the big thread about cube decisions - Page 5 - Backgammon Forum - Discuss Backgammon Strategy, Clubs and Books

I dont agree with the statement about the 5-2 "primed" backgame. Im sure (you may check) that if you put the white men on the 2 point from the 1 point making a 2-5 backgame with a man in the bar it turns into a massive take. The only bad thing here imo it is the fact that black has three men that can hit white on the ace and deuce point Messing up the position because of the gammon factor.

Quote

04-02-2015 , 05:10 PM

#102

_Z_

adept

Join Date: Jul 2004 Posts: 720

Double looks pretty clear to me, since there's a obvious big edge here and many immediate sequences will really crush White.

I think White has enough counter-play to take. Black has some work to do getting his back checkers around and White may be able to establish a second anchor. On the other hand, White really has his forces separated and doesn't have much of an offence. So double/take, but not sure about the take.

Quote

04-03-2015 , 12:07 AM

#103

Taper_Mike

adept

Join Date: Sep 2009 Posts: 1,030

Quote:

Originally Posted by Fllecha

The only bad thing here imo it is the fact that black has three men that can hit white on the ace and deuce point Messing up the position because of the gammon factor.

Although that will happen in many variations, I see another possibility that I think happens more often. I see Black gaining outfield control, and hitting White every time he jumps out. The result is White crashes, and/or White is forced to run off his 5pt anchor. Either of these variations is highly favorable to Black.

Mike

Quote

04-03-2015 , 11:34 AM

#104

yogiman

adept

Join Date: Jun 2012 Posts: 784

As a matter of fact taking the cube is a double whopper.

Code:

Cube analysis
Rollout cubeless equity  +0,668

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,199  ( +0,199)
3. No double            +0,977  ( -0,023)
Proper cube action: Double, pass

As usual, you guys have deserved some background information.

Adding outfield control for white with the same lousy 2-board makes it a take ofcourse:

Code:

Cube analysis
Rollout cubeless equity  +0,553

Cubeful equities:
1. Double, take         +0,854
2. Double, pass         +1,000  ( +0,146)
3. No double            +0,780  ( -0,074)
Proper cube action: Double, take

And less outfield control is compensated by building capacity:

Code:

Cube analysis
Rollout cubeless equity  +0,564

Cubeful equities:
1. Double, take         +0,924
2. Double, pass         +1,000  ( +0,076)
3. No double            +0,846  ( -0,078)
Proper cube action: Double, take

Creating a 3-board by moving white's builders from the 19wpoint to the 23point:

Code:

Cube analysis
Rollout cubeless equity  +0,623

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,053  ( +0,053)
3. No double            +0,930  ( -0,070)
Proper cube action: Double, pass

This 3-board is as clumsy as it may get. A 3-board is enough to make it a take,as far as me concerns.

Moving the 22-point builder to the 19-point makes the take official:

Code:

Cube analysis
Rollout cubeless equity  +0,571

Cubeful equities:
1. Double, take         +0,943
2. Double, pass         +1,000  ( +0,057)
3. No double            +0,859  ( -0,084)
Proper cube action: Double, take

So it is a pass in absence of an outfield point or a 3-board.

And this position is for Fllecha and Taper_Mike:

Code:

Cube analysis
Rollout cubeless equity  +0,561

Cubeful equities:
1. Double, take         +0,930
2. Double, pass         +1,000  ( +0,070)
3. No double            +0,905  ( -0,025)
Proper cube action: Double, take

Quote

04-03-2015 , 03:33 PM

#105

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

Thanks for the analysis, I got some confirmations on the 5-2 backgame, even if its a close take. The fact is that, even if a 5-2 backgame isn't appealing, it has resources if opponent has some men back and we have all checkers in play and some timing.

Quote

04-04-2015 , 01:58 AM

#106

yogiman

adept

Join Date: Jun 2012 Posts: 784

You all spotted the error: 23 means 22 and 22 means 21

Quote

04-06-2015 , 12:44 PM

#107

yogiman

adept

Join Date: Jun 2012 Posts: 784

Position ID: TRsA+AmY7VgCYA Match ID: UQkAAAAAAAAE

White - Pips 173

Black - Pips 156
Black on roll. Cube action?
Created with www.BGdiagram.com

Quote

04-06-2015 , 02:14 PM

#108

uberkuber

veteran

Join Date: Jul 2009 Posts: 3,032

Double/?

Not even sure it's a take. Yes Black has 2 men on the bar, has the worse homeboard and White has the golden point. But White is disconnected and has surrendered outfield control. And I learned with the last problem that the golden point doesn't automatically equal take.

Quote

04-06-2015 , 04:11 PM

#109

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

This is a very rare position and its imo only for accademic use, almost impossible to store as a reference position. Whatever: otb id probably double and I think at least 90% of 1600-1800 players would drop ( i call this "confusion drop") because of lack of outfield control and too many men back.

Quote

04-06-2015 , 04:59 PM

#110

Taper_Mike

adept

Join Date: Sep 2009 Posts: 1,030

Everything Fllecha said.

I'd like to apply Woolsey's law, and declare this a double based on the fact that I am not sure whether to take or pass. My uncertainty, however, may have more to do with the unfamiliarity of the position than anything else.

OTB, I cube this, for sure. Looking at Black's 9 dances, and his other entering numbers that fail to hit, I'd try a take. If hit, I would hope to make a second anchor. If not, I would hope for some tactical action in the outfield.

Mike

Quote

04-07-2015 , 03:36 AM

#111

Kamba

enthusiast

Join Date: Feb 2015 Posts: 93

I would take this, based on the confusion factor.

Also I'm thinking that if I don't get primed or crunch I might just have time to build a game.

Last edited by Kamba; 04-07-2015 at 03:37 AM. Reason: spelling

Quote

04-07-2015 , 04:36 AM

#112

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

In reality i lean on a real drop, because with out midpoint at least or some outside point i dont think that white has real counter possibility. I Also think that taking here is somewhat a sign of too much loose takes imo. Usually a lot of position with an anchor has some chanches but with out some structure to go with youre better of consider dropping. Here for me its

Sure double/pass

Quote

04-07-2015 , 09:11 AM

#113

Robertie

old hand

Join Date: May 2004 Posts: 1,954

The double is easy, so now let's look at the take/drop decision.

Hmm -- down 17 in the race, 6 guys stacked up on one point, a double shot at my loose blots, no outfield presence -- I feel confusion coming on. I drop.

Quote

04-07-2015 , 01:02 PM

#114

yogiman

adept

Join Date: Jun 2012 Posts: 784

Code:

Cube analysis
Rollout cubeless equity  +0,613

Cubeful equities:
1. Double, take         +0,987
2. Double, pass         +1,000  ( +0,013)
3. No double            +0,872  ( -0,115)
Proper cube action: Redouble, take

So even Bill Robertie has got it wrong.(lol)

But the position is very sensitive to alterations, as I will show.

Moving from 11 to 10, so that fives have to be played in the homeboard, leads to a very convincing pass:

Code:

Cube analysis
Rollout cubeless equity  +0,712

Cubeful equities:
1. Double, pass         +1,000
2. Double, take         +1,233  ( +0,233)
3. No double            +0,945  ( -0,055)
Proper cube action: Redouble, pass

Creating a third point in black's homeboard makes it even too good to double:

Code:

Cube analysis
Rollout cubeless equity  +0,778

Cubeful equities:
1. No double            +1,013
2. Double, take         +1,366  ( +0,353)
3. Double, pass         +1,000  ( -0,013)
Proper cube action: Too good to redouble, pass (3,6%)

Moving 24w to 22 results in a strong improvement in white's position:

Code:

Cube analysis
Rollout cubeless equity  +0,547

Cubeful equities:
1. Double, take         +0,868
2. Double, pass         +1,000  ( +0,132)
3. No double            +0,806  ( -0,063)
Proper cube action: Redouble, take

Putting a builder on 19w makes doubling a whopper whopper:

Code:

Cube analysis
Rollout cubeless equity  +0,305

Cubeful equities:
1. No double            +0,559
2. Double, pass         +1,000  ( +0,441)
3. Double, take         +0,319  ( -0,240)
Proper cube action: No redouble, take (35,2%)

Quote

04-07-2015 , 04:10 PM

#115

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

Mm can you perform (yogi) a serious rollout of the original position? It Will surely lead to a pass imo

Quote

04-07-2015 , 07:26 PM

#116

uberkuber

veteran

Join Date: Jul 2009 Posts: 3,032

On the contrary Yogi, we pretty much all acknowledged that it was a solid double and a borderline take/pass. I think we more or less nailed it in fact.

Quote

04-08-2015 , 12:52 AM

#117

Taper_Mike

adept

Join Date: Sep 2009 Posts: 1,030

Quote:

Originally Posted by Fllecha

Mm can you perform (yogi) a serious rollout of the original position? It Will surely lead to a pass imo

You very well may be right. Perhaps you could help Yogiman and me carry the load. I know we have done our share of rollouts. If you cannot, there may be someone else who can help out.

Mike

Quote

04-08-2015 , 04:12 AM

#118

yogiman

adept

Join Date: Jun 2012 Posts: 784

Thank you, Mike. My dedication is in no way a professional one. I have other interests than backgammon alone. As said before, one might argue whether it is relevant to do precision rollouts, when even the best players are making mistakes.

When everything is behind and I have not more than a broken 3-board, like in this case, I propose for myself as a general rule to make it a pass.

Quote

04-08-2015 , 02:32 PM

#119

Karol Szczerek

enthusiast

Join Date: Mar 2015 Posts: 92

A rollout for original position #13.

Code:

No redouble
  Player Winning Chances:   69,99% (G:32,72% B:1,64%)
  Opponent Winning Chances: 30,01% (G:8,25% B:0,39%)
Redouble/Take
  Player Winning Chances:   69,81% (G:32,23% B:1,49%)
  Opponent Winning Chances: 30,19% (G:8,22% B:0,39%)

Cubeless Equities: No Double=+0,657, Double=+1,295

Cubeful Equities:
       No redouble:     +0,894 (-0,106)
       Redouble/Take:   +1,056 (+0,056)
       Redouble/Pass:   +1,000

Best Cube action: Redouble / Pass

Rollout:
  2592 Games rolled with Variance Reduction.
  Moves: 3-ply, cube decisions: XG Roller
  Confidence No Double: ± 0,011 (+0,883..+0,905)
  Confidence Double:    ± 0,020 (+1,036..+1,076)

  Double Decision confidence: 100,0%
  Take Decision confidence: 100,0%

  Duration: 1 hour 22 minutes

eXtreme Gammon Version: 2.19.208.pre-release

Quote

04-08-2015 , 03:18 PM

#120

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

Thanks for the rollout. So Its a pass not a bare take and 0.06 its not that much borderline. Ok not a serious error but not definitively a toss up

Quote

04-08-2015 , 06:04 PM

#121

Taper_Mike

adept

Join Date: Sep 2009 Posts: 1,030

Quote:

Originally Posted by Fllecha

Thanks for the rollout. So Its a pass not a bare take and 0.06 its not that much borderline. Ok not a serious error but not definitively a toss up

Thanks, Karol.

I know of several Backgammon Giants who say that they cannot do much better than ±0.06 on a great many cube decisions. For checker plays, ±0.015 or ±0.02 may be a close choice, but for cubes, ±0.06 is often as "close" as they can get.

In my own play, I don't worry too much about a cube error of that size. I've got bigger holes to plug!

@Bill: If you read this, would you mind telling us what you generally aim for? Thanks.

Mike

Last edited by Taper_Mike; 04-08-2015 at 06:09 PM.

Quote

04-09-2015 , 01:57 AM

#122

Fllecha

journeyman

Join Date: Mar 2010 Posts: 321

Dont get the carol. Dont want to Be like the pope but when you discuss equity difference in checker play around 0.07 in this forum i Will say you the same thing: better leaks to fill. But in that case it matters because everybody spot the difference here dont matters. Peace.

Quote

04-09-2015 , 04:20 AM

#123

yogiman

adept

Join Date: Jun 2012 Posts: 784

As a matter of fact I posed this problem because 2-ply evaluation presented a pass. Before posting the solution I did the apparently incomplete rollout. As we are aware of, in unusual complex situations our judgements are better than the best rollouts of the bot. This position has the potential to develop into a jungle situation. So how sure are we that our polish comrade's rollout is adequate? And suppose we could be sure that the bot served the exact data on basis of perfect play, how many of us would be really helped by it, as a little percentage of the viewers here are the top of the top. In a position like this, I tend to see the rollout as an indication, and rather listen to the experience expert, in this case our moderator.

But it is a great thing that we can discuss those issues here. Unless objected, I will continue with the same rollout settings, and you are very welcome to correct it with your own rollout in case it is really off target.

Quote

04-09-2015 , 06:03 AM

#124

Taper_Mike

adept

Join Date: Sep 2009 Posts: 1,030

Bill Robertie and the many other users of Snowie back in the day when computers were a whole lot slower know the value of a truncated rollout. When the number of trials is small, a trunc is usually better than a full rollout.

The trunc used by Yogiman is very good. It uses GnuBg's "world class" 2-ply evals to a depth of 10 plies. The only thing I would change is the number of trials. I would like to see a minimum of 324, and twice that many for "difficult" positions. Note that 324 = 1296/4, while 648 = 1296/2. A good compromise might be to use the same number of trials as XG does with its XGR++ trunc. That number is 360.

Just my two cents.

Mike

Quote

04-09-2015 , 09:21 AM

#125

bleep69

grinder

Join Date: Sep 2008 Posts: 483

I hesitate to post problems, because what is confusing to someone of my level is seldom confusing to the majority of the posters here, but this has me stumped. I've run it past another, equally inexperienced player, and he's also stumped, so here goes...

Black on roll, cube action and white response.

Black on roll, cube action and white response.

Quote

Page 5 of 13

First

1 2 3 4 5 6 7 8 9 10

Last

Post Reply Subscribe

...

Page 5 of 13

First

1 2 3 4 5 6 7 8 9 10

Last