Endgame split 2

[Tim Chow]

XGID=-ACBBBB---------b--ccBcdA-:1:1:1:41:0:0:0:0:10

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-+
| O | | O O X O O X |
| O | | O O X O O |
| | | O O O O |
| | | O |
| | | |
| |BAR| |
| | | |
| | | |
| | | X | +---+
| | | X X X X X | | 2 |
| | | X X X X X X | +---+
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 109 O: 68 X-O: 0-0
Cube: 2, X own cube
X to play 41

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

> X:109 O:68, X to play 41

6/1

--bks

[michae...@gmail.com]

Well, I can see the benefits of making a closed board with 2/1 21/17. There are also additional benefits of maximizing chances to hit her if she rolls something like 21,62,53. The question is how many more gammons does that lose??
Impossible for me to tell. In this case I won't be surprised if this is the bots choice, but since I always play matches usually to 7 and a gammon loss with the cube at 2 usually means killing the whole match, i would play safe.Same move as Bradley's

[BlueDice]

So (a)21/17 2/1 or (b)6/1
I see that (a) duplicates all of Os pointing rolls and completes Xs board. On the down side it gives up the anchor and looks like inviting a gammon. I would chicken out and play (b)

[Tim Chow]

To be honest I still don't quite understand all the subtleties of this one.
Before you try to "explain" it, check out the variant position, where X's
spare on his 2pt has been shifted back to his 3pt. One thing this does is
to unduplicate X's aces to enter and cover if he is hit. It also means his
board is slightly less likely to be forced to crack with a bad roll.

These subtleties aside, in both the original position and the variant,
coming out does win more games. So it's a matter of estimating gammons,
which as michael said is no easy task.

1. Rollout¹ 21/17 2/1 eq:-0.740
Player: 22.80% (G:1.97% B:0.04%)
Opponent: 77.20% (G:24.97% B:0.54%)
Confidence: ±0.003 (-0.743..-0.736) - [100.0%]

2. Rollout¹ 6/1 eq:-0.803 (-0.063)
Player: 16.11% (G:1.38% B:0.03%)
Opponent: 83.89% (G:20.09% B:0.20%)
Confidence: ±0.005 (-0.808..-0.798) - [0.0%]

¹ 1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller

eXtreme Gammon Version: 2.19.208.pre-release

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

XGID=-ABCBBB---------b--ccBcdA-:1:1:1:41:0:0:0:0:10

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

Pip count X: 110 O: 68 X-O: 0-0

Cube: 2, X own cube
X to play 41

1. Rollout¹ 6/1 eq:-0.796
Player: 16.47% (G:1.43% B:0.03%)
Opponent: 83.53% (G:20.13% B:0.21%)
Confidence: ±0.005 (-0.801..-0.791) - [100.0%]

2. Rollout¹ 21/17 3/2 eq:-0.817 (-0.020)
Player: 19.98% (G:1.60% B:0.05%)
Opponent: 80.02% (G:29.18% B:0.78%)
Confidence: ±0.005 (-0.822..-0.811) - [0.0%]

3. Rollout¹ 5/1 3/2 eq:-0.817 (-0.021)
Player: 15.97% (G:1.15% B:0.02%)
Opponent: 84.03% (G:20.87% B:0.24%)
Confidence: ±0.004 (-0.821..-0.813) - [0.0%]

4. Rollout¹ 5/4 5/1 eq:-0.826 (-0.030)
Player: 14.63% (G:1.15% B:0.02%)
Opponent: 85.37% (G:18.71% B:0.15%)
Confidence: ±0.004 (-0.830..-0.822) - [0.0%]

¹ 1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller

eXtreme Gammon Version: 2.19.208.pre-release

---
Tim Chow

[bananab...@gmail.com]

It's a very easy task to estimate gammons, don't do it. Make the dmp play and get on with life. Unless there's a monstrous swing on gammons that comes out and slaps you upside the head ... dmp play, #simplegame.

As for the difference between the original position and the variant, one makes a full six point board and comes out, the other doesn't, that's a pretty huge difference.

Stick

[Tim Chow]

On Saturday, May 21, 2016 at 11:40:54 PM UTC-4, bananab...@gmail.com wrote:
> It's a very easy task to estimate gammons, don't do it. Make the dmp play and
> get on with life. Unless there's a monstrous swing on gammons that comes out
> and slaps you upside the head ... dmp play, #simplegame.

The DMP rule makes even less sense than usual in this series because the
difference in gammon losses can be large in these endgame split positions.

---
Tim Chow

[bananab...@gmail.com]

Look at your 4 positions so far there is only 1 position where there's a huge gammon difference and it's quite obvious that there's little/no gain from splitting there so unless you have more up and coming positions where there's a huge gammon difference and the play at dmp is clear it makes total sense.

In #1 there isn't a big difference in gammons and the dmp play is correct anyway.
In #2 there isn't a huge difference and the dmp play is correct anyway.
In #3 there isn't a real difference and the dmp play is correct anyway.
in #4 there _is_ a big difference and it smacks you in the face. The plays are TCTC at dmp and pretty obvious splitting is bad when gammons are involved which can be attested to by nobody yet getting it wrong or even strongly considering it.

Show me where the rule makes less sense than usual.

Stick

[Tim Chow]

On Sunday, May 22, 2016 at 8:34:37 PM UTC-4, bananab...@gmail.com wrote:
> Show me where the rule makes less sense than usual.

25% of the time, ignoring gammons gets you the wrong answer.

Are you claiming that your rule works only 75% of the time?

---
Tim Chow

[bananab...@gmail.com]

The rule is when the dmp play is CLEAR. You know this right? I've defined clear before as anywhere from 2-3% or more but as you know, it's almost always right, barring exceptions, even with a smaller dmp difference. In that problem the dmp play is far from clear, TCTC as I stated, which means with any common sense one would realize the obnoxious and obvious difference in gammons would be the deciding factor.

Stick

[Tim Chow]

On Monday, May 23, 2016 at 9:01:12 PM UTC-4, bananab...@gmail.com wrote:
> The rule is when the dmp play is CLEAR. You know this right?

I do know that you change the goalposts to suit yourself. In this particular
thread, you said, "It's a very easy task to estimate gammons, don't do it.

Make the dmp play and get on with life. Unless there's a monstrous swing on
gammons that comes out and slaps you upside the head ... dmp play, #simplegame."

Nothing here about "when the dmp play is CLEAR."

---
Tim Chow