
1. No double            +0,560
2. Double, pass         +1,000  ( +0,440)
3. Double, take         +0,463  ( -0,097)
Proper cube action: No double, take
----------

Fast simplified conclusion:
If black has at most 1 checker to cover for the 5-prime blot, and white has:
killed checker, then black can play the player, and white has a take;
no killed checker, then black has no double.

If black has at least 2 checkers to cover for the 5-prime blot, then white better passes.
the big thread about cube decisions Quote
07-29-2015 , 12:31 PM
Position ID: u7gxABhsmxjABQ
Match ID: cAkAAAAAAAAE

White - Pips 109

Black - Pips 147
Black on roll. Cube action?
Created with www.BGdiagram.com
the big thread about cube decisions Quote
07-30-2015 , 09:31 AM
Double/take.

Black has some good chances to make a 5 prime, or 4+1 prime. At the same time, he has checkers back that give him some time, and are also going to contribute to the prime. White is looking like he likely start crunching, unless he wants to try running soon.

White is pretty far ahead, and some decent dice will see him suddenly through all black's pieces, with a clear lead.
the big thread about cube decisions Quote
07-31-2015 , 07:43 AM


Code:
Cube analysis 
Rollout cubeless equity  +0,545 

Cubeful equities: 
1. Double, take         +0,880 
2. Double, pass         +1,000  ( +0,120) 
3. No double            +0,798  ( -0,081) 
Proper cube action: Double, take
In fact this is an unrealistic situation, because what is that blot doing there on the 8-point? Maybe white played 11 last turn, or entered from the bar. By the way, only with that many black checkers behind it will be a take, otherwise it's a pass.

These are more relevant positions (144 trials likely suffices in these cases):




Code:
Cube analysis
Rollout cubeless equity  +0,653

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,162  ( +0,162)
3. No double            +0,902  ( -0,098)
Proper cube action: Double, pass

With a 4-board and a killed checker, white has a narrow pass:


Code:
Cube analysis
Rollout cubeless equity  +0,590

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,020  ( +0,020)
3. No double            +0,825  ( -0,175)
Proper cube action: Double, pass
With a 4-board and no killed checker, white has a strong take:



Code:
Cube analysis
Rollout cubeless equity  +0,510

Cubeful equities:
1. Double, take         +0,838
2. Double, pass         +1,000  ( +0,162)
3. No double            +0,776  ( -0,063)
Proper cube action: Double, take
However, with 4 checkers behind, black has no double:


Code:
Cube analysis
Rollout cubeless equity  +0,366

Cubeful equities:
1. No double            +0,594
2. Double, pass         +1,000  ( +0,406)
3. Double, take         +0,509  ( -0,084)
Proper cube action: No double, take

15w to 17:


Code:
Cube analysis
Rollout cubeless equity  +0,522

Cubeful equities:
1. Double, take         +0,857
2. Double, pass         +1,000  ( +0,143)
3. No double            +0,800  ( -0,057)
Proper cube action: Double, take
Containing the 20-point has a slight advantage.




White has a 5-board:


Code:
Cube analysis
Rollout cubeless equity  +0,521

Cubeful equities:
1. Double, take         +0,889
2. Double, pass         +1,000  ( +0,111)
3. No double            +0,805  ( -0,084)
Proper cube action: Double, take
White's position is a little worse compared to a 4-board, because she runs out of flexibility.



Code:
Cube analysis
Rollout cubeless equity  +0,558

Cubeful equities:
1. Double, take         +0,966
2. Double, pass         +1,000  ( +0,034)
3. No double            +0,813  ( -0,153)
Proper cube action: Double, take

Fast simple conclusion:
If white has a 4 or 5-board, and black has a defensive point, white will have a take. Black will have no double if he has 4 checkers behind.




PS I hope you guys will come up with positions that contradict my statements, so that I can refine them.
the big thread about cube decisions Quote
08-12-2015 , 09:19 AM
Position ID: 93sBAAjmtmEIQA
Match ID: AxkAAAAAAAAE
White - Pips 114

Black - Pips 53
White doubles. Take or Pass?
Created with www.BGdiagram.com
the big thread about cube decisions Quote
08-12-2015 , 12:29 PM
Wow, tough one.
I would assess the gammon threat from White as small (moderate at worst), so I would take since I'm not sure at all.

44 and 41 are obviously great rolls for White. If White dances, even 55 and 66 will leave a shot, but at least 66 will eliminate the gammon threat and provide Black with gammon opportunities for himself if White misses the blot.

I buckle up and take.
the big thread about cube decisions Quote
08-12-2015 , 07:21 PM
White has 9 dances, 11 numbers that hit the guy on the 21, and 16 other numbers.

On the 9 dances, I think Black might be a favorite when holding the cube. So he gets 4.5+ wins there.

The 11 hits, Black might not win any after accounting for gammons, but actually I think he does better a little than a single loss on average. I'm thinking in the ball park of 20% gammons, but he wins over 10%. so maybe half a win there.

So we're already up to 5 wins, maybe more. He just needs to win 4 of the 16 other numbers to have a take, which I think he clearly can, even though there's a few decent numbers that hit on White's 3 point. I mean I would definitely take if I knew White was going to roll one of those 16 numbers, so I must think Black wins more than 25% of the 16 numbers.

So I'm feeling pretty solid on take.
the big thread about cube decisions Quote
08-14-2015 , 05:58 AM
Thanks a lot for your analysis, _z_.

This was the original position:


Code:
Cube analysis 
Rollout cubeless equity  +0,466 

Cubeful equities: 
1. Double, take         +0,809 
2. Double, pass         +1,000  ( +0,191) 
3. No double            +0,654  ( -0,155) 
Proper cube action: Redouble, take
Sorry guys, it is awkward to play around without reversing the board. I put the cube in the middle.
Position ID: 93sBAAjmtmEIQA
Match ID: cAkAAAAAAAAE


Black's front needs to get pretty crowdy in order for white to have a pass:


Code:
Cube analysis
Rollout cubeless equity  +0,576

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,046  ( +0,046)
3. No double            +0,664  ( -0,336)
Proper cube action: Double, pass
For practical purposes, black will always have a double:


Code:
Cube analysis
Rollout cubeless equity  +0,237

Cubeful equities:
1. No double            +0,402
2. Double, pass         +1,000  ( +0,598)
3. Double, take         +0,357  ( -0,045)
Proper cube action: No double, take
Situation:
Black is on the bar and can hit a blot, while another white checker is just behind black's 3point-board.
Fast and simple conclusion:
If white has not more than a 3-board, black has a double.
If black's front position is not too strong, white will have a take.
the big thread about cube decisions Quote
08-14-2015 , 10:01 AM
Looks like my numbers were pretty close except for the gammons after the hit. It's more like 30%+, not 20%. So the taker really wins more like -0.5 games of the 9 hits. But there's still enough wins elsewhere to make up for that.
the big thread about cube decisions Quote
08-15-2015 , 11:11 AM
Position ID: 33cQgAB2uwMEAQ
Match ID: UQkAAAAAAAAE

White - Pips 50

Black - Pips 87
Black on roll. Cube action?
Created with www.BGdiagram.com
the big thread about cube decisions Quote
08-15-2015 , 01:01 PM
Black hits 19/36, barely over 50%. White should have a strong redouble if missed.

I don't think it's a beaver however, because of gammon threats if White is hit.

No double/Huge Take
the big thread about cube decisions Quote
08-15-2015 , 01:14 PM
Double is clear per O'Hagan's Law (Your market losers minus anti-jokers count has to be 9 for initial double and 11-12 for redouble).

We have 18 hitters, which are mostly market losers.
If we don't hit, he mostly won't be strong enough to double, which is a definition for anti-joker for O'Hagan's Law, that I've learned from Stick.

Now the pass/take decision.
After 18 hits we win, say, 16 times.
After 18 non-hitters, I think we will average around 35% wins, so 6 wins.

That leaves 14/36 wins for the opp.
Our gammon chances after hits aren't that high, I'd say 20% (2-checker closeout is ca. 40%). So 10% before the roll. Not quite high enough to spoil opp's 14 wins.
Easy take.

EDIT:
I've missed 66, which gives 19 hitters. That doesn't change above calculation for easy double and easy take.

Last edited by Karol Szczerek; 08-15-2015 at 01:17 PM. Reason: counting mistake
the big thread about cube decisions Quote
08-17-2015 , 02:46 PM
White - Pips 50

Code:
Cube analysis
Rollout cubeless equity  +0,504

Cubeful equities:
1. Double, take         +0,798
2. Double, pass         +1,000  ( +0,202)
3. No double            +0,566  ( -0,232)
Proper cube action: Redouble, take
White has a very strong take.

Moving white 17 to 15 holds a surprise:



Code:
Cube analysis
Rollout cubeless equity  +0,630

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,071  ( +0,071)
3. No double            +0,689  ( -0,311)
Proper cube action: Redouble, pass
Black hits only with an additional 31 and 22. But if black misses, and white hits with her rear checker, it's also more difficult to safety that checker. This is made clear, if black 19 goes to 21:


Code:
Cube analysis
Rollout cubeless equity  +0,539

Cubeful equities:
1. Double, take         +0,861
2. Double, pass         +1,000  ( +0,139)
3. No double            +0,600  ( -0,261)
Proper cube action: Redouble, take
However, the more black is in the back, the better:


Code:
Cube analysis
Rollout cubeless equity  +0,511

Cubeful equities:
1. Double, take         +0,804
2. Double, pass         +1,000  ( +0,196)
3. No double            +0,576  ( -0,228)
Proper cube action: Redouble, take
So white has a fold, if she is 7 pips removed from safety, and black is in white's homeboard. However, this situation is so rare, that it is better to forget.

--------

In the next two positions white's board is improved.

White has a 4-board:



Code:
Cube analysis
Rollout cubeless equity  +0,415

Cubeful equities:
1. Double, take         +0,622
2. Double, pass         +1,000  ( +0,378)
3. No double            +0,487  ( -0,135)
Proper cube action: Redouble, take
White has a 5-board:


Code:
Cube analysis
Rollout cubeless equity  +0,325

Cubeful equities:
1. No double            +0,396
2. Double, pass         +1,000  ( +0,604)
3. Double, take         +0,317  ( -0,079)
Proper cube action: No redouble, take
With a white 4-board black's decisions are likely the same, but with a 5-board black has no double.

----------

open 4-point:



Code:
Cube analysis
Rollout cubeless equity  +0,321

Cubeful equities:
1. No double            +0,428
2. Double, pass         +1,000  ( +0,572)
3. Double, take         +0,395  ( -0,033)
Proper cube action: No redouble, take
open 5-point:


Code:
Cube analysis
Rollout cubeless equity  +0,286

Cubeful equities:
1. No double            +0,417
2. Double, pass         +1,000  ( +0,583)
3. Double, take         +0,348  ( -0,069)
Proper cube action: No redouble, take
These are more difficult to assess with o'hagans law, because the volatility is less, i.e. a black hit is less decisive. Better don't double with an open 3, 4, or 5-point.

---------

White has a pass, if black's board is closed:



Code:
Cube analysis
Rollout cubeless equity  +0,613

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,029  ( +0,029)
3. No double            +0,651  ( -0,349)
Proper cube action: Redouble, pass
It's a lot easier now to pick up a second checker.

---------

If black holds the 14-point, his double is somewhat stronger, because if he misses it's more difficult for white to hit back:



Code:
Cube analysis
Rollout cubeless equity  +0,528

Cubeful equities:
1. Double, take         +0,889
2. Double, pass         +1,000  ( +0,111)
3. No double            +0,592  ( -0,297)
Proper cube action: Redouble, take
Black holds the 19-point:


Code:
Cube analysis
Rollout cubeless equity  +0,604

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,063  ( +0,063)
3. No double            +0,681  ( -0,319)
Proper cube action: Redouble, pass
If black misses, white can safety her front blot. With three split checkers she will have a hard time to get home.




Situation:
Black has at least a 5-board, and black's rear checker has a direct shot and his front checker has an indirect shot.
White's homeboard is crunched.

Fast and simple conclusion:
Black has no double if he has a broken 5-board, or in case white has a 5-board.
White has a pass if black has a rear point, or has a closed board.

Last edited by yogiman; 08-17-2015 at 02:51 PM. Reason: i am just not infallible
the big thread about cube decisions Quote
09-01-2015 , 04:24 AM
Position ID: tm0GAwLemG0BAA
Match ID: cAkAAAAAAAAE


White - Pips 102

Black - Pips 91
Black on roll. Cube action?
Created with www.BGdiagram.com
the big thread about cube decisions Quote
09-01-2015 , 02:30 PM
double/take
the big thread about cube decisions Quote
09-02-2015 , 02:14 AM
Based on the race alone, I think this is definitely a double. Add to this the chance of making points on white's head as we bear in, and I think that it is perilously close to a fold. However, white has a good trap, and there are going to be a lot of combinations of black rolls that end up leaving a shot for white's lone checker, which pushes me into taking. also, blacks heavy 2 point must lean us in a taking direction as well.

Welcome back yogi. Hopefully no romantic chances were hurt in making this post!
the big thread about cube decisions Quote
09-02-2015 , 09:02 AM
i would guess double / take.

I am thinking that black has some good rolls, but is also looking pretty awkward with two checkers on each point and big gaps in his home. So in my mind the pros and cons neutralise each other and I'll base my decision purely on the pip counts.
the big thread about cube decisions Quote

      
m