5a 5a cube decision

[Walt]

XGID=--ABBbDBB---bB-----ccbab--:0:0:1:00:1:0:0:5:10

X:You O:O
Score is X:1 O:0 5 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O O O O |
| X | | O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| O X X | | X O X X |
| O X X | | X O X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 96 O: 114 X-O: 1-0/5
Cube: 1
X on roll, cube action

Do you double as X? Do you take as O?

//Walt

[badgolferman]

Double/Pass

[Paul]

"Do you take as O?" is surely a beginner's question. [Not criticising you for posing the question though, because it's conventional to always ask about both sides for any cube action problem.] Very obvious yes. X has great difficulty moving past O's two points and there is no reason at all to give this one up. The race is still fairly close.

The cubing decision is very tricky. There are some market losers. For money, I think gnu cubes this type of position and I certainly wouldn't argue.

The tricky thing is that we don't know the score. The title of your post is not consistent with the score on the diagram.

What would Humpty Dumpty say about that?

If it's really 5a 5a, I see no reason to deviate from money so double/take.

If X is actually ahead 1-0, then X should be a bit conservative. Since the double is ultra-marginal (if correct at all), I'd argue for a hold when leading 4a 5a.

Paul Epstein

[Walt]

On 8/21/2012 3:13 PM, Paul wrote:

>
> The tricky thing is that we don't know the score. The title of your post is not consistent with the score on the diagram.

You are correct - I got the title wrong.

Please respond to the diagram, not the title, since that's what actually
happened in the match.

But I'll provide a rollout for both when the time comes.

> What would Humpty Dumpty say about that?

"All the King's horses and all the King's men can't fix the title of a
usenet post once you hit send."

>


--
//Walt

[Bradley K. Sherman]

Walt <walt_...@yahoo.com> wrote:
>Score is X:1 O:0 5 pt.(s) match.
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X | | O O O O O |
> | X | | O O O O |
> | | | O O |
> | | | |
> | | | |
> | |BAR| |
> | | | |
> | | | X |
> | | | X |
> | O X X | | X O X X |
> | O X X | | X O X X X |
> +12-11-10--9--8--7-------6--5--4--3--2--1-+
>Pip count X: 96 O: 114 X-O: 1-0/5

>X on roll, cube action

Double/drop.

--bks

[Stick]

Score makes all the difference here. One should be able to accurately
estimate wins and net gammons for such a reference type position.
With this information if you understand how the score affects take
points and gammon values the decisions become trivial.

Stick

[Walt]

Yep, it's just as easy as playing the flute: you just blow in one end
and move your fingers around on the outside to choose which notes to
play. What could be easier?



--
//Walt

[Walt]

On 8/21/2012 1:59 PM, Walt wrote:

To expand on Stick's observation, this is a reference-like position
where X is about 75% to win and gammons are few.

At 0-0 to 5, the raw takepoint is 7/30 or 23.3%
At 1-0 to 5, the raw takepoint is 10/32 or 31.3%

So it's a pass if you go by the diagram and a take if you go by the
original title.

I'm struggling a bit with applying the lice cube takepoints. At 5a 5a
it's 18%, so it makes sense that it's a take. At 4a 5a the live cube
takepoint is 21% for the trailer, so this would seem to be a take. Do
the 2% gammons matter that much?

The cube is clear for both scores, although I failed to cube otb.

-Paul

//////////// 4a 5a /////////////////////////////

XGID=--ABBbDBB---bB-----ccbab--:0:0:1:00:1:0:0:5:10

X:Player 1 O:Player 2

Score is X:1 O:0 5 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O O O O |
| X | | O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| O X X | | X O X X |
| O X X | | X O X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 96 O: 114 X-O: 1-0/5
Cube: 1
X on roll, cube action

Analyzed in Rollout
No double
Player Winning Chances: 75.42% (G:1.97% B:0.06%)
Opponent Winning Chances: 24.58% (G:1.19% B:0.04%)
Double/Take
Player Winning Chances: 75.46% (G:1.94% B:0.07%)
Opponent Winning Chances: 24.54% (G:1.20% B:0.04%)

Cubeless Equities: No Double=+0.522, Double=+1.274

Cubeful Equities:
No double: +0.950 (-0.050)
Double/Take: +1.086 (+0.086)
Double/Pass: +1.000

Best Cube action: Double / Pass

Rollout:
1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: � 0.009 (+0.941..+0.959)
Confidence Double: � 0.010 (+1.076..+1.096)

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

Duration: 13 minutes 31 seconds

eXtreme Gammon Version: 2.03, MET: Kazaross XG2


/////////////// 5a 5a //////////////////////////////


XGID=--ABBbDBB---bB-----ccbab--:0:0:1:00:0:0:0:5:10

X:Player 1 O:Player 2
Score is X:0 O:0 5 pt.(s) match.

+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O O O O |
| X | | O O O O |
| | | O O |
| | | |
| | | |
| |BAR| |
| | | |
| | | X |
| | | X |
| O X X | | X O X X |
| O X X | | X O X X X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+

Pip count X: 96 O: 114 X-O: 0-0/5

Cube: 1
X on roll, cube action

Analyzed in Rollout
No double
Player Winning Chances: 75.31% (G:2.06% B:0.05%)
Opponent Winning Chances: 24.69% (G:1.30% B:0.04%)
Double/Take
Player Winning Chances: 75.51% (G:1.89% B:0.05%)
Opponent Winning Chances: 24.49% (G:1.20% B:0.04%)

Cubeless Equities: No Double=+0.513, Double=+0.994

Cubeful Equities:
No double: +0.794 (-0.065)
Double/Take: +0.859
Double/Pass: +1.000 (+0.141)

Best Cube action: Double / Take

Rollout:
1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
Confidence No Double: � 0.007 (+0.787..+0.802)
Confidence Double: � 0.008 (+0.851..+0.867)

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

Duration: 13 minutes 53 seconds

eXtreme Gammon Version: 2.03, MET: Kazaross XG2



--
//Walt

[Tim Chow]

On Aug 22, 6:05 pm, Walt <walt_ask...@yahoo.com> wrote:
> I'm struggling a bit with applying the lice cube takepoints. At 5a 5a
> it's 18%, so it makes sense that it's a take.  At 4a 5a the live cube
> takepoint is 21% for the trailer, so this would seem to be a take.  Do
> the 2% gammons matter that much?

Somehow I missed this one when it was first posted.

The raw take point is (35-25)/(57-25) ~ 30% according to the MET I
know. I'm not sure where you're getting 21% from. The trailer has
some recube vig but wins very few gammons so it shouldn't be
surprising that she can't take this cube.

---
Tim Chow

[Walt]

Agree that the raw takepoint is 10/32 ~ 31%

The "live" (not "lice" as I wrote) takepoint of 21% is from

http://bigwillgammon.site90.com/bgcardv2.pdf

I'm not sure how the author calculated it, but it must include
assumptions about volitility and gammon rates that don't apply here.



--
//Walt

[Tim Chow]

On Aug 23, 9:44 am, Walt <walt_ask...@yahoo.com> wrote:
> The "live" (not "lice" as I wrote) takepoint of 21% is from
>
> http://bigwillgammon.site90.com/bgcardv2.pdf
>
> I'm not sure how the author calculated it, but it must include
> assumptions about volitility and gammon rates that don't apply here.

I never know what people mean by the live cube take point. Seems like
there are many different definitions floating around.

I'd rather think in terms of how many gammons the taker can win and
what the taker is going to do with the cube later in the game. In a
textbook 5pt holding game like this one, we know that the taker is not
going to win many gammons, and I can't believe that the cube ownership
would drive the takepoint down ten percentage points.

---
Tim Chow

[Michael Petch]

On 2012-08-23 17:47, Tim Chow wrote:
> I'd rather think in terms of how many gammons the taker can win and
> what the taker is going to do with the cube later in the game. In a
> textbook 5pt holding game like this one, we know that the taker is not
> going to win many gammons, and I can't believe that the cube ownership
> would drive the takepoint down ten percentage points.

Earlier in the day I contacted the author of that cheat sheet (a friend
of mine) to find out how that number was arrived at. Hopefully if he has
time he'll respond.

[bigwil...@gmail.com]

I did not calculate them at all, I used the XG Cube Information to find the take points. I went back and double checked to be sure there was not a typo on my table. XG Gives the trailer live cube take point for 5 away 4 away, non gammon adjusted, as 21.3%. The table is rounded for easier reference and memorizing. The dead cube tp calculated by xg is 30.3%. These values are using the Kazaross XG2 MET.

Big Will

[Stick]

Using the dead cube take point (What Timothy seems to currently be
doing) and then trying to guesstimate how much owning the cube is
worth seems flat out the wrong way to go to me. I'd rather know the
live cube take point, as reported w/XG using its recube efficiency
(tweeking it in any way I see fit) and then simply adjust for gammons
to arrive at my fully adjusted take point.

Stick

[Walt]

On 8/23/2012 10:53 PM, bigwil...@gmail.com wrote:
> On Thursday, August 23, 2012 6:47:53 PM UTC-5, Tim Chow wrote:
>> On Aug 23, 9:44 am, Walt <walt_ask...@yahoo.com> wrote:
>>

>>> The "live" takepoint of 21% is from

>>
>>> http://bigwillgammon.site90.com/bgcardv2.pdf
>>
>>>
>>
>>> I'm not sure how the author calculated it, but it must include

> I did not calculate them at all, I used the XG Cube Information to find the take points. I went back and double checked to be sure there was not a typo on my table. XG Gives the trailer live cube take point for 5 away 4 away, non gammon adjusted, as 21.3%. The table is rounded for easier reference and memorizing. The dead cube tp calculated by xg is 30.3%. These values are using the Kazaross XG2 MET.

I have XG2 and it says the live cube takepoint is 22.16 AtS, which is
close enough to your value of 21.3 for my purposes.

What I don't understand is how it's a pass when the rollout says O wins
~24% of games.

Note that XG2 4-ply analysis says it's a take. There's not much
difference in the Win/Gammon/Backgammon rates or the cubeless equities
between the two, but a rather large difference in the cubeful equities.
Why is that?

I feel I'm missing something here, and that if I can fill it in it will
greatly help my understanding.


Analyzed in 4-ply
Player Winning Chances: 75.32% (G:2.16% B:0.01%)
Opponent Winning Chances: 24.68% (G:1.25% B:0.03%)

Cubeless Equities: No Double=+0.522, Double=+1.273

Cubeful Equities:
No double: +0.855 (-0.111)

Double/Take: +0.967
Double/Pass: +1.000 (+0.033)

Best Cube action: Double / Take

Analyzed in Rollout
No double
Player Winning Chances: 75.42% (G:1.97% B:0.06%)
Opponent Winning Chances: 24.58% (G:1.19% B:0.04%)
Double/Take
Player Winning Chances: 75.46% (G:1.94% B:0.07%)
Opponent Winning Chances: 24.54% (G:1.20% B:0.04%)

Cubeless Equities: No Double=+0.522, Double=+1.274

Cubeful Equities:
No double: +0.950 (-0.050)
Double/Take: +1.086 (+0.086)
Double/Pass: +1.000

Best Cube action: Double / Pass

--
//Walt

[Stick]

XG figures in too good of a recube efficiency. The live cube take
points in general, esp. from 5a 5a and in, should be adjusted. I have
done this on my own but if you simply add ~1 - 1.5% to its live cube
numbers you'll be closer to the truth. According to my numbers the
actual live cube take point at 5a 4a is 24.5%.

Stick

[Tim Chow]

The only difference I can see between what you're doing and what you
say is "flat out the wrong way to go" is that you're not attempting to
adjust the live cube takepoint according to the nature of the
particular position. Right?

If that's the case then I don't see why it's flat out wrong to adjust
for the particular position.
---
Tim Chow

[Stick]

No, this is not how I see it. I know going in what the cube is
generally worth, you do not. You see a fully dead take point and then
guess somehow what the cube is worth? I may not be making myself
clear but these two methods seem as different as night and day to
me.

Stick

[Tim Chow]

On Aug 26, 10:24 pm, Stick <checkmug...@yahoo.com> wrote:
> No, this is not how I see it.  I know going in what the cube is
> generally worth, you do not.  You see a fully dead take point and then
> guess somehow what the cube is worth?  I may not be making myself
> clear but these two methods seem as different as night and day to
> me.

It's true that memory beats intelligence every time. It's certainly
better to memorize a single number for the recube value, based on the
collective experience of the BG community, than to try to guess what
it is OTB.

However, it doesn't follow that what I was trying to do was "flat out
the wrong way to go." Let me try an analogy.

Suppose John Doe approaches positions by memorizing a single number
for each score (and cube location) that he calls the "gammon-adjusted
takepoint." (Let's ignore recube vig for the moment, for
simplicity.) He makes no effort to judge how many gammons are won or
lost in any particular position, but simply tries to figure out
whether he can win as often as the "gammon-adjusted takepoint" tells
him he needs to win.

Now suppose Jane Roe approaches positions by memorizing the raw
takepoint, then trying to estimate gammons won on either side, and
adjusting the estimated wins by the estimated gammons, weighted by the
gammon values. Jane, however, happens not to have memorized the
correct gammon values.

Is Jane going about things in "flat out the wrong way"? I'd say no.
In fact, I'd say that Jane is going about things the right way. Yes,
*of course* Jane ought to go off and memorize the correct gammon
values. However, that doesn't mean that her approach is fundamentally
unsound.

Getting back to recube vig, one approach is to memorize a single
number that allows you to compute what we'll call the "live cube take
point." (Let's ignore gammons for the moment, for simplicity.) You
then make no effort to judge how often the cube will be sent back, or
taken/dropped if sent back, in any particular position. Instead, you
just judge whether you win as often as the "live cube take point"
tells you you need to. This is basically John Doe's approach. It'll
do just fine most of the time, just as John Doe's approach to gammons
will do fine most of the time. But, I would argue that the Jane Roe
approach is ultimately better: Try to judge how often the cube will be
sent back and taken/dropped, recognize that redouble/take/win single
is mathematically equivalent to a gammon win, and use the gammon price
to make the appropriate adjustment.

And *of course* Jane Roe ought to learn the conventional wisdom about
how often the cube gets sent back and taken/dropped ATS. In this
case, though, it's not just a matter of her looking up a table and
memorizing the numbers, because the conventional approach in the BG
community doesn't present the numbers in this fashion. I'm convinced
that Jane Roe's method is the right way to go, but I admit I'm too
lazy to actually construct the relevant tables, mostly because the
bots are not set up to provide the relevant information very easily.
---
Tim Chow