Gammonfever BG Quiz - Problem #3 (and repost of 1&2)
Micke Nilsson
position (two decisions) has already been posted and will follow as a reply
to this post.
Problem #3, O on roll (home board top left in this and following posisions),
Cube action?
Money session. Score X-O: 0-0
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| X O | | O O O X X |
| X O | | O O O X |
| O | | O O X | S
| O | | | n
| O | | | o
| |BAR| | w
| | | | i
| | | | e
| X X | | X |
| X X | | X |
| O X X | | X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 151 O: 133 X-O: 0-0/Money (1)
CubeValue: 1
Micke Nilsson
#2) Cube action after O:s move?
Money session. Score X-O: 0-0
O to play (4 2)
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| O O O O O | | X |
| O O O O O | | X |
| | | | S
| | | | n
| | | | o
| |BAR| | w
| | | | i
| X | | O | e
| X X | | X O |
| X X X | | X O |
| O X X X | | X X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 113 O: 115 X-O: 0-0/Money (1)
CubeValue: 1
Micke Nilsson
Money session. Score X-O: 0-0
O to play (4 2)
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| O O O O O | | X |
| O O O O O | | X |
| | | | S
| | | | n
| | | | o
| |BAR| | w
| | | | i
| X | | O | e
| X X | | X O |
| X X X | | X O |
| O X X X | | X X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 113 O: 115 X-O: 0-0/Money (1)
CubeValue: 1
1. R 13/9 13/11 -0,592
Live cube rollout: -0,646
95% confidence interval:
- money cubeless eq.: -0,422 ±0,020,
- live cube: -0,646 ±0,046.
Rollout settings:
Full rollout,
324 games (equiv. 14809 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
2. R 5/1 5/3 -0,946 (-0,354)
Live cube rollout: -1,000
95% confidence interval:
- money cubeless eq.: -0,583 ±0,021,
- live cube: -1,000 ±0,000.
Rollout settings:
Full rollout,
324 games (equiv. 13722 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
3. R 13/9 3/1 -1,000 (-0,408)
Live cube rollout: -1,000
95% confidence interval:
- money cubeless eq.: -0,626 ±0,021,
- live cube: -1,000 ±0,000.
Rollout settings:
Full rollout,
324 games (equiv. 15853 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
4. R 6/2 6/4 -1,000 (-0,408)
Live cube rollout: -1,000
95% confidence interval:
- money cubeless eq.: -0,608 ±0,022,
- live cube: -1,000 ±0,000.
Rollout settings:
Full rollout,
324 games (equiv. 13422 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
"Micke Nilsson" <na...@home.se> skrev i meddelandet
news:zXjTe.145793$dP1.5...@newsc.telia.net...
Micke Nilsson
Money session. Score X-O: 0-0
X on roll, cube action
+24-23-22-21-20-19-------18-17-16-15-14-13-+
| O O O O O | | X O O |
| O O O O O | | X |
| | | | S
| | | | n
| | | | o
| |BAR| | w
| | | | i
| X | | | e
| X X | | X |
| X X X | | X O |
| O X X X | | X X O |
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
Pipcount X: 113 O: 109 X-O: 0-0/Money (1)
CubeValue: 1
Rollout Money equity: 0,422
0,5% 22,0% 64,2% 35,8% 8,5% 0,3%
95% confidence interval:
- money cubeless eq.: 0,422 ą0,020,
- live cube no double: 0,533 ą0,034,
- live cube double take: 0,636 ą0,054.
Rollout settings:
Full rollout,
324 games (equiv. 14803 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
Evaluations
1. Double, take 0,592
2. No double 0,510 (-0,082)
3. Double, pass 1,000 (+0,408)
Proper cube action: Double, take
Live cube
1. Double, take 0,636
2. No double 0,533 (-0,104)
3. Double, pass 1,000 (+0,364)
Proper cube action: Double, take
"Micke Nilsson" <na...@home.se> skrev i meddelandet
news:zXjTe.145793$dP1.5...@newsc.telia.net...
Micke Nilsson
you automatically expect something fishy. There is a completely different
thing if you are up against the positions in a real match. Try to picture
yourselves up against them during a match, otherwise you might get it wrong
just because you try to find an unexptected solution.
And please, no bot solutions right from the start. Give everyone a chance to
find the problem and try to find an own solution...
"Micke Nilsson" <na...@home.se> skrev i meddelandet
news:0UjTe.145792$dP1.5...@newsc.telia.net...
Raccoon
>
> Problem #3, O on roll
>
>
> O on roll, cube action
> +-1--2--3--4--5--6--------7--8--9-10-11-12-+
> | X O | | O O O X X |
> | X O | | O O O X |
> | O | | O O X | S
> | O | | | n
> | O | | | o
> | |BAR| | w
> | | | | i
> | | | | e
> | X X | | X |
> | X X | | X |
> | O X X | | X O |
> +24-23-22-21-20-19-------18-17-16-15-14-13-+
> Pipcount X: 151 O: 133 X-O: 0-0/Money (1)
> CubeValue: 1
I started to reply "13/9 13/11" and "Double/Take" to Problems 1 and 2,
but then I realized they looked familiar :)
Here, for 0: Up 18 pips, a blot, some joker 5-primers, and X's Swiss
cheese board: Double.
For X: An anchor, 0's open 5 and 4 points, all X's checkers in play,
priming potential: Take.
For Snowie users: This looks like a very easy take. If Snowie
evaluation says "no double," I could be wrong and Snowie could be
right. But keep in mind that Snowie generally overvalues 22-point
anchor games, so evaluation and rollout "no double" equities may be
inflated by wrong takes later on.
paulde...@att.net
double/take. I'm fairly confident it's a double, less sure of the
take.
Paul Epstein
Grunty
O's main source of strenght is not rolling a 5, but the 10 rolls that
make the 4 or 5 point without breaking his prime. (I'm not counting 54
and 52, which would rather hit than making a point.) O's block has a
big potential, despite the stack on the 6 point. Should O make either
point, X probably had to pass.
O rolling a 5, X doesn't get positionally much hurt having another man
sent back. This could contribute to either advance his anchor or land
on the very rear, thus covering all the inner ground. X has lots of
counter-hits, and O's re-hits could lead to establish a second anchor.
So, with a third man back X doesn't stand bad in a *positional* sense.
But, this improvement in defensive resources will cost him more
gammons, and this is a *money* downside.
Except 66 (market-loser?) and 65 (hit), any other 6 permits O to anchor
outside. These 8 rolls keep some contact. Still, O would have a good
double next turn. The remaining outside contact, combined with the gap
in front of his defensive anchor, gives X a take.
So, computing:
O has 11 rolls close to market-losers + 8 rolls securing blots + 11
hits = 30 improving rolls, too much for not doubling.
X has 36 - 11 = 25 rolls that holds the position, but 11 of them are
somewhat gammonish, and 8 of them give O a clear edge.
Over the board, I would have taken. But in quiet analysis, those
gammons lead me to (marginally) pass.
paulde...@att.net
affect the cube actions in marginal cases, creating a bias towards
conservatism. Marginal doubles become holds and marginal takes become
passes. I was ignoring the rake factor, and leaving aside my knowledge
that the position is from an online game. (And I assume, Grunty also
was considering the pure position without this messy consideration)
Paul Epstein
Micke Nilsson
Money session. Score X-O: 0-0
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| X O | | O O O X X |
| X O | | O O O X |
| O | | O O X | S
| O | | | n
| O | | | o
| |BAR| | w
| | | | i
| | | | e
| X X | | X |
| X X | | X |
| O X X | | X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 151 O: 133 X-O: 0-0/Money (1)
CubeValue: 1
3-Ply Money equity: 0,351
0,6% 12,1% 65,0% 35,0% 7,2% 0,3%
1. No double 0,515
2. Double, take 0,447 (-0,068)
3. Double, pass 1,000 (+0,485)
Proper cube action: No double, take 12%
Rollout:
Money session. Score X-O: 0-0
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| X O | | O O O X X |
| X O | | O O O X |
| O | | O O X | S
| O | | | n
| O | | | o
| |BAR| | w
| | | | i
| | | | e
| X X | | X |
| X X | | X |
| O X X | | X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 151 O: 133 X-O: 0-0/Money (1)
CubeValue: 1
Rollout Money equity: 0,382
0,5% 12,9% 66,4% 33,6% 7,8% 0,2%
95% confidence interval:
- money cubeless eq.: 0,382 ą0,021,
- live cube no double: 0,621 ą0,053,
- live cube double take: 0,566 ą0,077.
Rollout settings:
Full rollout,
324 games (equiv. 12444 games),
played 2-ply (fast), cube 2-ply,
settlement 0,550 at 16 pts,
seed 1, without race database.
Evaluations
1. No double 0,571
2. Double, take 0,519 (-0,052)
3. Double, pass 1,000 (+0,429)
Proper cube action: No double, take 11%
Live cube
1. No double 0,621
2. Double, take 0,566 (-0,055)
3. Double, pass 1,000 (+0,379)
Proper cube action: No double, take 13%
"Micke Nilsson" <na...@home.se> skrev i meddelandet
news:0UjTe.145792$dP1.5...@newsc.telia.net...
Grunty
> 3ply:
> Proper cube action: No double, take 12%
> Rollout:
Humm...er... Snowie seems to be way off my mark.... time to revise its
playing algorithms.
Now seriously, I just CAN'T believe that position is NO double ! Or
should I say no double YET ?
Possibly the current position is already strong enough, but since next
turn is likely to be still a take, it'd be more efficient to wait a
turn?
I'd like to know the rationale behind Snowie's evaluation.
Micke Nilsson
might have done something wrong, but earlier Raccoon kind of saw this
coming.
Tomorrow I will try to post quiz 4 of the 18 decisions that Gammonfever got
all right. According to Snowie at least...
"Grunty" <grunti...@yahoo.com> skrev i meddelandet
news:1126371003.4...@z14g2000cwz.googlegroups.com...
Raccoon
paulde...@att.net wrote:
> An important general point is that, in online backgammon, the rakes
> affect the cube actions in marginal cases, creating a bias towards
> conservatism. Marginal doubles become holds and marginal takes become
> passes.
How so?
010010 vs 110000 is a marginal double, right?
With a $10 stake and 2.5% per player PER POINT rake,
ND -> 19/36*$9.50 - 17/36*$10.00 = $5.01 - $4.72 = $0.29
DP -? 36/36*9.50 = $9.50
DT -> 19/36*$19.00 - 17/36*$20.00 = 10.03 - 9.44 = $0.59
Double/Take.
With a $10 stake and $0.25 per player PER GAME rake,
ND -> 19/36*$9.50 - 17/36*$10.00 = $5.01 - $4.72 = $0.29
DP -? 36/36*9.50 = $9.50
DT -> 19/36*$19.50 - 17/36*$20.00 = 10.29 - 9.44 = $0.85
Double/Take.
020000 vs 110000 is a marginal take, right?
With a $10 stake and 2.5% per player PER POINT rake,
ND -> 25/36*$9.50 - 11/36*$10.00 = $6.60 - $3.05 = $3.45
DP -? 36/36*9.50 = $9.50
DT -> 25/36*$19.00 - 11/36*$20.00 = 13.19 - 6.11 = $7.08
Double/Take.
With a $10 stake and $0.25 per player PER GAME rake,
ND -> 25/36*$9.50 - 11/36*$10.00 = $6.60 - $3.05 = $3.45
DP -? 36/36*9.50 = $9.50
DT -> 25/36*$19.50 - 11/36*$20.00 = 13.54 - 6.11 = $7.43
Double/Take.
paulde...@att.net
I'd like to start by saying how much I enjoy your postings and
writings. I like the precision and the detailed calculations.
It must be the case that the rakes can affect the cube actions.
However, this effect is only clear to me in the rake-per-point case.
Let us then do some more calculations for this case. Your
rake-per-point analysis of 010010 vs 110000 is perfect. However, I can
make my point with a case that is even more marginal.
010010 vs 101000 is redouble/take, right?
With a $10 stake and 2.5% per player PER POINT rake, and cube at 2,
ND -> (19/36 + 17/36 * 1/18) * $19.00 - 17/36* 17/18 * $20.00 = $1.61
DT -> 19/36 * $38.00 - 17/36*$40.00 = $1.17
So the action changes from a redouble/take to a hold.
O.k., so perhaps I can only make my point with an ultra-marginal case.
However, it's still relevant because the context of the original
discussion was an analysis of a particular player's cube actions. And
one of these does appear to be ultra-marginal, perhaps to the same
degree as the one I cite above.
Paul Epstein
paulde...@att.net
No, wrong! Of course, there is no completely precise definition of
"marginal", and nor is there a precise definition of "blunder".
However, I do think there is a precise connection between the two
terms.
Surely, a "marginal take" means "a technically correct take in a
position where passing is an error but not a blunder." Here, passing
loses 1/9 of a point. So your "marginal" designation is inconsistent
with what you've said before (and which is a fairly common standard) --
that a play losing 0.1 points is a blunder.
Despite my disagreements here, and in an earlier posting, I appreciate
your precise analysis.
Paul Epstein