 |
| |
 |
| |
BG Quiz 30 | The Weekly Quiz #30
13
November 2002
by Paul Money
Author's Introduction: Welcome quiz fans, another selection
of five interesting positions that good players got wrong in live
play. Take a couple of minutes on each one and pick the play you
would make over the board in an important match. No prizes, but
something to learn for everyone. Starting with Quiz 30, rollouts
are now done with Snowie 4 Pro. Good luck!
Instructions: Have a pen and paper handy to keep track of
your choices, then see how you scored under "Comments &
Rollouts" where Paul also provides his perspective and a rollout by
Snowie.
The arrow on the right side of the board indicates which player's
turn it is to move or double. The match length, score and pip count
are there as well. Please post your comments on the strategy in the
forum thread at the end of the article. This quiz is published every
Wednesday. Enjoy!
The Weekly Quiz
#30
Problem 1:
Problem 2:
Problem 3:
Problem 4:
Problem 5: Cube Action?
Comments &
Rollouts
Problem 1:
Green has just doubled and got White to make
a terrible take. Now he must press on and extract the maximum
advantage from the situation and that means hitting and going for
the gammon, 13/7, 6/1*. The move of 24/13 will, as you might
suspect, win the most games, but it's gammons that win matches and
putting a second checker on the roof is mandatory. This is often
right, even when it means leaving two blots inboard, so it's quite
clear here. There is a great tendency to play for safety after
doubling; it is usually the reverse of correct strategy. Aggressive
backgammon wins matches.
You Score:
13/7, 6/1* = 2 points
Anything else = 0 points.
| # |
Ply |
Move |
Equity |
| |
1 |
R |
13/7 6/1* |
0.666 |
|
|
|
2.1% 38.0% 71.1% 28.9%
6.5% 0.3% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.754
±0.026.
Rollout settings:
Full rollout,
324 games
(equiv. 10029 games),
played 2-ply,
seed 1, with race
database.
|
| |
2 |
R |
24/18 13/8 |
0.599
(-0.067) |
|
|
|
2.6% 29.5% 71.5% 28.5%
5.9% 0.3% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.691
±0.024.
Rollout settings:
Full rollout,
324 games
(equiv. 9254 games),
played 2-ply,
seed 1, with race
database.
|
| |
3 |
R |
24/13 |
0.598
(-0.069) |
|
|
|
1.6% 27.1% 72.6% 27.4%
5.1% 0.3% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.684
±0.021.
Rollout settings:
Full rollout,
324 games
(equiv. 10528 games),
played 2-ply,
seed 1, with race
database.
|
| |
4 |
R |
24/18 6/1* |
0.589
(-0.077) |
|
|
|
1.6% 32.5% 70.2% 29.8%
5.9% 0.3% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.682
±0.027.
Rollout settings:
Full rollout,
324 games
(equiv. 9277 games),
played 2-ply,
seed 1, with race
database.
|
| |
5 |
M |
13/7 13/8 |
0.564
(-0.102) |
|
|
|
2.0% 31.6% 69.9% 30.1%
7.2% 0.5% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.658
±0.011.
Rollout settings:
Truncated rollout, depth
7,
108 games (equiv. 45360 games),
played 2-ply
(fast),
seed 1, with race
database.
|
Problem 2:
Don't be too disappointed if you blundered
here. This is a very tough problem indeed. The key is to see that
Green must deny White freedom of movement on her next turn. If Green
doesn't hit, then all White's numbers play well. The move of 24/23,
3/1* seems to be the best way to do this. Although Green doesn't get
to the edge of White's prime, that isn't necessarily fatal. Only
nine numbers will break Green's own prime if White fans. After this
play, Green wins 25% of the games, an astonishingly high figure. A
nice feature is that all Green's wins will look "lucky" (very
irritating for White) and some of them will be match-winning
gammons.
You Score:
24/23, 3/1* = 2 points
24/22, 2/1* = 1 point
Anything else = 0 points.
| # |
Ply |
Move |
Equity |
| |
1 |
R |
24/23 3/1* |
-0.354 |
|
|
|
0.0%
6.9% 25.0% 75.0% 21.5%
1.2% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.658
±0.018.
Rollout settings:
Full rollout,
324 games
(equiv. 17937 games),
played 2-ply,
seed 1, with race
database.
|
| |
2 |
R |
24/22 2/1* |
-0.397
(-0.043) |
|
|
|
0.1%
6.8% 22.9% 77.1% 28.0%
2.3% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.775
±0.021.
Rollout settings:
Full rollout,
324 games
(equiv. 13364 games),
played 2-ply,
seed 1, with race
database.
|
| |
3 |
R |
24/22 24/23 |
-0.438
(-0.084) |
|
|
|
0.2%
5.7% 22.1% 77.9% 18.0%
0.8% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.688
±0.021.
Rollout settings:
Full rollout,
324 games
(equiv. 11775 games),
played 2-ply,
seed 1, with race
database.
|
| |
4 |
R |
3/1* 3/2 |
-0.441
(-0.087) |
|
|
|
0.1%
6.1% 21.5% 78.5% 20.9%
1.2% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.729
±0.018.
Rollout settings:
Full rollout,
324 games
(equiv. 14237 games),
played 2-ply,
seed 1, with race
database.
|
| |
5 |
R |
6/5 3/1* |
-0.441
(-0.087) |
|
|
|
0.1%
5.9% 21.7% 78.3% 20.4%
1.2% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.723
±0.017.
Rollout settings:
Full rollout,
324 games
(equiv. 17627 games),
played 2-ply,
seed 1, with race
database.
|
Problem 3:
Several conflicting plans present themselves
to Green here. Moving 6/3, 5/3 makes the six prime. The 8/3 move
slots the valuable three point and stays even on the highest points.
Moving 8/5, 6/4 also stays even, without directly doing anything to
make the three point. However, when bearing in against an anchor,
nothing beats counting the shots next turn. Move 8/3 leaves an
immediate shot of course and all of them leave eight or nine shots
next turn. However, the ticket is 7/4, 7/5, leaving an ideal no
shots next turn, clearing a point, which has to be cleared anyway,
and killing sixes, except 6-1. It does create a second gap, but it
is a relatively innocuous one and the four active builders ready to
make the three point amply compensate. Count the shots! If you don't
do it, practice.
You Score:
7/4, 7/5 = 2 points
8/3 or 6/3, 5/3 = 0 points
Anything else = -1 point.
| # |
Ply |
Move |
Equity |
| |
1 |
R |
7/4 7/5 |
0.630 |
|
|
|
0.2%
9.4% 80.4% 19.6%
0.8% 0.0% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.696
±0.013.
Rollout settings:
Full rollout,
324 games
(equiv. 17602 games),
played 2-ply,
seed 1, with race
database.
|
| |
2 |
R |
8/3 |
0.578
(-0.052) |
|
|
|
0.3% 10.0% 78.5% 21.5%
2.0% 0.1% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.651
±0.014.
Rollout settings:
Full rollout,
324 games
(equiv. 19900 games),
played 2-ply,
seed 1, with race
database.
|
| |
3 |
R |
6/3 5/3 |
0.572
(-0.058) |
|
|
|
0.1%
8.4% 78.8% 21.2%
1.8% 0.1% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.642
±0.013.
Rollout settings:
Full rollout,
324 games
(equiv. 19798 games),
played 2-ply,
seed 1, with race
database.
|
| |
4 |
M |
8/6 7/4 |
0.514
(-0.116) |
|
|
|
0.6% 15.4% 73.6% 26.4%
2.8% 0.1% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.603
±0.024.
Rollout settings:
Truncated rollout, depth
7,
108 games (equiv. 10812 games),
played 2-ply
(fast),
seed 1, with race database.
|
| |
5 |
M |
8/5 6/4 |
0.428
(-0.201) |
|
|
|
0.3%
9.4% 72.9% 27.1%
3.4% 0.1% |
|
|
|
95%
confidence interval:
- money cubeless eq.: 0.520
±0.016.
Rollout settings:
Truncated rollout, depth
7,
108 games (equiv. 15264 games),
played 2-ply
(fast),
seed 1, with race
database.
|
Problem 4:
A very good player chose 23/20, 6/4 in this
position, I hope that you didn't. White will attack the blot any way
that she can, after she has doubled. Green must wait patiently with
13/8. Of course the straggler will still have to tunnel out somehow,
but at least he won't be on the bar and Green may be able to
strengthen his board and give a good account of himself later. Even
23/20, 13/11 does better than the move made by the player,
presumably because Green at least gets the chance to anchor
somewhere.
You Score:
13/8 = 2 points
23/20, 13/11 = 0 points
Anything else = -1 point.
| # |
Ply |
Move |
Equity |
| |
1 |
R |
13/8 |
-0.648 |
|
|
|
0.2%
5.7% 34.2% 65.8% 18.4%
0.4% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.446
±0.019.
Rollout settings:
Full rollout,
324 games
(equiv. 13248 games),
played 2-ply,
seed 1, with race
database.
|
| |
2 |
R |
23/20 13/11 |
-0.746
(-0.098) |
|
|
|
0.2%
7.4% 34.3% 65.7% 25.0%
0.6% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.494
±0.024.
Rollout settings:
Full rollout,
324 games
(equiv. 11072 games),
played 2-ply,
seed 1, with race
database.
|
| |
3 |
R |
23/20 6/4 |
-0.850
(-0.202) |
|
|
|
0.3%
7.6% 32.9% 67.1% 27.3%
0.6% |
|
|
|
95%
confidence interval:
- money cubeless eq.: -0.542
±0.022.
Rollout settings:
Full rollout,
324 games
(equiv. 11180 games),
played 2-ply,
seed 1, with race
database.
|
| |
4 |
2 |
8/5 8/6 |
-0.868
(-0.221) |
|
|
|
0.3%
7.1% 31.5% 68.5% 24.1%
0.8% |
| |
5 |
1 |
13/10 6/4 |
-0.862
(-0.214) |
|
|
|
0.3%
8.0% 32.2% 67.8% 26.2%
1.2% |
Problem 5: Cube Action?
A powerful double, clear take for White. Many
good players would refrain from doubling here, but White's position
is not particularly strong. The combination of an anchor and an
outfield point ties up too many checkers in defence. The points
don't work together to give the attacker problem numbers, in the way
that a good backgame does. Also the defender is sometimes forced to
run from one of the points at a difficult moment. In this particular
position, most of White's equity derives from Green's slightly
awkward checker placement; note the stack on the six point and the
two men out of play on the ace. If Green had better distribution,
the take would be marginal at best.
You Score:
Double/take = 2 points
No double/take = 1 point
Double/pass = 0 points
No double/pass = -1 point.
| Cube
action equity |
| Rollout |
Money
equity: |
0.532 |
| |
0.3% 12.8% 72.3% 27.7%
4.3% 0.1%
|
| |
95%
confidence interval:
- money cubeless eq.: 0.532
±0.021.
Rollout settings:
Full rollout,
324 games
(equiv. 12810 games),
played 2-ply,
seed 1, with race
database.
|
| 1. |
Double, take |
0.900 |
|
| 2. |
No double |
0.833 |
(-0.066) |
| 3. |
Double, pass |
1.000 |
(+0.100) |
| Proper cube
action: Double,
take |
|
Conclusion: And how did you do? 10/10 is quite an
achievement, a clear world class result. 9 or 8 is certainly not to
be sneezed at, an expert result and even 7 or 6 is an advanced mark.
Please take a moment to rate the column 1 to 5 and post a note to
let us know how you did. Thanks for playing, look forward to seeing
you next week, enjoy the game!
7 users have given this article an average rating of 5.0 out of
5.
Feedback about this article:
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
14 Nov 2002 11:26 EST |
Practicing.... In problem 3, I only count 6 shot numbers
after 6/3 5/3 (54, 65, 55 and 66). Rolls like 33 and 44
aren't pretty, but they aren't shots. Am I missing
something obvious?
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
14 Nov 2002 15:45 EST |
Was Problem #2 a Crawford game? If yes then the rollout
results can't be trusted until the new patch comes out.
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
14 Nov 2002 20:14 EST |
It's essentially Crawford 2-away. I don't think that
triggers the bug you are thinking of, but it makes cubeless
money play rollouts suspect since the gammon prices of 0
and 1+ are very different from the 0.5 for both sides of
cubeless money play.
Douglas Zare
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
15 Nov 2002 02:43 EST |
Ermm, in problem 3, there are nine shot numbers after
playing 6/3, 5/3. They are 55, 65, 54, 53 and 51. More
practice DB!
Problem 2 is post crawford, White has the cube. I am not
aware of the bug to which you refer. I will investigate
this position further, possibly with good ol' snowie 3.
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
15 Nov 2002 08:56 EST |
snowie 3 produces broadly similar results for problem 2. I
think the roll-outs supports the analysis quite well, the
plays to win the game are usually the best plays to win the
gammon and I believe that the hit and split plays are
clearly best.
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
16 Nov 2002 00:52 EST |
well, er, not exactly. 5-1, and 5-3 don't leave shots,
although they are unpleasant.
| Subject: |
Re: The Weekly Quiz #30 |
| From: |
|
| Date: |
16 Nov 2002 09:12 EST |
thx Dave, apologies to DB, obviously I have a recurring
blind spot when looking at this position. You are quite
correct, there are only 5 immediate blotting numbers if you
make the three point.
|
|
