How to analyze a cube action?
Banjo Bluesman
A question on evaluating positions -- I've read all about how I should
double if I have corrected winning chances of 75% and I should take if I
have 25% chance. But I've never really seen anything on how to know when I'm
there.
I've read some books, most of which show a position, say "Cube action?" and
then give a very vague summation of one side's structure .vs. the other
side's deceivingly strong-looking position and then just pontifically says,
"Black should double, white has a close take." How do you get to that point?
If Snowie and Jellyfish do it, then there must be some structured way of
looking at a position and getting some sense of each side's winning chances.
Even if it's impractical sitting in the chair, just knowing how it works in
theory can help guide the practice.
Any takers on explaining it in simple terms?
--
Daniel Murphy
Kit Woolsey made up a helpful and simple rule: If you look at a
position and you're not sure the other side should take, then you MUST
double. If that helps you, great. If your game's a little more
sophisticated than that, also great
But I've tried to explain this rule to some novice players who just
say: Take? Drop? How the hell should I know?
So, to keep it very simple, your first goal in analyzing a position is
to answer this question: Who's winning?
And then try for a little more detail in your answer: are you winning
just a little, winning by a lot? Do you have a "can't lose" position?
Do you threaten to improve so much your opponent will wish he's never
seen a cube?
And here's your thinking structure:
Is there still contact between your opposing armies of checkers? If
not, you've got a simple race. Count the pips. If you lead by about
9%, double. 10%, redouble.
(If you've counted pips once and decide not to double, remember the
numbers. Subtract from that pipcount with each roll until you reach a
count that seems like a double. That way, you shouldn't have to
tediously count pips more than once in a game.)
If there is still contact, you need to do two things. The first is:
compare positions. Who's got more checkers back? Who's got the better
anchor? Who's got the better prime? Who's got the better flexibility
to turn a good position into a great one? Or the awkward position that
threatens to get worse? Who's got many rolls that do good things?
Who's relying on a miracle? Who's theatening to turn a contact
position into a no contact race? And who's the favorite if that
happens?
The second thing is: figure out if your position threatens to improve
so much that -- if you're successful -- your opponent wouldn't dream
of taking a cube. This is tricky: a lot of novices see a couple of
good numbers coming and turn the cube. Optimists. In backgammon it's
good to be an optimist but things don't always go our way. So, another
and often helpful way of thinking about this is: If you have several
numbers that will give you an overwhelming position, and most of the
the rest of the numbers leave you no worse than even, then you
probably have a double.
This may seem unsophisticated. I haven't even mentioned "math." But it
works. And believe me, if you can ask and answer the basic question:
who's winning? -- or even just consistently remember to ASK -- you're
a lot farther along than a lot of your opponents.
Daniel Murphy
rac...@best.com, Raccoon on FIBS, GamesGrid
Vi ses, og tak for alle fiskene!
13th Nordic Open, 12-16 April 2001
http://www.nordicopen.dk
Leo Bueno
wrote:
>
>Kit Woolsey made up a helpful and simple rule: If you look at a
>position and you're not sure the other side should take, then you MUST
>double.
What's the rule for making the *take* decision?
======================================================================
Leo Bueno leob...@usa.net 305-818-9129 305-669-5260
P.O. Box 440545 Miami FL 33144-0545 U.S.A.
AOL Instant Messenger ("buddy"): leobueno
Castro Fall Poll - http://home.earthlink.net/~leobueno/adiosfidel.html
Cuba Books - http://home.earthlink.net/~leobueno/cubabooks.html
Examen de Cubania - http://home.earthlink.net/~leobueno/cubania.html
======================================================================
Ric Gerace
Ric
--
============================
Life is a dark ride.
Visit me at www.oncapecod.net/rickgerace
"Daniel Murphy" <rac...@best.com> wrote in message
news:3a7bca18...@news.cybercity.dk...
Daniel Murphy
Bueno) wrote:
>What's the rule for making the *take* decision?
Well, take decisions are more difficult to assess accurately than
doubling decisions. If I don't know if you should take, I double.
That's easy; YOU have to decide what to do. "I double, you think!"
You can apply the same comparision method you use for doubling
decision to take decisions, too. But I it's mostly a matter of
experience -- you keep taking cubes until you realize which ones you
just don't win often enough to make taking worthwhile. There's no easy
way to learn where the drop/take line is on middle game contact
doubles. Few players can estimate equity precisely in complex
positions, but they don't need to, if experience tells them what to
do.
No contact racing positions are easier: a simple rule, usually
accurate, is that you take with no more than a 12% pipcount deficit.
If there's one general rule for taking it is:
When in doubt, take.
An experienced player who is aware of his own tendencies to err might
benefit from "when in doubt, take" or "when in doubt, drop." I, for
example, know I take too many drops, and I drop a take about twice a
year. So when I'm in doubt, I drop -- usually :)
But "when in doubt, take" must be right for most novice and
intermediate players. First, because (my impression is that) bad drops
are more costly to many novices than bad takes. And second, because
playing for little or no stakes except ratings points, you don't learn
anything by dropping takes.
There are a few other techniques you can also apply.
Consider what your opponent must do -- how well he must perform -- to
make you wish you had dropped his double. What is he threatening to
do? How likely is he to roll that well? And consider -- if you take
the cube, how do you plan to win? What's your winning strategy? Do you
have one? If you can't come up with a plan, drop.
A common intermediate mistake is to double with a position that
certainly threatens to improve a good position, but isn't too likely
to immediately become overpowering. For example, your opponent
threatens to make either of two points, but he really needs to make
both of them (and that's not likely) to wipe you out. So take.
Similarly, the longer the game is from ending, the more likely it is
that you should take. The more good things your opponent must do, the
more chances he has to err or roll badly, the more chances you have to
turn the game around, the more willing you should be to take. For
example, if you need to hit to win and can expect to get only one
indirect shot, you should drop. But if you can anticipate two winning
indirect shots, you probably have a take. Count the ways you might win
the game. A couple of indirect shots, maybe a direct shot, together
with some racing chances usually add up to a take.
Another way to improve take decisions is to think about the cube on
every roll. While your opponent is shaking the dice, think: what would
I do if he doubled? Drop? Take gladly? Take confidently? Take
reluctantly? Drop immediately?
The cube, when it comes, should never be a complete surprise. One
benefit you get from thinking about the cube every roll is this:
Sometimes you'll be thinking about a possible take and figure you
should drop, but your opponent doesn't double. Well, if the cube comes
2 or 3 rolls later and your position hasn't improved, drop!
Another benefit from thinking about the cube every roll is that it
should help you become more aware of how positions develop from no
double, to double/take, to double/drop. That should help you with both
doubling and taking decisions and also further your awareness of
development and timing.
BGtallrock
>Well, take decisions are more difficult to assess accurately than
>doubling decisions.
Snowie suggests that this isn't true. Most people who save their matches will
find over time that their double decisions have a higher error rate than their
take decisions.
I suspect that the reason for this is that people generally evaluate positions
identically from either side of the board, but that since they are never
"forced" to make a double decision, but are always forced to make a take/drop
decision, means more errors [ie missed cube decisions when they forgot about
the cube].
Chase
Bueno) wrote:
|What's the rule for making the *take* decision?
To the other fine advice that has been posted, I will add a quote from
Kit Woolsey that I have found helpful (emphasis mine):
"If you are not in IMMEDIATE danger of being blitzed, primed,
crunched, or outraced, then it is probably a take."
Chase
_______
To respond via email, replace "USERNAME" with "demiga" in my address.
Mark Driver
Backgammon Puzzles),concise cube advice to be effective, simple and easy to
remember.
Paul uses the acronym PRAT as a guide to doubling and accepting:
PRAT stands for - Postion, Race And Threats
Paul recomends to double when you have an advantage in two of these three
areas. If you have a clear advantage in all 3, then your opponents should
pass.
Daniel Murphy
>Daniel Murphy suggests:
>
>>Well, take decisions are more difficult to assess accurately than
>>doubling decisions.
>
>Snowie suggests that this isn't true. Most people who save their matches will
>find over time that their double decisions have a higher error rate than their
>take decisions.
That's an interesting point, but it doesn't contradict me, even if I'm
wrong to say that it's easier to accurately assess a double decision
than a take or drop problem.
You are probably right that most people suffer more from the slings
and arrows of Snowie's error ratings because of their doubling
decisions than their taking decisions. But I think the simplest
explanation for that is (1) in our matches we are faced with many,
many more doubling decisions than taking decisions and (2) bad
doubling decisions (not doubling doubles) is the most harmful of all
doubling errors. But more numerous and more harmful does not
necessarily mean more difficult.
It should also be true that doubling and taking/dropping are two sides
of the same coin and if we can solve one half of the problem, we
should also have the other. Perhaps what's happening is that doubling
and taking are indeed equally difficult to "accurately" assess, but we
don't need as much accuracy in the doubling decision, since once we
(finally) decide we have a double, we don't have to decide whether or
not it is also a take. That's our opponent's headache.
In any case, the point I was trying to make in the sentence you quote
isn't very clear, but it's a sidenote, I think, to the rest of it all,
and won't be corrected until the next draft :)
Daniel Murphy
rac...@best.com, Raccoon on FIBS, GamesGrid
13th Nordic Open, 12-16 April 2001
50,000 DKK added http://www.nordicopen.dk
Douglas Zare
> On Sat, 03 Feb 2001 09:06:37 GMT, rac...@best.com (Daniel Murphy)
> wrote:
>
> >
> >Kit Woolsey made up a helpful and simple rule: If you look at a
> >position and you're not sure the other side should take, then you MUST
> >double.
>
> What's the rule for making the *take* decision?
You have to have a lot of experience. You have to have a feeling for how
likely it is for you to get gammoned from the given position, and how
likely it is for you to be able to turn the game around. I would focus on
a few common areas first: the blitz, the race, and bearing in against a
high anchor. Make sure you understand these well enough that you are not
making huge errors.
If gammons were not an issue and volatility did not exist, then you should
take if you have a 50% chance of evening up the game, taking into account
that you have control of the cube.
Try to figure out how your position might improve, or how you can put
together an adequate defense. If it will definitely look worse in the
future, e.g., you closed out your opponent but have a checker stuck behind
a 6-prime, then your position is already bad now.
Douglas Zare
Webby
> Paul uses the acronym PRAT as a guide to doubling and accepting:
>
> PRAT stands for - Postion, Race And Threats
>
> Paul recomends to double when you have an advantage in two of these three
> areas. If you have a clear advantage in all 3, then your opponents should
> pass.
You beat me to it Mark :-)
I just want to add a little more about the "PRAT" method. First of all it is
easy to remember as you just ask yourself not to be a "PRAT" on doubling
decisions :-) "Prat" btw in english, english is a derogatory term meaning
"Idiot" Not sure if you can say it across the pond though :-)
Since I read Paul Lamford's book, about 6 months ago now, I have used this
method pretty much extensivly. I tested it in fact recently against the
doubling positions in Roberties "501 essential bg positions" by not relying
on experience but simply solely using the method described. I found that the
results were impressive and I made a marked improvement from when I first
tried them using just experience, especially in contact positions.
I have tweaked the general principal slightly myself in that I attach
weightings to each of the 3 main areas. Meaning for example, if all things
being equal, I will weigh in the pip count higher than the threat and
position areas. I also take good account of the score in match play and
gammon potential both ways.
The method itself is not that much good to a real beginner however as you
still need to know if your position is in fact better and still need to know
if you have any threats and what they are. For those with a reasonable
understanding of the game, say 1550+ I believe it can help a long way. It
isn't perfect as there are some positions where other factors override the
result of the method, particular prime vs prime positions, but it has helped
my game along nicely and I'm making far fewer doubling errors than I used
to.
Just to illustrate things let's see the guidelines in action...
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| X O O O | | O O X |
| X O O O | | O X |
| O | | X | S
| O | | X | n
| | | | o
| |BAR| | w
| | | | i
| | | | e
| X | | |
| X O X | | X X O |
| X O X | | X X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 150 O: 132 X-O: 0-0/3 (1)
CubeValue: 1
X O
---------------------------
Position| - * Better board, high anchor.
|
Race | - * O is 18 pips ahead and on roll
|
and |
|
Threats | - * Positioned well to make bar or 4 point.
From the above we gather that O has the better Position, is ahead in the
Race and Threatens to make the bar or 4 point.
So if we use the criteria that 3 areas in your favour should be double /
pass, then the correct cube action must be double / pass. Snowie agrees...
1. Double, pass 1,000
2. No double 0,920 (-0,080)
3. Double, take 1,210 (+0,210)
How about this one...
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| O X O | | O O X |
| O O | | O O X |
| O | | O X | S
| | | X | n
| | | X | o
| |BAR| | w
| X | | | i
| X | | O | e
| X | | X O |
| X | | X O |
| O X X | | X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 143 O: 148 X-O: 1-1/3 (3)
CubeValue: 1
X O
----------------------
Position| - * Better board, 11 point
|
Race | * - O trails by 5 pips.
|
and |
|
Threats | - * Pointing on the 5 pnt, hitting on 21, 2 checkers up
So O has 2 out of the 3 areas which according to the general rule is a
double / take.
Snowie sayssss...
1. Double, take 0,732
2. No double 0,666 (-0,067)
3. Double, pass 1,000 (+0,268)
Proper cube action: Double, take
Last but not least...
O on roll, cube action
+-1--2--3--4--5--6--------7--8--9-10-11-12-+
| O O O O O | | O X |
| O O O O | | |
| O | | | S
| | | | n
| | X | | o
| |BAR| | w
| | | | i
| | | | e
| X | | O |
| X X X X | | X X O |
| X O X X X | | X X O |
+24-23-22-21-20-19-------18-17-16-15-14-13-+
Pipcount X: 108 O: 113 X-O: 0-0/5 (1)
CubeValue: 1
X O
----------------------
Position| - * O clearly has the better board and remains flexible
|
Race | * - O trails by 5 pips.
|
and |
|
Threats | - * Numerous. Covering, hitting, escaping etc.
So what we have here is 2 out of 3 of the areas so according to the
guideline this is a double/take right?
1. Double, pass 1,000
2. No double 0,731 (-0,269)
3. Double, take 1,151 (+0,151)
Proper cube action: Double, pass
Wrong!
This is why you should only use it as a general guideline. Here the position
and threats are so great that they override the small pip deficit. Gammon is
another factor and X is so inflexible with buried checkers and ...etc etc.
If you are going to use the method then look out for positions where one
area clearly overrides the others.
Also beware that not all "good in all 3 areas" are double / passes. They may
not even be doubles of course as you may well be too good to double.
Anyway, it works for me :-)
regards
Alan
Webby's Backgammon Site
quizzes, downloads, tips, annotated matches, links...