Opening moves in a 7 point match at all possible scores

2 views
Skip to first unread message

Back4U2 BBL

unread,
Jul 27, 2002, 11:11:05 PM7/27/02
to
I analysed the opening moves in a 7 point match with Snowie (3Ply, Huge,
50%).
From match score 0-0 up to to 6-6.
The equities are the average equities of that move at all possible scores.

65: 24/13 (0.000), 24/13 13/8 (-0.043)

24/13 13/8 comes close (-0.014 - -0.020) at scores:
5-6, 5-6 Cr(awford), 4-6, 3-6, 3-6 Cr, 1-6 Cr and 0-6

64: 24/14 (-0.006), 24/18 13/9 (-0.012), 8/2 6/2 (-0.020)

24/14 is always best at equal scores and while leading,
is also best trailing at most scores,
but is an error at scores:
5-6 Cr (-0.040) (but best move at 5-6), 3-5 (-0.068), 4-6 (-0.041), 3-6
(-0.037).
8/2 6/2 is to play at 5-6 Cr and 3-6 Cr (but *not* at other Cr scores), 3-5,
4-6, 3-6, and 1-6.
24/18 13/9 is best (or close to best) at 4-5 and trailing with 3 points (but
not at 3-6)

63: 24/18 13/10 (-0.003), 24/15 (-0.005)

24/18 13/10 if favorite while trailing. Also if leading or equal up to a
score of 3 (1-x, 2-x, 3-x).
24/15 is favorite at other scores.

62: 24/18 13/11 (0.000), 24/16 (-0.023)

24/18 13/11 is best at all scores.

61: 13/7 8/7 (0.000)

Only other possible move (-0.044) is 24/18 24/23 at score 5-6.

54: 24/20 13/8 (0.000), 13/8 13/9 (-0.025), 24/15 (-0.033)

24/20 13/8 is always good.
24/15 is only best at 5-6.
13/8 13/9 comes in first at 5-6 Cr, 3-5 and 4-6.

53: 8/3 6/3 (-0.001), 24/21 13/8 (-0.021)

8/3 6/3 is OK at all scores.
Yet: 24/21 13/8 is best at 5-5, 6-5 Cr, 6-2 (Cr), 6-1 (Cr), 6-0 (Cr)

52: 24/22 13/8 (-0.004), 13/8 13/11 (-0.007)

24/22 13/8: when leading or equal
13/8 13/11 while trailing, but *not* at 5-6, 4-6 Cr, 2-6 Cr and 0-6 Cr (but
the equity is max -0.008)

51: 24/23 13/8 (0.000), 13/8 6/5 (-0.032), 24/18 (-0.037)

At all scores: 24/23 13/8.
13/8 6/5 comes close at 6-5 (cube is coming) and 3-5.

43: 24/20 13/10 (-0.003), 24/21 13/9 (-0.007), 13/9 13/10 (-0.008), 24/20
24/21 (-0.012)

Leading and equal, 24/20 13/10 is always best or close best. 24/21 13/9 is
second there.
Trailing....
24/20 24/21 at score 5-6.
13/9 13/10 is best at other scores, but *not* 4-6 Cr, 2-6 Cr, 1-6 and 0-6
Cr.
24/20 13/10 is error at 3-5.

42: 8/4 6/4 (0.000)

Only at score 5-6 24/20 13/11 is thinkable (-0.042)

41: 24/23 13/9 (0.000), 13/8 (-0.022), 24/20 24/23 (-0.024)

24/23 13/9 at all scores, except (again!) 5-6.
24/20 24/23 is best at 5-6.

32: 24/21 13/11 (-0.001), 13/10 13/11 (-0.014), 24/22 13/10 (-0.026)

24/21 13/11 covers for all scores.
13/10 13/11 is better at 5-6 Cr, 3-5, 4-6, 3-6, 2-6, 1-6 Cr and 0-6.

31: not typing this one.

21: 24/23 13/11 (-0.001), 13/11 6/5 (-0.016)

24/23 13/11 for almost all scores.
13/11 6/5 at 3-5 (+0.028) and 4-5 (+0.015).

Overall....
It's remarkable ... at least for one being new to backgammon :
for scores up to 3 for both players (x-y with 0 <= x <= 3 and 0 <= y <= 3)
one notices the same moves coming in as best.

Once one player reaches -3, -2 or -1, there are some 'surprises'.

Especially the trailer - while at -4 and -2 - has to be carefull to play
'automatically'. Some 'normal' moves are errors here.

Crawford and post-crawford games ask for 'be carefull' (no gammon needed for
leader, but gammon could come handy for trailer).

Remark:
Don't shoot me if some figures are wrong.
I just finished this. It is not 'full proof' yet.

Louis Nardy Pillards
BBL


moorg

unread,
Jul 28, 2002, 9:38:58 AM7/28/02
to
great stuff!

Thanks for your effort! :-)

"Back4U2 BBL" <nardy.p...@skynet.be> wrote in message
news:3d4360c6$0$180$ba62...@news.skynet.be...

Gregg Cattanach

unread,
Jul 29, 2002, 7:05:25 AM7/29/02
to
Good work. But some long rollouts would tell a more accurate story, even
though 3-ply Huge probably gets them 'right' anyway. I suggest lots of
games (>500) at 3-ply small, 33% and truncate at 9 or 11 moves.

Gregg

"Back4U2 BBL" <nardy.p...@skynet.be> wrote in message
news:3d4360c6$0$180$ba62...@news.skynet.be...

Back4U2 BBL

unread,
Jul 29, 2002, 10:33:45 AM7/29/02
to

"Gregg Cattanach" <gcattana...@prodigy.net> wrote in message
news:Vj919.5354$Zk3.71...@newssvr15.news.prodigy.com...

> Good work. But some long rollouts would tell a more accurate story, even
> though 3-ply Huge probably gets them 'right' anyway. I suggest lots of
> games (>500) at 3-ply small, 33% and truncate at 9 or 11 moves.
>
> Gregg

Thanks.

The opening moves I 'selected' first, were the seven best moves at a score,
having an equity up to -0.150.
Afterwards I deleted all moves that got a -0.150 or up (down?) error at all
scores.
Only the 21 opening has over 7 moves at some scores.

I started with the rollouts, both with GNUbg and Snowie. But I didn't think
of doing rollouts for the -0.090 and up errors. Correct me if I see this
wrong.

And also: moves within -0.005.... you really think rollouts could 'widen'
that equity? (We are talking about the first move in a game).

... and yes (sighing here):
The answer to that opening move? (21 rolls to consider here)
I do have those at score 0-0. I'm not looking forward to the weeks (months?)
it will take to get them for all scores.
LOL. Wish me strenght?

Louis Nardy Pillards


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.380 / Virus Database: 213 - Release Date: 24/07/2002


Martin Hemming

unread,
Jul 29, 2002, 7:19:25 PM7/29/02
to
I'm not looking forward to the weeks (months?)
> it will take to get them for all scores.
> LOL. Wish me strenght?
>
> Louis Nardy Pillards
>
> > I do indeed. And it's very public-spirited of you to post the results of your work. But is what you're doing the best use of your (presumably limited) analysis time?

First, what you do in the opening moves of a game at scores below -4
-4 is not going to matter as much as opening moves in games at higher
match scores. And even opening moves in end-of-match games are dwarfed
in importance by later moves in all games.

Second, the problem with trying to learn the "best" openings is that
the bots assume you (and your opponent) will continue to play just as
well no matter what type of game ensues. Humans are not like that so
it's best to avoid setting up game types that you are less comfortable
with and seek game types that your opp is less comfortable with. Most
of the books advise this too. Getting into a strategic battle where
you're uncomfortable is going to lose more equity than making the
'wrong' opening move or even the 'wrong' reply.

Martti

unread,
Jul 29, 2002, 7:35:42 PM7/29/02
to

"Martin Hemming" <mhem...@lineone.net> wrote in message
news:4e9131b9.02072...@posting.google.com...

I disagree. I think that attempting to steer games into types that
don't fit the rolls loses more equity then you'd gain by trying for
types you are more familiar with.

I also think that getting off to a good solid start is very
important and that the equity differences, while small, at the
beginnings of games are crucial since you see them over and over
again so in the long run those small differences really add up.
Every game has a beginning. Best learn how to play it.

Martti

unread,
Jul 29, 2002, 7:40:48 PM7/29/02
to

"Martti" <mar...@spamremovejunkbellsouth.net> wrote in message
news:d4k19.76208$Og3.17...@e3500-atl1.usenetserver.com...

> I disagree. I think that attempting to steer games into types that
> don't fit the rolls loses more equity then you'd gain by trying
for
> types you are more familiar with.
>
> I also think that getting off to a good solid start is very
> important and that the equity differences, while small, at the
> beginnings of games are crucial since you see them over and over
> again so in the long run those small differences really add up.
> Every game has a beginning. Best learn how to play it.
>
>

I just realized I do agree with part of the post I just disagreed
with. I have started to play several of the opening moves
differently then what the best roll outs I've seen have indicated.
Like bringing down two instead of splitting on some rolls. I started
doing this just to learn how to play them but I feel I am successful
sometimes in large part because players don't see that approach as
often as they used to and are not as good at playing against that
style. I have some "alternate moves" for the opening that I really
love.

Gregg Cattanach

unread,
Jul 30, 2002, 7:05:15 AM7/30/02
to
Doing all the 2nd roll rollouts and transcribing them took me 7 months
(often using 2 computers to do the rollouts). I did a total of 2504
rollouts and recorded the results for all 4 match scores ($$, DMP, GS, GG).
It might be interesting to see if Snowie 4 gets any different results; she
apparently plays priming and backgames better and might not be so 'fearful'
of slotting plays.

Gregg

"Back4U2 BBL" <nardy.p...@skynet.be> wrote in message

news:3d455652$0$206$ba62...@news.skynet.be...

Back4U2 BBL

unread,
Jul 30, 2002, 12:36:36 PM7/30/02
to

"Martin Hemming" <mhem...@lineone.net> wrote in message
news:4e9131b9.02072...@posting.google.com...

I do agree Martin.
And this was one of the reasons I started it.
There is no need for a World Class player to know the -.002 difference
between two opening moves.

Yet... for the weaker player (advanced?) it does matter knowing the 21
opening slot or split is best move or huge error at some scores of a match.

It does matter to *know*. It is up to him/her to do what he/she wants to do
with it :)
Maybe to prevent the World Class opponent to get his/her game too easily?

BTW: GNUbg opens a lot more aggressive than Snowie.

Louis Nardy

Back4U2 BBL

unread,
Jul 30, 2002, 12:38:05 PM7/30/02
to

"Martti" <mar...@spamremovejunkbellsouth.net> wrote in message
news:_8k19.76258$Og3.17...@e3500-atl1.usenetserver.com...

> I just realized I do agree with part of the post I just disagreed
> with. I have started to play several of the opening moves
> differently then what the best roll outs I've seen have indicated.
> Like bringing down two instead of splitting on some rolls. I started
> doing this just to learn how to play them but I feel I am successful
> sometimes in large part because players don't see that approach as
> often as they used to and are not as good at playing against that
> style. I have some "alternate moves" for the opening that I really
> love.

Maybe post those "alternate moves"?

Louis Nardy

Martti

unread,
Aug 1, 2002, 1:52:51 AM8/1/02
to

"Gregg Cattanach" <gcattana...@prodigy.net> wrote in message
news:Lpu19.14510$YR.48...@newssvr16.news.prodigy.com...

> Doing all the 2nd roll rollouts and transcribing them took me 7
months
> (often using 2 computers to do the rollouts> >
> > ---

Catt, you're my hero... :-)

Reply all
Reply to author
Forward
0 new messages