2 Post-Crawford games: why are these passes?
Scott Steiner
I have a question concerning post crawford doubling strategy. I was
playing a match to 7 against gnubg at the preset expert level. We just
finished the crawford game and this is the first game post crawford.
Gnubg is leading 6-3. I double at the first occassion:
gnu (O, 6 pts) vs. user (X, 3 pts) (Match to 7)
Game number 6
Move number 4: X doubles to 2
GNU Backgammon Position ID: 4NvgATDgc/ADIA
Match ID : cBHgAGAAGAAA
+13-14-15-16-17-18------19-20-21-22-23-24-+ O: gnu
| X O O | | O X | 6 points
| X O O | | O | Cube offered at 2
| X | | O |
| X | | O |
| 6 | | O |
v| |BAR| |
| | | X |
| O | | X |
| O X | | X |
| O X | | X O |
| O X | | X O | 3 points
+12-11-10--9--8--7-------6--5--4--3--2--1-+ X: user
Pip counts: O 160, X 156
* gnu accepts
Alert: wrong take ( -0,020)!
Cube analysis
2-ply cubeless equity +0,179 (Money: +0,052)
0,523 0,120 0,004 - 0,477 0,116 0,002
1. Double, pass +1,000
2. Double, take +1,020 ( +0,020)
3. No double +0,902 ( -0,098)
Proper cube action: Double, pass
Output generated Fri Nov 29 16:58:52 2002
by GNU Backgammon 0.12 (Text Export version 1.27)
So it seems that gnubg made an error here, analysis is preset to
worldclass++. My question here is why is this a free drop? Is O's game
here so unpromising and too risky? Incidently, 3ply also says it's a
drop and by a wider margin (double, take is +0.061 there).
So gnubg takes and I win the game making the score 6-5. In the next
game I again double on my first occassion:
gnu (O, 6 pts) vs. user (X, 5 pts) (Match to 7)
Game number 7
Move number 4: X doubles to 2
GNU Backgammon Position ID: mGfwATDgc+IBIQ
Match ID : cBHgAGAAKAAA
+13-14-15-16-17-18------19-20-21-22-23-24-+ O: gnu
| X O | | O X O X | 6 points
| X O | | O O | Cube offered at 2
| X | | O |
| X | | O |
| | | |
v| |BAR| |
| O | | X |
| O | | X |
| O X | | X |
| O X | | X O |
| O X X | | X O | 5 points
+12-11-10--9--8--7-------6--5--4--3--2--1-+ X: user
Pip counts: O 161, X 160
* gnu rejects
Cube analysis
2-ply cubeless equity +0,323 (Money: +0,068)
0,524 0,138 0,006 - 0,476 0,120 0,004
1. Double, pass +1,000
2. Double, take +1,096 ( +0,096)
3. No double +0,930 ( -0,070)
Proper cube action: Double, pass
Output generated Fri Nov 29 16:59:47 2002
by GNU Backgammon 0.12 (Text Export version 1.27)
This time gnubg rejects my double, and correctly as the analysis says
(3ply says the same and again by a wider margin). My question here is
again, why is this a free drop? Is O's game again here so unpromising
and too risky? O even made his 4 point and has a stronger home board
than me.
That these 2 decisions are both drops baffled me a little, but then
again I'm no expert on cube handling. Any comments that puts more light
on the matter are mostly appreciated.
thx!
Gregg Cattanach
post-Crawford cube when you are an even number of points away. This can be
modified a bit if you are 6-away or 8-away as you make get the opportunity
to drop as a bigger underdog later.
In position #1, the 4 pip lead and that X is playing with only one man back
makes him the favorite, and in position #2, X is threatening to make good
home board points or the 20 point anchor (and most importantly IS ON ROLL);
in both O is the underdog so Gnu drops. You don't have to have an
'unpromising' game, just if you are any kind of underdog. Generally losing
the opening roll and then failing to roll doubles or point on your opponent
is enough to pass these post-Crawford cubes.
Gregg C.
"Scott Steiner" <nos...@nospam.nospam> wrote in message
news:3DE793DF...@nospam.nospam...
Scott Steiner
>
> Simply put, any time you are ANY kind of an underdog, you should drop a
> post-Crawford cube when you are an even number of points away.
Let me see if I understood this properly. You mean any time the leader
is any kind of underdog, he should drop a post-crawford cube when the
trailer is an even number of points away, is that right? I don't
understand the even points part though. If in position #1 (7 point
match) the score was 6-4 or 6-2 instead of 6-3, would that be a take
then instead and why?
--
Email replies to following address: bpp at chello dot at
Hugh McNeil
at an even-away, Post-Crawford score. Without thinking much about it, I
could probably find you about a dozen examples in ten minutes (with a small
wager attached, of course...). No tricks, the statement is simply wrong.
Hugh
"Gregg Cattanach" <gcattana...@prodigy.net> wrote in message
news:U%1G9.251$Pp5.19...@newssvr15.news.prodigy.com...
Adam Stocks
post-crawford game, it will be assumed at the outset that some 'default'
gammon rate will apply (eg 26% for Snowie's match equity table), and that
the leader only has a finite number of post-crawford 'chances' to win the
one game he needs. Therefore, if the current post-crawford game turns into
a non-gammonish position, then that non-gammonish game is a more valuable
than average chance, hence a lower takepoint. (the gammon rate for the
current game needs to be significantly lower than average for the takepoint
to be lower than 50%, in order to outway the average gammon value for the
other post-crawford games).
Adam
"Hugh McNeil" <tall...@sympatico.ca> wrote in message
news:Lz5G9.67887$e%.1788931@news20.bellglobal.com...
Gregg Cattanach
"Scott Steiner" <nos...@nospam.nospam> wrote in message
> Let me see if I understood this properly. You mean any time the leader
> is any kind of underdog, he should drop a post-crawford cube when the
> trailer is an even number of points away, is that right? I don't
> understand the even points part though. If in position #1 (7 point
> match) the score was 6-4 or 6-2 instead of 6-3, would that be a take
> then instead and why?
>
that immediately reduces the number of games the trailer needs to win to win
the match. If the trailer has an odd number to go, you must take.
I still believe with 2 points to go, if you are more than a 51.5% underdog
(the value of the free drop has been calculated to be around 1.5%), you
should ALWAYS drop regardless. If you have 4, 6 or 8 to go, you may retain
your free drop for a future game if you are just a tiny underdog in the
current game and go ahead and take.
HUGH: I'd be interested to see any position at 2-away 1-away post Crawford
where the leader is the underdog and Snowie or Gnu still recommends a take.
Gregg
Maik Stiebler
> Let me see if I understood this properly. You mean any time the leader
> is any kind of underdog, he should drop a post-crawford cube when the
> trailer is an even number of points away, is that right? I don't
> understand the even points part though. If in position #1 (7 point
> match) the score was 6-4 or 6-2 instead of 6-3, would that be a take
> then instead and why?
>
The difference between post-Crawford odd-away and post-Crawford
even-away cannot be stressed enough. At odd-away, the leader actually
has a "free TAKE"!
Would you ever drop if you were doubled at 6-6 in a 7-point-match?
Well, dropping while leading 6-4 is usually nearly as bad. If you
cannot be gammoned, you can take with game winning chances as small as
3-4%.
It's quite common for the trailer to delay doubling until he thinks
his hopefully less knowledgeable opponent will erroneously pass. The
leader can avoid this trap by deciding (and maybe even announcing)
before the game that he will take trailer's double no matter when it
comes. Doing so would not reduce leader's equity, though it is not
optimal in the case that trailer errs by losing his market.
Gregg Cattanach
news:60225e11.02120...@posting.google.com...
If the trailer (at 3-away 1-away post-Crawford) delays doubling, there are
positions that the leader should drop. I think if you end up in a position
where the trailer is winning the game 66% of the time and gammons 33% of the
time, then the gammon threat leads to a proper drop by the match leader. So
it would be wrong for the leader to take EVERY cube no matter when it comes.
This ploy (waiting for the position to become gammonish before doubling at
odd-away, 1-away post-Crawford) can net you a free point or two with an
improper drop, but the flip side is that you do risk losing your market if
it becomes too gammonish too fast.
Gregg C.
Scott Steiner
>
> "Scott Steiner" <nos...@nospam.nospam> wrote in message
> news:3DE8B03E...@nospam.nospam...
> > Let me see if I understood this properly. You mean any time the leader
> > is any kind of underdog, he should drop a post-crawford cube when the
> > trailer is an even number of points away, is that right? I don't
> > understand the even points part though. If in position #1 (7 point
> > match) the score was 6-4 or 6-2 instead of 6-3, would that be a take
> > then instead and why?
> >
> First, if the score is 6-4/7 or 6-2/7, you NEVER have a free drop, because
> that immediately reduces the number of games the trailer needs to win to win
> the match. If the trailer has an odd number to go, you must take.
>
> I still believe with 2 points to go, if you are more than a 51.5% underdog
> (the value of the free drop has been calculated to be around 1.5%), you
> should ALWAYS drop regardless. If you have 4, 6 or 8 to go, you may retain
> your free drop for a future game if you are just a tiny underdog in the
> current game and go ahead and take.
Thanks a lot Gregg! I gave the subject a bit of thought after reading
your post and now it makes all sense. I'm sure this will improve my
game. Thanks again :-)
>
> HUGH: I'd be interested to see any position at 2-away 1-away post Crawford
> where the leader is the underdog and Snowie or Gnu still recommends a take.
Hmmm, I would like to see that too now...should what I've just learned
be shattered to pieces again? ;-)
>
> Gregg
Hugh McNeil
"Gregg Cattanach" says...:
> HUGH: I'd be interested to see any position at 2-away 1-away post
Crawford
> where the leader is the underdog and Snowie or Gnu still recommends a
take.
>
> Gregg
Being bots, I suspect with some industriousness you could find a few.
However, I can come up with several examples of real life situations where
you would agree that the leader is an underdog in the game, and the leader
should take, with the match score 1A, 2A....
Gregg Cattanach
looking at the cube being turned at the first opportunity for the trailer.
It just seems odd that as more than a 51.5% underdog (somewhere I read that
the free drop is worth 1.5%), that you'd ever want to take this cube.
Gregg C.
Hugh McNeil
"Gregg Cattanach" <
> > > HUGH: I'd be interested to see any position at 2-away 1-away post
> > Crawford
> > > where the leader is the underdog and Snowie or Gnu still recommends a
> > take.
> > >
> > > Gregg
> >
> >
> > Being bots, I suspect with some industriousness you could find a few.
> > However, I can come up with several examples of real life situations
where
> > you would agree that the leader is an underdog in the game, and the
leader
> > should take, with the match score 1A, 2A....
> >
> >
> I'd definitely be interested in seeing such a position. I'm assuming
we're
> looking at the cube being turned at the first opportunity for the trailer.
> It just seems odd that as more than a 51.5% underdog (somewhere I read
that
> the free drop is worth 1.5%), that you'd ever want to take this cube.
>
> Gregg C.
That would be a poorish assumption. It's my impression that you are
generally correct that if the cube is being turned at the first available
opportunity, then you should drop with any deficit in the position (as the
Leader, when the Trailer is an Even number of points away from winning).
However, if the Trailer errs, and waits even just one turn before cubing,
then interesting things happen.... not least of which, the leader may make
an advanced anchor, which reduces the gammon risk. This allows the leader to
take a cube with only a ~42% Game winning chance when the score is 4A-1A.
For a two Away-one Away score...imagine yourself in a simple-to-play
position (ie you are't going to make any mistakes in the rest of the game),
where you figure you only have a 47% Game Winning Chance. Playing a very
strong checker player, Grandell for eg, do you take, or do you start a new
and final game? Methinks you are better off taking...
Conversely, should Grandell take a last roll cube (2A-1A) where he was a
19-17 favorite against you?
My point was that while the rule of dropping with any positional deficit
when Even Away, Post Crawford, is usually valid (out of the opening rolls),
you do have to know why it's correct in a specific position.
Douglas Zare
Gregg Cattanach wrote:
> "Hugh McNeil" <tall...@sympatico.ca> wrote in message
> news:KDSG9.3399$mj2.4...@news20.bellglobal.com...
> > However, I can come up with several examples of real life situations where
> > you would agree that the leader is an underdog in the game, and the leader
> > should take, with the match score 1A, 2A....
Apparently you mean opponent considerations, but I had thought you meant
positions in which one is a money underdog due to gammons or poor cube
efficiency, but a DMP favorite.
> I'd definitely be interested in seeing such a position. I'm assuming we're
> looking at the cube being turned at the first opportunity for the trailer.
> It just seems odd that as more than a 51.5% underdog (somewhere I read that
> the free drop is worth 1.5%), that you'd ever want to take this cube.
The value of the free drop is irrelevant (but probably smaller than the 1.5% in
Snowie's table). You never get to post-Crawford 2-away from post-Crawford
2-away. The choice is DMP or playing the game for the match, so the take point
is 50%.
Much more interestingly, the racing take point is lower than 50% at
post-Crawford 4-away (35% according to Snowie's MET). The gammon price is very
high on a 2-cube, though. With proper play, there can be real cube decisions
after the first few rolls because some opening exchanges leave the trailer too
good to double.
Douglas Zare
Larrikin
> If the trailer (at 3-away 1-away post-Crawford) delays doubling, there are
> positions that the leader should drop. I think if you end up in a position
> where the trailer is winning the game 66% of the time and gammons 33% of the
> time, then the gammon threat leads to a proper drop by the match leader. So
> it would be wrong for the leader to take EVERY cube no matter when it comes.
>
> This ploy (waiting for the position to become gammonish before doubling at
> odd-away, 1-away post-Crawford) can net you a free point or two with an
> improper drop, but the flip side is that you do risk losing your market if
> it becomes too gammonish too fast.
>
> Gregg C.
This begs the question: Can you be "too good to double" in this or
similar odd-away situations? Winning 2 points does not help the
trailer much in terms of MWC. But winning an undoubled backgammon
certainly does. Therefore, if Trailer looses his market, as in the
case above, should he play on hoping for a backgammon? I think that
as long as you don't risk an exchange where you let the Leader back
into "take" territory you could reasonably play on the backgammon.
Ian Dunstan.
Gregg Cattanach
"Larrikin" <larri...@optusnet.com.au> wrote in message
news:8c2ca95f.02120...@posting.google.com...
I think playing on for the BACKgammon regardless of the score has a very low
expected value. The real solution is to not lose your market by waiting for
SOME gammon threat, doubling there and hope for a bad pass. If you do lose
your market because of a large gammon threat, you'll gain far more by
cashing (which was what you were hoping for when you delayed doubling,
anyway), then trying to win some low percentage backgammon.
Gregg C.
Douglas Zare
Yes, definitely you can be too good to double. This is true even without the possibility of backgammons. In some positions, you
will have a huge gammon threat even after a setback, so your opponent will be unable to take after any exchange, so you should
play on for the gammon. With enough backgammon potential, it may be worth it to risk regaining your market.
If you close 3 checkers out in ordinary circumstances (money play with your opponent holding the cube) and ideally placed
spares, you should win more than 8% backgammons. It depends on the state of your opponent's offense, thus how much you have to
avoid getting hit. With 4 checkers closed out you should win at least 25% backgammons. Most players, including current bots, do
not play aggressively enough for the backgammon. They view it as a freak occurence, even though when you close 3 out winning a
backgammon should be more common than losing. I've won 3 valuable backgammons in 14 live tournaments, including from trailing
2-away 5-away with the cube on 2 in a final, so I don't discount their importance.
By the time you close out 3 checkers, you are certainly too good to double at post-Crawford 3-away. The question is how you
could get to such a situation. You can't with perfect play by both sides, but I think it might be reasonable not to double if
you only have a couple of market losing exchanges in 1296 if you are sure that your opponent will take now. Then if you get one
of the market losers, you might be too good to double if the position is not volatile, e.g., you have manufactured a 6-prime
that will still be there next turn. (Be careful that a 6-prime is far from gin, though, so if the gammon chances go down your
opponent may have a take.)
At post-Crawford 5-away, backgammon wins are huge on a 2-cube. The difference between winning a backgammon and winning a gammon
is the same as the difference between winning a gammon and losing! Although I think there is more room to benefit from bad
passes of delayed doubles, a real risk of backgammons (count your opponent's blots) is a strong argument against actually
losing your market.
Douglas Zare