Which double shot = double whopper?

[Axel Reichert]

GNU Backgammon Position ID: /+wAAAX7xhCEEA
Match ID : UQllASAACAAE
+24-23-22-21-20-19------18-17-16-15-14-13-+ O: gnubg
| O X O O X | | X | 2 points
| O O O | | |
| O O | | |
| O | | |
| 8 | | |
| |BAR| |v 11 point match
| X | | |
| X | | |
| X | | |
| X X X X | | | Rolled 21
| X X X O O X | | X | 1 point
+-1--2--3--4--5--6-------7--8--9-10-11-12-+ X: axel (Cube: 2)
Pip counts: O 67, X 97

Hint (again): It's not the match score.

Best regards

Axel

[Stick Rice]

What a wreck of a roll. Just hit two and go from there.

Stick

[MK]

On March 10, 2023 at 1:01:00 PM UTC-7, Axel Reichert wrote:

> GNU Backgammon Position ID: /+wAAAX7xhCEEA
> Match ID : UQllASAACAAE

> O: gnubg
> O 2 points
> 11 point match
> Rolled 21
> X 1 point
> X: axel (Cube: 2)

> Hint (again): It's not the match score.

Another one with the same hit already..? Allow me,
then, to repeat my reply to your first example that
was not about the match score either:

Why the need for the hint?

Are you hinting at my recent posts about humans
and/or bots adjusting their checker decisions to
the match score?

Do you think you could post a position from an 11
point match, with either side owning the cube at 2,
where the checker move would vary on the score?

(Preferably use the above position and just change
the scores as you wish but a similar one or even a
completely will be acceptable also.)

I know you don't work for me and you probably have
been Waltized by the clergy but I fugured it may be
worth a try to dare and expose you in front of your
arithmetician ilk... ;)

MK

[MK]

On March 11, 2023 at 1:36:02 AM UTC-7, MK wrote:

> On March 10, 2023 at 1:01:00 PM UTC-7, Axel Reichert wrote:

>> GNU Backgammon Position ID: /+wAAAX7xhCEEA
>> Match ID : UQllASAACAAE
>> O: gnubg
>> O 2 points
>> 11 point match
>> Rolled 21
>> X 1 point
>> X: axel (Cube: 2)

>> Hint (again): It's not the match score.

> Another one with the same hit already..? Allow
> me, then, to repeat my reply to your first example
> that was not about the match score either:

> Why the need for the hint?

What I said to clarify/correct myself about your other
example applies to this one also. You don't need to
repeat your response but I'll repeat my reply to it just
for the record.

The score was not lop-sided, (nor the cube was high),
and because of what was on my mind when I read it,
I took it as alluding to my recent posts about match
score, etc.

While thinking of two different things at once, I totally
confused what I wrote and exaggerated in my daring
you, a little. :)

I should have at least stuck to my objection to Tim and
Bob in another thread that the position itself should not
the variant and not go beyond saying "a similar position".
Otherwise, of course, anything is possible...

Your hint that got me going was about score. So I should
have limited my challenge to position and score only. You
may do a retake based on this or just for get about it...

MK

[Timothy Chow]

On 3/10/2023 3:00 PM, Axel Reichert wrote:
> GNU Backgammon Position ID: /+wAAAX7xhCEEA
> Match ID : UQllASAACAAE
> +24-23-22-21-20-19------18-17-16-15-14-13-+ O: gnubg
> | O X O O X | | X | 2 points
> | O O O | | |
> | O O | | |
> | O | | |
> | 8 | | |
> | |BAR| |v 11 point match
> | X | | |
> | X | | |
> | X | | |
> | X X X X | | | Rolled 21
> | X X X O O X | | X | 1 point
> +-1--2--3--4--5--6-------7--8--9-10-11-12-+ X: axel (Cube: 2)
> Pip counts: O 67, X 97
>

We're way behind in the pip count and the opponent's board is
weak, so we should be looking for bold plays that gain a lot if
we're missed. Seems like we should hit. A pick-and-pass play
leaves fewer shots but kills a checker. I think I'd try the
double hit instead.

---
Tim Chow

[Axel Reichert]

Axel Reichert <ma...@axel-reichert.de> writes:

> GNU Backgammon Position ID: /+wAAAX7xhCEEA
> Match ID : UQllASAACAAE
> +24-23-22-21-20-19------18-17-16-15-14-13-+ O: gnubg
> | O X O O X | | X | 2 points
> | O O O | | |
> | O O | | |
> | O | | |
> | 8 | | |
> | |BAR| |v 11 point match
> | X | | |
> | X | | |
> | X | | |
> | X X X X | | | Rolled 21
> | X X X O O X | | X | 1 point
> +-1--2--3--4--5--6-------7--8--9-10-11-12-+ X: axel (Cube: 2)
> Pip counts: O 67, X 97

The banana split is correct. Moving up the checker exposed to a double
direct shot and unstacking point 2 is close. Pick and pass is close to a
double whopper.

1. Rollout 6/5* 6/4* Eq.: -0.082
0.435 0.052 0.000 - 0.565 0.160 0.004 CL -0.244 CF -0.082
[0.001 0.001 0.000 - 0.001 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
867 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
2. Rollout 10/8 2/1 Eq.: -0.099 (-0.017)
0.409 0.051 0.001 - 0.591 0.122 0.001 CL -0.255 CF -0.099
[0.002 0.001 0.000 - 0.002 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
829 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
3. Rollout 19/18 10/8 Eq.: -0.125 (-0.043)
0.402 0.052 0.001 - 0.598 0.134 0.001 CL -0.282 CF -0.125
[0.002 0.001 0.000 - 0.002 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
989 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
4. Rollout 10/7 Eq.: -0.139 (-0.058)
0.392 0.041 0.000 - 0.608 0.115 0.001 CL -0.292 CF -0.139
[0.002 0.001 0.000 - 0.002 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
742 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
5. Rollout 15/14 10/8 Eq.: -0.190 (-0.108)
0.383 0.047 0.001 - 0.617 0.146 0.001 CL -0.336 CF -0.190
[0.002 0.001 0.000 - 0.002 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
750 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
6. Rollout 19/17 10/9 Eq.: -0.229 (-0.148)
0.382 0.053 0.001 - 0.618 0.187 0.002 CL -0.376 CF -0.229
[0.002 0.001 0.000 - 0.002 0.001 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
1030 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune
7. Rollout 6/5*/3 Eq.: -0.267 (-0.186)
0.366 0.032 0.000 - 0.634 0.165 0.003 CL -0.410 CF -0.267
[0.002 0.001 0.000 - 0.002 0.002 0.000 CL 0.004 CF 0.005]
Truncated cubeful rollout (depth 7) with variance reduction
696 games, Mersenne Twister dice gen. with seed 449002682 and quasi-random dice
Stop when std.errs. are small enough: limit 0.0050 (min. 36 games)
Play: 1-ply cubeful prune
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Cube: 1-ply cubeful prune
Different evaluations after 2 plies:
Play: 0-ply cubeful [expert]
Cube: 1-ply cubeful prune

[Stick Rice]

I would recommend strengthening your rollout parameters. 0 ply or 1 ply just won't cut it. The second best play here is worse by ~.045.

Stick

[Axel Reichert]

Stick Rice <bananab...@gmail.com> writes:

> I would recommend strengthening your rollout parameters. 0 ply or 1
> ply just won't cut it. The second best play here is worse by ~.045.

Quite a difference, thanks for the reminder. If I recall correctly,
these settings are roughly the equivalent of XG's Roller+ or
Roller++. Normally this is good enough for my level of play.

Best regards

Axel