The below is an extract from Kit Woolsey's special report "How to Play
Tournament Backgammon". If you want a copy, you can contact Kit on
FIBS.
"THE TWO-AWAY VERSUS TWO-AWAY SCORE"
"Before going into general match strategy, there is one match score
which deserves special discussion. That is the score where both
players are two away from match. What makes this score unique is that
whoever turns the cube first kills the cube. His opponent has no
value holding the cube. Under normal circumstances one of the reasons
you do not immediately double when you have an advantage is that you
are giving your opponent control of the cube - he has the option of
turning it to four in the future while you do not. When both players
are two away, that argument against doubling does not exist. In
addition you cannot prevent the cube from being turned by refusing to
turn it; since your opponent can turn the cube without giving away
anything, it will be of no value to you to own the cube.
Consequently, at two-away versus two-away you should always double if
there is any chance that after the next exchange you will lose your
market (i.e. your opponent will no longer have a take). For example,
suppose your opponent wins the opening roll of 4-1, and plays 13/9,
6/5. Even though he is probably the favorite, you should still
double! If you roll 4-4 and he flunks, you will clearly lose your
market, so you must turn the cube. Note that you don't gain anything
by not doubling, since if you don't hit the blot on his five point he
should immediately double if he knows what he is doing, since he might
lose his market, and you will still have to take. The only argument
for not doubling in this situation is if you believe that your
opponent does not understand this principle, so by waiting you give
him a chance to make the mistake of not doubling when he should and
losing his market if things go his way. If you are playing against a
perfect opponent, it is always correct to double at this match score
if there is any way you can lose your market on the next exchange. It
logically follows that if both players know what they are doing, this
must be the last game of the match."
Not the easiest of principles to understand. :) (I still have trouble)
Hope your question is answered (you can always debate it with Kit
himself on FIBS).
JF makes every move assuming it's playing against itself -- it tries to
make the best move based on the games it has played while training, and
it was trained against itself.
Depending on your discipline and skill level, you might do better not
cubing immediately at -2:-2 if you're fairly certain your opponent
doesn't fully understand the intricacies of the position. However you
can soon end up costing yourself equity when doing this.
(For the sake of discussion below, assume that a market loser puts you
at 80% to win, a non-market loser leaves you at 50% to win.)
Suppose you reach a position in which it's a huge take but if you roll a
44 and your opponent rolls a 66, you lose your market. The
theoretically correct action is double/take. But suppose you assume
your opponent doesn't know proper cube handling at this score. You
decide to risk losing some equity on that 1/1296 parlay in order to give
your opponent a chance to make a bigger mistake later.
Then you roll a 44.
Your opponent might now have some sequences that are market losers.
Let's suppose there are 35/1296 market losing sequences. However your
opponent is better off not doubling, losing equity on the 35/1296 market
losing sequences but gaining equity on the 36/1296 sequences that start
with 66. (Recall that 44, 66 was a market loser for you, so after your
opponent rolls a 66, you have a cash. If your opponent doesn't double,
he can drop your cube.)
Then your opponent rolls a number that's the beginning of a market
losing sequence for him.
You're now in the same situation your opponent was in -- you have some
market losers, but you also have some rolls after which you would drop.
You've suddenly given yourself a lot more to think about on every turn.
How certain are you that the potential gain from waiting and giving your
opponent a chance to make an error outweighs the risk of rolling a
market loser without doubling?
JF doesn't try to figure out the opponent's mistakes and gain equity from
playing into those, so it assumes you know what you're doing, and
doubles immediately.
-Michael J. Zehr
>can understand if a weaker player doubles a stronger player, but why
>would the stronger player double the weaker player? If I was playing a
>weaker player, I would much rather have the option of resigning, if the
>game went sour, and be able to play on in the match. I'm assuming JF is
>a much better player than I, so why does he double? Comments please.
First: Do you understand that, if you are a even a slight favorite to
win the game, even at your second roll, that you must double? Suppose
you are a 55-45 favorite. If you do not turn the cube, "because the
game might go sour", you will win one point 55% of the time, and be a
70-30 favorite in the match, and lose one point 45% of the time and be
a 30-70 underdog. Summing up your chances: .55 x .7 + .45 x .3 = .52.
That is, you will win the match 52% of the time.
But if you turn the cube, you will win the match 55% of the time. No
choice, is there?
And if you are not ahead - see the quotation from Mr. Woolsey, given
in another post on this thread.
deekay
>On Sun, 01 Mar 1998 19:33:41 -0500, "Richard P. Reasin"
><rpre...@computer-connection.net> wrote:
>
>>can understand if a weaker player doubles a stronger player, but why
>>would the stronger player double the weaker player? If I was playing a
>>weaker player, I would much rather have the option of resigning, if the
>>game went sour, and be able to play on in the match. I'm assuming JF is
>>a much better player than I, so why does he double? Comments please.
>>
>
>The below is an extract from Kit Woolsey's special report "How to Play
>Tournament Backgammon". If you want a copy, you can contact Kit on
>FIBS.
> It
>logically follows that if both players know what they are doing, this
>must be the last game of the match."
This being the case, the only argument I see for not doubling immediately is
that you think your opponent doesn't know what they are doing, or doesn't know
it as well as you do. Thus it seems that for less skilled players it would
always be correct to double immediately. Right?
All email sent to the address used for this post is deleted unread
(although headers may be used in my spam filters). To reach my real
email box, send to personal@ at the above domain.
This does not seem to always be the case. Consider the following
contrived example. Half your rolls lose the game outright, and half
leave you with an 80% chance of winning the game (clearly a market
loser).
No gammons for this example. If you double, you will win 40% of the
matches.
If you don't double, then 50% probability you lose one point (opponent
either wins or doubles you out) and 50% probability you win one point
(you double out opponent on next turn). So not doubling leaves you with
50% match equity, which is clearly better.
- Jim
Jim Williams wrote:
(snip)
> Consider the following
> contrived example. Half your rolls lose the game outright, and half
> leave you with an 80% chance of winning the game (clearly a market
> loser).
> No gammons for this example. If you double, you will win 40% of the
> matches.
> If you don't double, then 50% probability you lose one point (opponent
> either wins or doubles you out) and 50% probability you win one point
> (you double out opponent on next turn). So not doubling leaves you with
> 50% match equity, which is clearly better.
>
> - Jim
This looks a little bit TOO contrived. Your opponent DOES have the
option of cubing, too, right? Shouldn't s/he send it over after your roll?
In REAL backgammon, I'd like to see REAL positions....
Chuck
bo...@bigbang.astro.indiana.edu
c_ray on FIBS
: This does not seem to always be the case. Consider the following
: contrived example. Half your rolls lose the game outright, and half
: leave you with an 80% chance of winning the game (clearly a market
: loser).
: No gammons for this example. If you double, you will win 40% of the
: matches.
: If you don't double, then 50% probability you lose one point (opponent
: either wins or doubles you out) and 50% probability you win one point
: (you double out opponent on next turn). So not doubling leaves you with
: 50% match equity, which is clearly better.
: - Jim
This would be correct if you ever reached such a position. But how can
you get there? If 50% of your rolls now lose outright your opponent must
have had many market losing sequences before his previous roll, so he
clearly should have doubled. With correct play on both sides, you just
can't get from the starting position to the end of the game without the
cube being turned and accepted at this match score.
Kit
Lotus wrote in message <34fa66c7...@news.supernews.com>...
>On Sun, 01 Mar 1998 19:33:41 -0500, "Richard P. Reasin"
><rpre...@computer-connection.net> wrote:
>
>>can understand if a weaker player doubles a stronger player, but why
>>would the stronger player double the weaker player? If I was playing a
>>weaker player, I would much rather have the option of resigning, if the
>>game went sour, and be able to play on in the match. I'm assuming JF is
>>a much better player than I, so why does he double? Comments please.
>>
>
>The below is an extract from Kit Woolsey's special report "How to Play
>Tournament Backgammon". If you want a copy, you can contact Kit on
>FIBS.
>
>"THE TWO-AWAY VERSUS TWO-AWAY SCORE"
>
>"Before going into general match strategy, there is one match score
>which deserves special discussion. That is the score where both
>players are two away from match. What makes this score unique is that
>whoever turns the cube first kills the cube. His opponent has no
>value holding the cube. Under normal circumstances one of the reasons
>you do not immediately double when you have an advantage is that you
>are giving your opponent control of the cube - he has the option of
>turning it to four in the future while you do not. When both players
>are two away, that argument against doubling does not exist. In
>addition you cannot prevent the cube from being turned by refusing to
>turn it; since your opponent can turn the cube without giving away
>anything, it will be of no value to you to own the cube.
>
>Consequently, at two-away versus two-away you should always double if
>there is any chance that after the next exchange you will lose your
>market (i.e. your opponent will no longer have a take). For example,
>suppose your opponent wins the opening roll of 4-1, and plays 13/9,
>6/5. Even though he is probably the favorite, you should still
>double! If you roll 4-4 and he flunks, you will clearly lose your
>market, so you must turn the cube.
>>>>>> I must be missing something here ..... after the 4-1 is played (13-9
6-5) and double 4's are played (best at this match score is 24-20* 20-16*
13-9(2) )and now flunk (6-6) ..... now comes the double ..... this is an
easy take at -2 -2 match score. The side getting doubled wins more than 33%
of the games according to Jellyfish rollout at level 6. I am not saying it
is wrong to double at -2 -2 after the first roll, but i can not find any
position sequence of 3 rolls (his roll, your roll, his roll) where he would
past at this match score. Have I missed something???? If not it seems there
is a great reason to NOT double after the opening roll because as in the
example above you opponent might pass.
> Note that you don't gain anything
>by not doubling, since if you don't hit the blot on his five point he
>should immediately double if he knows what he is doing, since he might
>lose his market, and you will still have to take. The only argument
>for not doubling in this situation is if you believe that your
>opponent does not understand this principle, so by waiting you give
>him a chance to make the mistake of not doubling when he should and
>losing his market if things go his way. If you are playing against a
>perfect opponent, it is always correct to double at this match score
>if there is any way you can lose your market on the next exchange. It
>logically follows that if both players know what they are doing, this
>must be the last game of the match."
>
>
>
>Lotus wrote in message <34fa66c7...@news.supernews.com>...
(snip)
>>The below is an extract from Kit Woolsey's special report "How to Play
>>Tournament Backgammon". If you want a copy, you can contact Kit on
>>FIBS.
>>
>>"THE TWO-AWAY VERSUS TWO-AWAY SCORE"
(snip)
>>
>>...At two-away versus two-away you should always double if
>>there is any chance that after the next exchange you will lose your
>>market (i.e. your opponent will no longer have a take). For example,
>>suppose your opponent wins the opening roll of 4-1, and plays 13/9,
>>6/5. Even though he is probably the favorite, you should still
>>double! If you roll 4-4 and he flunks, you will clearly lose your
>>market, so you must turn the cube."
(Hugh interjects:)
> I must be missing something here ..... after the 4-1 is played (13-9
>6-5) and double 4's are played (best at this match score is 24-20* 20-16*
>13-9(2) )and now flunk (6-6) ..... now comes the double ..... this is an
>easy take at -2 -2 match score. The side getting doubled wins more than 33%
>of the games according to Jellyfish rollout at level 6. I am not saying it
>is wrong to double at -2 -2 after the first roll, but i can not find any
>position sequence of 3 rolls (his roll, your roll, his roll) where he would
>past at this match score. Have I missed something???? If not it seems there
>is a great reason to NOT double after the opening roll because as in the
>example above you opponent might pass.
Sounds like Hugh has a strong point here. However, my rollouts
show that 24/20*/16*, 8/4(2) wins the same number of games (63%) as
Hugh's suggested play (hit twice and make the 9-point). Making the 4-point
does a little better at cubeless money play. (NOTE: Hugh then looked
at an entry-failure roll which is why he has responder winning 67%.)
I wonder what opener's chances are after entry failure with THIS play.
S/he may still have more than 30% chances, and if not, it must be
pretty close.
But I think I may have a couple three roll sequences which DO leave
opener under 30%. How about 42 (24/20, 24/22), 33, fan? And 31 (24/21,
24/23), 44, fan? (Just thought I'd infuse a bit of humor...) Actually
some people (probably not many) like the 24/20, 24/21 opening with 43.
44 (8/4*(2),13/5*) followed by entry failure--does that work? Then there
is the classic: 52 played 13/8, 24/22, 55 response, fan. Don't scoff.
This choice of playing the 52 does well in the rollouts at all match
scores. Still, opener may be above 30%, even with this "oh, woe is me"
sequence.
In any case, I'm really clawing hard trying to find market losers
here, which I think is Hugh's contention.