Double from the bar?

[Tim Chow]

XGID=aaB-BBBBA---c--C-abcb-b--A:0:0:1:00:0:0:3:0:10
X:Player 2 O:Player 1

Score is X:0 O:0. Money session, Jacoby
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O O | | O O O |
| X O | | O O O |
| X | | O |
| | | |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | |
| O X | | X X X X |
| O X X | | X X X X O |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pipcount X: 126 O: 144 X-O: 0-0
Cube : 1
X on roll, cube action

---
Tim Chow

[Bradley K. Sherman]

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

>Pipcount X: 126 O: 144,Cube:1, X on roll, cube action

Looks like a double/pass to me.

--bks

[Paul]

As is clear from the title of the post, the problem is all about the
double, with the take being trivial. O has plenty of counterplay
against X's blot and X doesn't have particularly large gammon
threats. Surely it's a take (if it's a double).
From playing gnu, I've learned that, if you do have an advantage (and
X is clearly ahead), you don't need that many market losers for an
initial double. If X enters and O dances, X has lost market. That
parlay actually happens 1/3 of the time. Also 62 and 44 for X is
likely to be devastating for O for most of O's replies.

That's enough market losers for an initial double.

Double/take.

I'm trying to earn some more points by taking on a related but unasked
problem.
If X owned the cube, should X redouble? My initial reaction was no.
However, when I actually count the market losers (as above), there are
more of them than I expected. I'm very undecided about the redouble
problem. I'll go with redouble/take as well but I won't be at all
surprised if that's wrong.

Summary:

Paul's opinions:
Double/take
Redouble/take

Outcomes that would significantly surprise Paul:
A pass
No initial double

Outcomes that Paul believes in but that he's very unsure about.
I'm about 52% sure that X should double if you change the position by
shifting the cube to X's side. By the way, if you do shift the cube
to X's side, you should do so quietly. X can be a bit timid and might
be alarmed if you bang down the cube on the table with maximum force.
That would make X reluctant to play again and you might also get
thrown out of the restaurant, particularly if other diners complain.

Paul Epstein

[Walt]

A thorny one. X is definitely ahead, and there are gammon chances on
both sides, with X's gammon rate higher. With Jacoby and centered cube,
that should mean that turning the cube gains more equity than at an
early match score. Whether that's enough is another matter.

O has a tough take decision, so I'm going to apply Kit's law and say
double. I'm not so sure about the take - with X on the bar against a
three point board O certainly has some winning chances. I'll say take,
but I'm not at all sure.

//Walt

[Paul]

As my post says, I don't see it as a tough take decision.
I'm going to add a numerical estimate to my assessment.

I estimate that dropping is about a 0.4 blunder.

Paul

[Tim Chow]

I didn't double here but the rollout says it's a big drop. Though X
is on the bar, he enters 3/4 of the time, and O is then in serious
trouble. X's four-point board and 11 checkers in the zone create a
strong blitzing threat with many gammons, and even if O enters, X has
an edge in the priming game. Of course X does have 9 dancing rolls
and then O will have fighting chances, but apparently it's not
enough. O does have winning chances but the gammon threat is too
much.

=======
Rollout
=======

Analyzed in Rollout
No Double
Player Winning Chances: 71.11% (G: 33.84% B: 1.92%)
Opponent Winning Chances: 28.89% (G: 7.28% B: 0.28%)
Double/Take
Player Winning Chances: 71.30% (G: 36.13% B: 1.78%)
Opponent Winning Chances: 28.70% (G: 7.52% B: 0.38%)

Cubeless Equities: No Double=+0.422, Double=+1.452

Cubeful Equities:
No Double: +0.839 (-0.161)
Double/Take: +1.215 (+0.215)
Double/Drop: +1.000

Best Cube action: Double / Drop

Rollout
1296 Games rolled with Variance Reduction.
Dice Seed: 2
First 4 moves and cube decisions: 4 ply
Remaining moves and cube decisions: 3 ply
Confidence No Double: ±0.020 (+0.819<E<+0.860)
Confidence Double: ±0.028 (+1.187<E<+1.244)

eXtreme Gammon Version: 1.21

---
Tim Chow

[Paul]

Good problem. I do think that O's blot on X's 17 point has a massive
negative impact on O's position. Since O is trying to enter, it's
quite difficult for O to safety that blot and X hits it immediately 8%
of the time.

Do you think it's a take if we Chow the position by moving the 17 blot
to the 18 point? I do. It's difficult to estimate how often X hits
that blot. Probably most of the hits happen in the next 2 or 3 rolls
if O struggles to enter.

I'll estimate 25% of the time. Since these hits are really crushing
and not at all rare, eliminating them by Chowing in the above manner
may well swing the .21 equity to create a take. Perhaps even an easy
take.

Paul Epstein

[Tim Chow]

On May 23, 7:13 am, Paul <pepste...@gmail.com> wrote:
> Do you think it's a take if we Chow the position by moving the 17 blot
> to the 18 point?  I do.

Good question. I haven't looked at this variation with the computer
but will try to remember to do so. In the meantime, I agree that
cleaning up that blot will improve O's position a lot. But knowing
how big a drop the original position is, I suspect that it will still
be a drop---a closer one, to be sure, but still a drop.

---
Tim Chow

[Tim Chow]

On May 23, 7:13 am, Paul <pepste...@gmail.com> wrote:

> Good problem.  I do think that O's blot on X's 17 point has a massive
> negative impact on O's position.  Since O is trying to enter, it's
> quite difficult for O to safety that blot and X hits it immediately 8%
> of the time.
>
> Do you think it's a take if we Chow the position by moving the 17 blot
> to the 18 point?  I do.

O.K., I took a look at this. Cleaning up the blot does boost O's
equity significantly, but apparently not enough to make it a take.
See the truncated rollout below. I actually moved the blot to the
20pt rather than the 18pt, which I think is better for O (but
evidently still not good enough).

XGID=aaB-BBBBA---c--C--bcc-b--A:0:0:1:00:0:0:3:0:10

X:Player 2 O:Player 1
Score is X:0 O:0. Money session, Jacoby
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X O | | O O O |

| X O | | O O O |

| X | | O O |

| | | |
| | X | |
| |BAR| |
| | O | |
| | | |
| O | | |
| O X | | X X X X |
| O X X | | X X X X O |
+12-11-10--9--8--7-------6--5--4--3--2--1-+

Pipcount X: 126 O: 141 X-O: 0-0

Cube : 1
X on roll, cube action

Analyzed in Rollout
No Double

Player Winning Chances: 70.97% (G: 28.37% B: 1.08%)
Opponent Winning Chances: 29.03% (G: 7.61% B: 0.30%)
Double/Take
Player Winning Chances: 70.59% (G: 28.80% B: 1.08%)
Opponent Winning Chances: 29.41% (G: 7.55% B: 0.29%)

Cubeless Equities: No Double=+0.419, Double=+1.265

Cubeful Equities:
No Double: +0.831 (-0.169)
Double/Take: +1.062 (+0.062)

Double/Drop: +1.000

Best Cube action: Double / Drop

Rollout Truncated at 7 moves using 3 ply.
648 Games rolled with Variance Reduction.
Dice Seed: 2
First 2 moves and cube decisions: 3 ply
Remaining moves: 2 ply, cube decisions: 3 ply
Confidence No Double: ±0.009 (+0.822<E<+0.839)
Confidence Double: ±0.016 (+1.046<E<+1.077)