one more dumb beginner question
paul new
point.
JF doubled.I know that 5/3 is not a great opener,nor the three point the
best move,but wasn`t this an "incorrect" double?
Paul New
JP White
> In a 5 point match against JF7 tied 3 3,I opened with 5/3 and made the 3
> point.
> JF doubled.I know that 5/3 is not a great opener
5-3 is pretty good in my book. Be thankful for small mercies.
> ,nor the three point the
> best move,
If your first on roll, it's the best move in my book.
> but wasn`t this an "incorrect" double?
>
Not at all.
See http://www.bkgm.com/articles/mpd.html#two-away
You may not understand all the terminology, but you were playing two-away
two-away (Ie you were both 2 away from winning the match). Jellyfish was
correct in doubling. If you don't double and you win a single point you have
to play again in order to win the match. Doubling puts the match on the line
straight away. So not doubling in most cases is counter productive.
I've not analyzed (no time) it but if you rolled 6-6, I'd suspect Jelly
wouldn't double since it would be a big underdog. Anyone know if this is the
case?
JP
--
JP White
mailto:jpwh...@bellsouth.net
-=Leslie=-
"JP White" <jpwh...@bellsouth.net> wrote in message
news:3AE617BC...@bellsouth.net...
If you rolled 6-6, you'd have to roll again, because it's the first roll.
Le吸ie
JP White
>
> If you rolled 6-6, you'd have to roll again, because it's the first roll.
> Le§lie
How true.
Typing and not thinking.
Alef Rosenbaum
then you roll double 6. Does JF then on its
first opportunity to double actually do so?
8^) Alef
JP White wrote:
>
> -=Leslie=- wrote:
>
> >
> > If you rolled 6-6, you'd have to roll again, because it's the first roll.
> > Le吸ie
JP White
> But you can since JF could win the opening roll
> then you roll double 6. Does JF then on its
> first opportunity to double actually do so?
>
Snowie does.
Match to 5. Score X-O: 3-3
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| O | | X O O X |
| O | | X O X |
| O | | O X | S
| O | | | n
| O | | | o
| |BAR| | w
| X | | | i
| X | | O | e
| X | | X O |
| X | | X X O |
| O O X | | X X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 143 O: 164 X-O: 3-3/5 (1)
CubeValue: 1
Rollout Money equity: -0.356
0.3% 6.6% 36.3% 63.7% 14.5% 0.6%
95% confidence interval:
- money cubeless eq.: -0.356 ą0.012,
- live cube no double: -0.739 ą0.014,
- live cube double take: -0.739 ą0.014.
Rollout settings:
Truncated rollout, depth 11,
180 games (equiv. 32153 games),
played 2-ply (medium), cube 3-ply,
seed 1, with race database.
Evaluations
1. Double, take -0.739
2. No double -0.739 (-0.000)
3. Double, pass 1.000 (+1.739)
Proper cube action: Optional double, take
Live cube
1. Double, take -0.739
2. No double -0.739 (-0.000)
3. Double, pass 1.000 (+1.739)
Proper cube action: Double, take
Svein Dybvik
a perfect player and a perfect player would certainly double in the
next turn if he/she got a 6-6 in the first throw. At 2-away 2-away
it is cruicial that you double at the first risk that you are going
to loose your market. So JF thinks that since you are going to double
in the next roll anyway, it might as well do it before you.
In a human-human match humans would probably not double if the opponent
got 6-6 in the first throw, since they know they are not playing a perfect
player.
Non-perfect players tend to forget things...for instance doubling.
They also may not be aware of the concepts at a 2-away 2-away match score.
Then you should wait until you yourself have a market looser.
SD
"Alef Rosenbaum" <al...@clara.co.uk> wrote in message
news:3AE62408...@clara.co.uk...
Daniel Murphy
wrote:
>paul new wrote:
>
>> In a 5 point match against JF7 tied 3 3,I opened with 5/3 and made the 3
>> point. JF doubled. ... wasn`t this an "incorrect" double?
>
>Not at all.
>
>See http://www.bkgm.com/articles/mpd.html#two-away
>
>You may not understand all the terminology, but you were playing two-away
>two-away (Ie you were both 2 away from winning the match). Jellyfish was
>correct in doubling. If you don't double and you win a single point you have
>to play again in order to win the match. Doubling puts the match on the line
>straight away. So not doubling in most cases is counter productive.
>
>I've not analyzed (no time) it but if you rolled 6-6, I'd suspect Jelly
>wouldn't double since it would be a big underdog. Anyone know if this is the
>case?
If you edit a JellyFish position to give yourself an opening 66 and
present the position to JellyFish as a two point match, JellyFish does
not double. But normally JellyFish doesn't do any calculating when
both players need two points for the match (2-away 2-away). It is
simply preprogrammed to always double at its first opportunity whether
or not it is a favorite. If you start with your best legal roll, 31,
JellyFish will double.
JellyFish's preprogrammed cube action at (2-away 2-away) implements a
theoretically correct strategy.
This is how JellyFish might explain its strategy:
"I'm a perfect player and so are you. Before either of us plays the
first move, we've each got a 50% chance to win the game. So I might as
well tell you now that no matter what you roll or I roll, I am going
to double the first chance I get. You might as well think of the cube
as already being on 2, and therefore we both have a 50% chance of
winning the match.
"This means that I might be doubling when you are the favorite. Silly?
Actually, it costs me nothing, because since you are a perfect player,
I know you would never make the biggest mistake possible at 2-away
2-away. That mistake is to allow your position to become so good that
when you double, I should pass. I know you will always double when I
should still take. I might as well double you first because if I
don't, you will double me and I will take anyway. So it doesn't matter
who doubles, since the cube will always end up on 2."
Although this strategy is not wrong, neither is it the best strategy
against real, imperfect, unequal opponents. JellyFish is correct if
neither player makes a mistake with the cube. But imperfect opponents
make mistakes.
What are these mistakes?
(1) Your opponent might make the mistake of not doubling until you
should pass. If your opponent starts with a good roll, you can give
him the opportunity to make this mistake by NOT doubling yourself.
Your opponent might double, and you will take. He might not double,
and you'll be glad. He might let his position get so good before
doubling that you should pass, and then you'll be very happy that you
can still drop. As long as there is no possibility that your own
position becomes so good that your opponent should drop, you might as
well leave the cube in the middle.
(2) Your opponent might make the mistake of dropping when he should
take. If he takes, you are no worse off than if you had doubled
earlier. But if he drops when he should take, you gain.
Do you see why you gain? By definition, your opponent should take and
play the game for the match if his chances of winning the game are
better than his chances -- after dropping -- of winning the next two
games, or a gammon in one. If his chances are worse after he drops,
then your chances must be better after he drops.
(3) Your opponent might make the mistake of taking when he should
drop. At first sight, this may seem like a good situation for you!
Your opponent is certainly worse off for his mistake. Unfortunately,
the only way to reach this situation is to make a mistake yourself, by
letting the position become so good that your opponent can now improve
his match winning chances by dropping and playing one or two more
games instead. You've made the biggest mistake possible at 2-away
2-away.
Here's the big question to ask yourself at 2-away 2-away: "If I don't
double and roll something good, and my opponent rolls something bad,
is it at all possible that if I double on my next roll my opponent
should drop?" In backgammon terminology, do you have some sequences of
rolls (one roll each by you and opponent) that lead to a market-losing
position (you have waited too long to "sell" your cube, and your
opponent should correctly refuse to "buy" it.)?
The best basic 2-away 2-away strategy, then, is to double whenever you
have any market-losing sequences. With correct play by both players,
all 2-away 2-away games should end the match with a 2-cube given and
accepted. You should always double before your position becomes too
good.
How good is too good? When have you lost your market? Most players use
a standard match equity table developed by Kit Woolsey and Hal
Heinrich to answer this question. This table tells us that opponent's
chances of winning a match trailing 2-away 1-away in the Crawford game
is 30%. Therefore, at 2-away 2-away, opponent should take your double
as long as he can win the game at least 30% of the time. If you let
your chances of winning the game rise to more than 70%, you have lost
your market.
Don't let this happen to you! At 2-away 2-away, market-losing
sequences abound for one or both players within the very first few
rolls of the game. At 2-away 2-away, the cube should usually be coming
quickly. Against equal or better opponents, be optimistic; double
early when you are the slightest of favorites or can foresee just a
handful of sequences that will give your opponent reason to drop.
Unless your opponent is so hopeless that you think he's very likely to
overshoot HIS market, doubling too late is much worse than doubling
too early. If you happen to double too early and the game turns sour,
the usual result is that the game ends with the cube on your
opponent's side of the table instead of yours. You've lost nothing.
But if you let him escape to play another game or two, you've given
your opponent a better chance than he deserved.
Once you've developed your evaluating skills, it's possible to refine
2-away 2-away strategy. Many players do not always double with a very
small number of market losers. They wait with the cube until a
position becomes difficult for opponent to evaluate, but -- and this
is very important -- it is still correct for opponent to take. They're
willing to risk the small chance that they will actually lose their
market in exchange for a larger chance of opponent dropping a takable
double.
But it's very difficult to evaluate positions -- looking two rolls
ahead to see whether it's possible that your winning chances will
shoot over 70%. It's hard to know what opponent should do, and harder
to predict what he will do. And it's a little too easy to overestimate
our own skills in comparison to our opponents'. The worse you yourself
are at evaluating, the more willing you should be to turn the cube to
2. By doubling, you give up the chance to drop a cube later, but most
of your opponents won't be losing their marke. If you're an expert, or
your opponent is very poor, you can be a little more cagy, as long as
you're very confident that you won't be the one to make the big
mistake.
Let's go back to JellyFish's double after you rolled 53 and made your
3 point. It happens to be true that there are no market-losing
sequences. No matter what JellyFish rolls and you roll, if JellyFish
doubles next time you will still have a take. In fact there's nothing
you could have rolled that, after JellyFish's roll and your next roll,
you should drop JellyFish's cube. So there's really no need for
JellyFish to double after your 53.
But it's also true that no matter what JellyFish rolls, if YOU double,
JellyFish will still have a take. And since JellyFish assumes you will
never make the mistake of not doubling when you should -- on your next
or any turn -- JellyFish doubles now.
After your 53, JellyFish -- the expert -- might have done better by
waiting. But if JellyFish had rolled the 53 against a novice player,
the double would have been perfect.
--
You can find the Woolsey-Heinrich match equity table at
www.gammonline.com (see the "demo issue"), and many articles on 2-away
2-away strategy at www.bkgm.com (Backgammon Galore). Besides the
article JP White mentions, see the "Newsgroup Archive" there.
---
Daniel Murphy
Raccoon on FIBS, GamesGrid
Vi ses, og tak for alle fiskene!
Daniel Murphy
wrote:
>Alef Rosenbaum wrote:
>
>> But you can since JF could win the opening roll
>> then you roll double 6. Does JF then on its
>> first opportunity to double actually do so?
>
>Snowie does.
After Snowie's 21 and Opponent's 66 at 2-away 2-away, Snowie DOESN'T
double.
Your Snowie live-cube rollouts, JP, say "double/take" (with no gain in
equity over no double), and if the position is set up for Snowie to
evaluate Snowie says "optional double/take" with equity of -0.619 for
both "double/take" and "no double", but if you actually play Snowie a
match with manual dice and the panels open, Snowie doesn't double
after its 21 and opponent's 66. It says "no double -0.535, double/take
-0.681". Doubling is a blunder.
JellyFish doesn't double either.
JellyFish's is hardwired to double immediately at 2-away 2-away when
Opponent wins the first roll of the game. If JF rolls first, though,
it "thinks" on its second roll and doesn't always double.
Daniel Murphy
>
>> In a 5 point match against JF7 tied 3 3,I opened with 5/3 and made the 3
>
>Not at all.
>
>See http://www.bkgm.com/articles/mpd.html#two-away
>
>You may not understand all the terminology, but you were playing two-away
>two-away (Ie you were both 2 away from winning the match). Jellyfish was
>correct in doubling. If you don't double and you win a single point you have
>to play again in order to win the match. Doubling puts the match on the line
>straight away. So not doubling in most cases is counter productive.
>
>I've not analyzed (no time) it but if you rolled 6-6, I'd suspect Jelly
>wouldn't double since it would be a big underdog. Anyone know if this is the
>case?
If you edit a JellyFish position to give yourself an opening 66 and
present the position to JellyFish as a two point match, JellyFish does
not double. But normally JellyFish doesn't do any calculating when
opponent wins the opening roll at (2-away 2-away). It is
simply preprogrammed to always double. If you start with your best
legal roll, 31, JellyFish will double.
JellyFish's preprogrammed cube action at (2-away 2-away) implements a
theoretically correct strategy.
This is how JellyFish might explain its strategy:
"I'm a perfect player and so are you. Before either of us plays the
first move, we've each got a 50% chance to win the game. So I might as
well tell you now that no matter what you roll, I am going
to double. You might as well think of the cube as already being on 2,
What are these mistakes?