Bidding System Design Contest

[tys...@yahoo.com]

Every two months or so, someone on rgb asks "what is the best system?"
The same flurry of answers usually follows starting with "whatever
you can really remember and use effectively" and "it's the skill of
the pair and not the system." Then the thread usually spins into
theory and if is there an optimal system for two computers to use.
Maybe there is...

Let's have a contest to see who can design the best system (at least
opening bids)!

I'm not sure this can be answered, but I'm going to attempt to come
close. If you make a simplified model of bridge bidding you can use
information theory to quantitatively rate how good a system is. The
trick is to make this model as close to "real" bridge as possible and
not to make it too simplified of a model. Matt Ginsberg first
proposed doing this back in 1996 and I think his theory is mostly
sound. For details of how Matt's information theory system works,
look here:

http://tinyurl.com/27rql

Be warned that this is not for the faint of heart or the
non-mathematically inclined. For those who do not need the details,
it talks about a way to take definitions for bids and quantitatively
state how good of a definition it is. It takes into consideration
several factors of bidding including:

1. how the bid helps us find the optimal contract
2. how the bid helps refine the definitions of *other* bids
3. how the bid hinders our opponents from finding their optimal
contract

It does have some limitations, and so it does not address:

1. right-siding the contract (all contracts are assumed to be
right-sided)
2. lead directing inferences

If you feel that these last two points are minor compared to what you
get with the first three, then this contest might be right for you.

So I propose the following system generation contest to be conducted
here on rgb. There is no prize for winning except for the pride.
After it's over, everyone here on the list can look at which systems
were rated the highest and I think it will generate some healthy
debate. We can discuss what points the program feels are the best and
decide if they would really apply at the table. I think it can be a
good learning experience for everyone. Plus I think it will be a lot
of fun.

Anyone who wants to participate in this contest should post their
system on this thread (don't send me an email with it, only posted
systems will be judged). Only post definitions for OPENING bids
(anything more would be too much to handle in this format). Keep the
following things in mind:

1. Any definitions are allowed; you can be as artificial as you want.
Just be sure the definitions are accurate enough that I can program a
computer with them.
2. Clearly specify all meanings for all bids (don't forget pass).
Include ways to sort out which bid will be used when more than one
definition can apply. For example if you have one bid that shows 5+
spades and one that shows 5+ hearts, make sure I know which one to use
if you are 55, 56, 65, etc.
3. You can specify meanings for 3C and higher, but *you don't have
to*. Just list definitions from Pass to 2NT. Higher bids will be
given to the "unassigned" hand definitions and the computer will
intelligently assign definitions to them. Your system will preempt
more aggressively if you have more unassigned hands. Just to make
things clear, you can specify the hands that you will open 3C+ so I
make sure to capture them, but they won't be inputted in specifically.
These higher bids will really have no effect on your overall score
since they will likely be rare. Also since these bids often take
combinations of suit quality, exact shape, and other things it's hard
to program them effectively.
4. Scoring is IMPs and vulnerability is none.
5. Only 1 system per person
6. All entries must be posted by June 20, 2004.

Here's an example of how a Standard American system might look (and
the level of detail I need):

Count HCP and add 3/2/1 points for void/singleton/doubleton. Points
include distribution unless specified otherwise.
Pass = all 0-12 point hands with 0-5 in spades, hearts, and diamonds,
and 0-6 clubs. All hands with 0-4 HCP and a 7-card suit.
1C = 13-22 points, 3+ C
1D = 13-22 points, 3+ D
1H = 13-22 points, 5+ H
1S = 13-22 points, 5+ S
1NT = 15-17 HCP any 5332, 4432, or 4333 shape
2C = 23+ points, any shape
2D = 7-12 points, 6 D, no 4cM
2H = 7-12 points, 6 H, no 4cS
2S = 7-12 points, 6 S, no 4cH
2NT = 20-21 HCP any 5332, 4432, or 4333 shape
*For hands that have more than one definition: 1NT & 2NT take highest
priority. Otherwise longest suit in the hand takes priority. If
there is a tie for longest suit, then higher ranking suit takes
priority except in the case of 33 in the minors, then bid 1C.
*Note that the unassigned meanings for weak hands with 7+ suits will
be intelligently assigned to bids of 3C+.

I'm sure everyone here can be much more creative than this when you
are not bounded by memory or system restrictions. Let's get creative
here!

I've played around with several definitions, so let me give you some
hints as to what the computer will rate highly:
1. Shape is more important than strength. A very specific shape bid
with wide strength (within reason) will do better than a nebulous bid
with narrow strength.
2. Open light, especially unbalanced hands
3. Bids with multiple definitions will do okay if the meanings are
very different so that further rounds of bidding can differentiate
them. If the opponents preempt, will your partner be able to tell
which definition you have?
4. The opponents will always have the best defense prepared, so you
can't rely on unfamiliarity with the system!

I think that's it. Email me only if you have private questions,
otherwise let's try to keep everything posted publicly so that
everyone can learn.

[Gerben Dirksen]

Okay here's mine:

PASS: a) 0 - 7 (8 balanced)
b) 16+ unbalanced or 18+ balanced
1C: a) unbalanced 8 - 15 HCP, 4+Clubs
b) 12 - 14 balanced without 4D or 5H/S *
c) 15 - 17 balanced (no 5-card major)
1D: 8 - 15 HCP 4+D, unbal. or 12 - 14 balanced *
1H: 8 - 15 HCP 5+H, unbal or 12 - 14 balanced *
5H + 4S only if 12 - 15 HCP.
1S: 8 - 15 HCP, 5+S
1NT: 9 - 11 balanced (5H or 4H5m22 possible), 12 - 14 vulnerable
2C: 7 - 11 HCP, 4+H 4+S or 21 - 22 balanced
2D: 4 - 8 HCP, 5 - 5 not both minors
2H: 5 - 9 HCP, Weak two in H or S (no 4-card other major)
2S: 5 - 9 HCP, 4S + 5+ minor
2NT: 5 - 9 HCP, 5+C 5+D

* Vulnerable: swap 12 - 14 and 9 - 11 NT.
Priority: Higher bids first.

Gerben

[Mike Bell]

In message <a1bce703.0406...@posting.google.com>
tys...@yahoo.com wrote:

[snip]

> Let's have a contest to see who can design the best system (at least
> opening bids)!

[snip]

How is being in the even seats handled?

Cheers
Mike

[richard willey]

OK.

Here is a description of the Modified MOSCITO structure that I play in
first second seat. I'm dumping the dealer script that I normally use
for simulations. If you prefer I can provide a description in
English.

There is a minor glitch in the code. The definition of opening
strength should allow for hands with
c13(north) >= 500 if the length of opener's 2 longest suits is >= 10

#predeal
#south SJ, HAJ862, DQT, CKJT52

######### Definition - Opening Points

# 11 is the index for c13
# This function allows the user to define valuations for cacluating
HCP
# Most hand evaluation is performed used a modified version of the
# 4 Aces Point Count.(Ace = 3, King = 2, ...)
# All values are multiplied by 100 to avoid fractions

altcount 11 300 200 100 50 20

############## STRONG OPENINGS

one_club =

(
hcp(north)>=17
or
c13(north) >=1000
or
(hcp(north)>=15 and cccc(north)>=1600 and c13(north)>=900)
or
(hcp(north)==14 and cccc(north)>=1900 and c13(north)>=900)
or
(hcp(north)==13 and cccc(north)>=2000 and c13(north)>=900)
or

(
hcp(north)>=15 and
shape(north, any 4333 + any 4432 + any 5332 + any
5422)
and
c13(north) >= 900
)

)

#####
#Opening strength defines the minimum strength for a limited opening
bid
#####

opening_strength =
(
(
cccc(north) - 25 * shape(north, any 4441 + any 5440) >= 1050

or hcp(north) >= 11
)

and
c13(north) >= 600
)

and not

one_club

#MOSCITO Opening structure

#################### SINGLE SUITED PREEMPTS

#################### MAJORS
four_diamonds =

(
(
spades(north) >= 8 and
hascard(north, AS) + hascard(north, KS) == 2
)

or

(
spades(north) == 7 and
hascard(north, AS) + hascard(north, KS) + hascard(north,QS) == 3
)
)

and

hascard(north,AC) + hascard(north,KC) +
hascard(north,AD) + hascard(north,KD) +
hascard(north,AH) + hascard(north,KH) <=1

four_spades =

hcp(north) <= 9

and

(
(
spades(north) >= 7 and

hascard(north, AS) + hascard(north, KS) +
hascard(north,QS) + hascard(north, JS) >= 3
)

or

spades(north) >= 8
)

and not four_diamonds

three_spades =

spades(north) >= 6

and

(
hascard(north, AS) + hascard(north, KS) +
hascard(north, QS) + hascard(north, JS) >= 2
)

and

(
hearts(north) <=1 or
diamonds(north) <=1 or
clubs(north) <= 1
)

and

hcp(north) <= 9

and not four_diamonds

and not four_spades

four_clubs =

(
(
hearts(north) >= 8 and
hascard(north,AH) + hascard(north,KH) == 2
)
or

(
hearts(north) == 7 and
hascard(north,AH) + hascard(north,KH) + hascard(north,QH) == 3
)
)

and

hascard(north,AC) + hascard(north,KC) +
hascard(north,AD) + hascard(north,KD) +
hascard(north,AS) + hascard(north,KS) <=1

four_hearts =

hcp(north) <= 9

and

(
(
hearts(north) >= 7 and

hascard(north, AH) + hascard(north, KH) +
hascard(north,QH) + hascard(north, JH) >= 3
)

or

spades(north) >= 8
)

and not four_clubs

three_hearts =

hearts(north) >= 6

and

(
hascard(north, AH) + hascard(north, KH) +
hascard(north, QH) + hascard(north, JH) >= 2
)

and

(
spades(north) <=1 or
diamonds(north) <=1 or
clubs(north) <= 1
)

and

hcp(north) <= 9

and not four_clubs

and not four_hearts

############## MINORS

three_nt =

hcp(north) <= 9

and

(
(
clubs(north) >= 8
and
hascard(north,AS) + hascard(north,KS) +
hascard(north,AH) + hascard(north,KH) +
hascard(north,AD) + hascard(north,KD) <= 1
)

or

(
diamonds(north) >= 8
and
hascard(north,AS) + hascard(north,KS) +
hascard(north,AH) + hascard(north,KH) +
hascard(north,AC) + hascard(north,KC) <= 1
)
)

three_clubs =

clubs(north) >= 6

and

hascard(north, AC) + hascard(north, KC) + hascard(north, QC) == 2 and

hascard(north, AD) + hascard(north, KD) +
hascard(north, AH) + hascard(north, KH) +
hascard(north, AS) + hascard(north, KS) == 0 and

hcp(north) <=9

and not three_nt

three_diamonds =

diamonds(north) >= 6

and

hascard(north, AD) + hascard(north, KD) + hascard(north, QD) == 2 and

hascard(north, AC) + hascard(north, KC) +
hascard(north, AH) + hascard(north, KH) +
hascard(north, AS) + hascard(north, KS) == 0 and

hcp(north) <=9

and not three_nt

two_nt =

(clubs(north) >=6 and

hascard(north, AC) + hascard(north, KC) + hascard(north, QC) +
hascard(north, JC) == 2

and

hascard(north, AD) + hascard(north, KD) +
hascard(north, AH) + hascard(north, KH) +
hascard(north, AS) + hascard(north, KS) == 0

and

hascard(north, QD) + hascard(north, QH) + hascard(north, QS) <= 1

and

not three_clubs)

or

(diamonds(north) >=6 and

hascard(north, AD) + hascard(north, KD) + hascard(north, QD) +
hascard(north, JD) == 2

and

hascard(north, AC) + hascard(north, KC) +
hascard(north, AH) + hascard(north, KH) +
hascard(north, AS) + hascard(north, KS) == 0 and

hascard(north, QC) + hascard(north, QH) + hascard(north, QS) <= 1 and

not three_diamonds)

and not three_nt

preempts =
four_spades or
four_clubs or
three_spades or
four_hearts or
four_clubs or
three_hearts or
three_diamonds or
three_clubs or
three_nt or
two_nt

###############################
## Weak Opening Bids
###############################

two_diamonds =

not opening_strength and
hcp(north) <= 12 and
c13(north) >= 325 and

shape(north, any 4432, any 54xx, any 55xx, any 65xx) and
clubs(north) <= 3 and
diamonds(north) >=4

two_hearts =

not opening_strength and
hcp(north) <= 12 and
c13(north) >= 325 and

shape(north, any 4432, any 54xx, any 55xx, any 65xx) and
hearts(north) >= 4 and

(
spade(north) >= 4 or
clubs(north) >= 5
)

two_spades =

not opening_strength and
hcp(north) <= 12 and
c13(north) >= 325

and

(
(
shape(north, any 6322, any 6331, any 7222) and
spades(north) >=6
)

or

(
spades(north) >=4 and clubs(north) >=5
and hascard(north, AS) + hascard(north, KS) + hascard(north, QS) >= 1
)
)

############## LIMITED OPENINGS

##### Pre-Definitions

bad_spades = hascard(north, AS) + hascard(north, KS) + hascard(north,
QS) == 0
bad_hearts = hascard(north, AH) + hascard(north, KH) + hascard(north,
QH) == 0
balanced = shape(north, any 4432, any 5332, any 4333)

########

two_clubs =
opening_strength
and not preempts

and

(

(
clubs(north) >= 6 and
spades(north) < 4 and
hearts(north) < 3
)

or

(
clubs(north) >= 6 and
spades(north) == 4 and
bad_spades
)

or

(
clubs(north) >= 6 and
hearts(north) == 4 and
bad_hearts
)

)

one_notrump =

opening_strength and

(

(
balanced and not
shape(north, 5xxx, x5xx, 44xx, 4x4x, 4xx4, x44x, x4x4)
)

or

(
hearts(north) == 4 and
balanced and
(hcp(north) == 11 or hcp(north) == 12)
)

or

(
spades(north) == 4 and
balanced and
hearts(north) <= 3 and
(hcp(north) == 13 or hcp(north) == 14)
)

)

one_spade =
opening_strength
and not preempts
and not one_notrump

and

diamonds(north) >= 4 and
not one_notrump and
hearts(north) <4 and
spades(north) <4

or

(

diamonds(north) >=6 and

(
(spades(north) == 4 and bad_spades) or
(hearts(north) == 4 and bad_hearts)
)

)

one_heart =

opening_strength and
not two_clubs and
not one_notrump and
not one_spade and

(

(
spades(north) == 4 and
spades(north) > hearts(north)
)

or

(
spades(north) >= 5 and
spades(north) >= hearts(north)
)

)

one_diamond =

opening_strength and
not two_clubs and
not one_notrump and
not one_spade and
not one_heart

and

hearts(north) >= 4

and not

(balanced and hcp(north) <= 11)

##########

[Ron Lel]

This is what I have played and enjoyed immensly:

Pass = 13+ any
1C = 8-12 4+H possibly canape
1D = 8-12 4+S, possibly longer m
1H = 0-7 all, (with some 2 suited exceptions - see below), or 8 flat
1S = 9-12 no M
1N = 6+C 8-12
2C = 6+D 8-12
2D = weak 2 in H or 5S+5m 5-10
2H = weak in S or 5H+5m 5-10
2S = weak m pre empt or 5/5 Ms 5-10
2N = 5/5 ms 5-10

Weak and GF relays over all the 1 level openings, full Symmetric relay
engine
Over Pass 1D = GF, 1H = neg, rest = semi positives.

Ron Lel

[Marcel den Broeder]

REGRESsion system

Count HCP and deduct unguarded Jack, stiff Q or J, 1J in 2 guarded
J-combinations.
Any Ace and King counts as 8 points.
Distribution does not influence point count at all. Good or bad points
specially count in 4333/4432 and 0-12 range.

Pass: 13+ any
12+ any 1-suiter: 8 up to 12-carder
1C : 8-12 any 4441, 5440, 5431, 6421, 6430
6331, 7321, 7330, 5521, 5530
7-11 any 1-suiter: 8 up to 12-carder
1D : 0-6 any 1-suiter: 8 up to 12-carder
0-7 the rest (as stipulated NO any Ace and King)
1H : 8-12 4333, 4432, 5332 (no 2 toph. of 3), 5M422
1S : 8-12 Clubs: 5332 (with 2 Toph.), 5422, 6322
1NT : 8-12 Diam.: 5332 (with 2 Toph.), 5422, 6322
2C : 8-12 Hearts: 5332 (with 2 Toph.), 6322
2D : 8-12 Spades: 5332 (with 2 Toph.), 5422, 6322
2H : 8-12 any 7222 (if 8 rubbish ànd V/nv treat as 6322)
2S : 7-9 7/8/9c Clubs + 4c?
10-11 4c Clubs + 7/8/9c ?
7-11 6+/5+ Clubs/Diamonds or vv
2NT : 7-9 7/8/9c Diam. + 4c?
10-11 4c Diam. + 7/8/9c ?
7-11 6+/5+ Diamonds/Hearts or vv

Regards,

Marcel

[Marcel den Broeder]

REGRESsion system

Count HCP and deduct unguarded Jack, stiff Q or J, 1J in 2 guarded
J-combinations.
Any Ace and King counts as 8 points.
Distribution does not influence point count at all. Good or bad points
specially count in 4333/4432 and 0-12 range.

Pass: 13+ any
12+ any 1-suiter: 8 up to 12-carder
1C : 8-12 any 4441, 5440, 5431, 6421, 6430
6331, 7321, 7330, 5521, 5530
7-11 any 1-suiter: 8 up to 12-carder
1D : 0-6 any 1-suiter: 8 up to 12-carder
0-7 the rest (as stipulated NO any Ace and King)
1H : 8-12 4333, 4432, 5332 (no 2 toph. of 3), 5M422
1S : 8-12 Clubs: 5332 (with 2 Toph.), 5422, 6322
1NT : 8-12 Diam.: 5332 (with 2 Toph.), 5422, 6322
2C : 8-12 Hearts: 5332 (with 2 Toph.), 6322

2D : 8-12 Spades: 5332 (with 2 Toph.), 6322

[Michael Clark]

On 2 Jun 2004 15:13:28 -0700, tys...@yahoo.com wrote:

>3. Bids with multiple definitions will do okay if the meanings are
>very different so that further rounds of bidding can differentiate
>them. If the opponents preempt, will your partner be able to tell
>which definition you have?

We don't have to specify the responses or continuations to these
opening bids, but does there have to be some logic behind it? For
instance, could I specify:

1NT = 15-17 balanced OR weak two in clubs

These two meanings are nicely differentiated and if partner knows
which one you have you'll be in a happy position. However, it's
obviously impossible to have a sensible set of responses to this.
Compare this to:

2C = 20-22 balanced OR weak two in hearts

Again, these meanings are quite distinguishable - there's certainly no
overlap! This time, though, it's quite easy to work out some responses
and you can even invite in hearts below the 2 level.

My question is: can your simulation tell that the 1NT definition is
rubbish and the 2C definition is fine?

Michael

[richard willey]

Here is the description of the 3rd seat opening structure that I
normally use playing MOSCITO

Bids from 2N+ As before

2S = 7-11 HCP, 5+ Spades and 5+ cards in either minor

2H = 7 - 11 HCP. 5+ hearts and 5+ cards in either minor

2D = 21-22 balanced or
17+ HCP with 4441 shape or
Weak 2 in Hearts or
Weak 2 in spades

2C = 6+ Clubs with 11+ - 16 HCP or
5+ Clubs and 4 cards in another suit with 14- 16 HCP

1N = 16-17 HCP balanced or
13-15 HCP with 3334 or 3325 shape

1S = 4+ Spades, might have a longer suit, 11+ - 16 HCP
show spades in preference to Hearts, Diamonds, or Clubs with

4432 shape

1H = 4+ Hearts, might have another suit, 11+ - 16 HCP
Show Hearts in preference to Diamonds or Clubs with
4432 shape

1D = 3+ Diamonds, might have longer suit

1C = 17+ HCP, forcing

In fourth seat, the structure is preety much the same with the
following exceptions

2N = 21-22 HCP
2M = Acol 2 bid

[Michal Rosa]

tys...@yahoo.com wrote:
> Every two months or so, someone on rgb asks "what is the best system?"
> The same flurry of answers usually follows starting with "whatever
> you can really remember and use effectively" and "it's the skill of
> the pair and not the system." Then the thread usually spins into
> theory and if is there an optimal system for two computers to use.
> Maybe there is...
>
> Let's have a contest to see who can design the best system (at least
> opening bids)!

[...]

Any well structured WOS - Regres, Delta, No Name for example.

--
Windows Media Player is inactive...

Michal Rosa - GG 1081218

[DavJFlower]

Here's mine - not a million miles different from precision:

1C 16+HCP any distribution
1D & 2C: 11-15HCP 6-card suit, or 5 with another 4-card suit
1H & 1S: 11-15HCP 5-card suit
1NT: 13-15 balanced (May be 2254 or 2245)
2D: 11-15 Roman
2NT: 22-24HCP

Dave Flower

[Marcel den Broeder]

Marc...@chello.nl (Marcel den Broeder) wrote

> Any Ace and King counts as 8 points.

may be confusing, meant is: Ace + any King

> Distribution does not influence point count at all.

because of strictly separation of balanced (=4333 up to 7222) and
singl/void shapes for all openings/pointranges.
S/V shapes have priority in 1st. shape-type call when reaching a
certain pointrange

> 2D : 8-12 Spades: 5332 (with 2 Toph.), 5422, 6322

paste error: 5332 (with 2 Toph.), 6322

Completely Symmetric Design.
Singl/void hand is never the Relayer taking into account the limits of
the two pointranges involved.
Interference handling as much as possible system-on.

Sorry Tyson for event. inconvenience, but above worthwhile to be
mentioned imo

[tys...@yahoo.com]

> How is being in the even seats handled?

I guess I forgot to mention that this is for 1st seat only. Let's see
what kind of results we get for this first run, then we can see if
there are other situations we'd like to get into.

[Tony Warnock]

Opening bids: (Uses HCP means 4321, LTC adjusted as players see fit,
strengthe requirements are only approximate as everyone knows anyway,
just "points" means Gorens 321 short suit adjustments and whatever is
wanted for excess Aces and unprotected honors)

1C: 12-20 Points primary suit Clubs (7 or fewer losers)
15-20 HCP Balanced; reason to choose Clubs
14-20 Points with 4+Clubs and 5+ in other suit (6 or fewer losers)

1D: 12-20 Points primary Diamonds (7 or fewer losers)
15-16 HCP Balanced 5 Diamonds
17-20 HCP Balanced; 4+Diamonds
12-20 Points 4+ Diamonds, 5+Clubs (7 or fewer losers)
15-20 Points 4+ Diamonds, 5+Major (KQxxx or AJ10XX at least, 6 or
fewer losers)

1H: 12-20 Points Primary Hearts (7 or fewer losers)
15-16 HCP Balanced 5 Hearts
17-20 HCP Balanced 4+Hearts
12-20 Points, 4+Hearts, 5+minor (7 or fewer losers)
15-20 Points, 4+Hearts, 5+Spades (KQxxx or AJ10xx at least, 6 or
fewer losers)

1S: 12-20 Points Primary Spades (7 or fewer losers)
17-20 HCP Balanced 4+Spades
12-20 Points, 4+Spades, 5+other (7 or fewer losers)

1NT: 13-14 HCP, Balanced (4333, 4432,5332) (really good 12 HCP)

2C: 23+HCP Balanced (4333,4432,5332,5422)
21+HCP 3 or fewer losers if unbalanced

2D: 17-24 HCP, 4441 distribution (any singleton)

2H: ACOL, 15+HCP (at least 1QT outside Hearts), 6+Hearts or 5+Hearts
& 5+other

2S: ACOL, 15+HCP (at least 1QT outside Spades), 6+Spades or 5+Spades
& 5+other

2NT: 21-22 HCP Balanced

Limit raises in Majors

2 over 1 game forcing (except 2C over 1D as in KS).
1C-2C and 1D-2D game forcing.

1NT forcing response to 1H or 1S.
1NT weak over 1C or 1D (as in KS.)

Void & singleton raises to all opening 1 level suit bids. (Suit just
above jump raise shows unspecified singleton, relays to identify;
higher jumps show voids.)

1x-2NT: Baron, 15+HCP balanced

Limit jump shifts: 1D-2D/2H/2S; 1D-2H/2S/3C, 1H-2S/3C/3D, 1S-3C/3D/3H.
Hands valuable only in the suit bid; shows 6+card suit and about 9-11
points. Minor suits guaranteed to have 2 of top three honors (so as to
be runnable in NT opposite Qx,Kx, or Qx).

[akhare1]

"Ron Lel" <ron...@msn.com> wrote in message news:<yaAvc.3567$rz4....@news-server.bigpond.net.au>...

How interesting -- this is almost identical to the Tresboof FP system
that we play during lunch (and sometimes on BBO)!!! The only
difference is that P = 14+ w/ 4+ controls and the 1S and 2C bids are
transposed w/ 2D as Wilcosz...

Atul

[tys...@yahoo.com]

> 1NT = 15-17 balanced OR weak two in clubs
>

[snip]

> 2C = 20-22 balanced OR weak two in hearts
>

[snip]

> My question is: can your simulation tell that the 1NT definition is
> rubbish and the 2C definition is fine?

The short answer is yes. Did you wade through Ginsberg's explanation
of how this works? Basically it's a function of the final contract
you want to arrive at and how much bidding space you have to reach it.
For the 1NT example, you are very likely going to want to stop in 1N
or 2C and responder has no bidding space to explore. So responder has
to guess a lot and will often guess wrong. The 2C bid is much better.
You'll want to stop in some level of NT or hearts and there is still
bidding space to determine which one. So you'll be right more often
and score better.

[tys...@yahoo.com]

davjf...@aol.com (DavJFlower) wrote in message news:<20040603092634...@mb-m18.aol.com>...

Please everyone make sure to specify which bid you choose when
multiple meanings apply. What do you open with 6c&5H and so forth.
Also if you don't list meanings for bids (like weak 2's) I'm going to
assume "standard." Assumptions can be dangerous, so you might not
want to do it.

[tys...@yahoo.com]

> Here is a description of the Modified MOSCITO structure that I play in
> first second seat. I'm dumping the dealer script that I normally use
> for simulations. If you prefer I can provide a description in
> English.

You might want to give me an English version. I can read the dealer
script, but some others in the group might not be able to. Also, I
can't easily use CCCC to determine valuations if that's a problem.

Tysen

[Mike Bell]

In message <a1bce703.0406...@posting.google.com>
tys...@yahoo.com wrote:

...

> Let's have a contest to see who can design the best system (at least
> opening bids)!

...

> 3. Bids with multiple definitions will do okay if the meanings are
> very different so that further rounds of bidding can differentiate
> them. If the opponents preempt, will your partner be able to tell
> which definition you have?

Am I right in thinking that bids with multiple definitions are likely to
fare less well than they would at the table because the opponents
bidding over them will be modelled as 100% efficient?

Is TSP ok for hand evaluation?

Cheers
Mike

[Ron Lel]

snipped

> How interesting -- this is almost identical to the Tresboof FP system
> that we play during lunch (and sometimes on BBO)!!! The only
> difference is that P = 14+ w/ 4+ controls and the 1S and 2C bids are
> transposed w/ 2D as Wilcosz...
>
> Atul

Ah well, great minds....

Ron

[DavJFlower]

>> Here's mine - not a million miles different from precision:
>>
>> 1C 16+HCP any distribution
>> 1D & 2C: 11-15HCP 6-card suit, or 5 with another 4-card suit
>> 1H & 1S: 11-15HCP 5-card suit
>> 1NT: 13-15 balanced (May be 2254 or 2245)
>> 2D: 11-15 Roman
>> 2NT: 22-24HCP
>
>
>Please everyone make sure to specify which bid you choose when
>multiple meanings apply. What do you open with 6c&5H and so forth.
>Also if you don't list meanings for bids (like weak 2's) I'm going to
>assume "standard." Assumptions can be dangerous, so you might not
>want to do it.
>

OK Weak 2's in the majors

On two suiters, normally open the longer suit, although, as in many systems, a
view may be taken with extreme disparity of strength. With equal length, open
the higher ranking suit.

I tried to make it concise, but I overdid it. Sorry!

Dave Flower

[richard willey]

Hi Tysen

Bother. Would have been so easy if you had the K+R hand evaluation
function built in. Here is an anglicize descrition of the schedule:

Start by defining two hand evaluation functions.

SLAM POINTS: A = 3, K = 2, Q = 1
Singleton King or Singleton queen count as 0 slam points

POINT COUNT: A = 3, King = 2, Q = 1, J = .5, T = .25

####

Next define three strength levels for opening bids

STRONG HANDS have point count >= 9 and either
9+ slam points or (8+ slam points and 10+ cards in the two longest
suits)

CONSTRUCTIVE HANDS have pont count >= and either
6+ slam points or (5+ slams points and 10+ cards in the to longest
suits)

PREMPTIVE HANDS hve pint count >= 3.25

###

Here is the core of the system

3m = Constructive preempts that promise two of the top three honors
and deny a side suit Ace or Queen

2N = Bad 3 level preempt in either minor

2S = Preemptive bid
Promises one of two hand types Either

6+ Spades, single suited OR
4+ Spades and 5+ Clubs (denies 5440 shape)
With 4 Spades, Opener promises Hxxx or better

2H = Preemptive bid
Promises one of two hand types. Either

4+ Hearts and 4+ Spades (denies 4441 or 5440 shape) OR
4+ Hearts and 5+ Clubs (denies 5440 shape)
With a 4 card major, Opener promises Hxxx or better

2D = Preemptive bid
Promises one of two hand types. Either

4+ Dimaonds and 4+ Hearts or
4+ Diamonds and 4+ Spades

With a 4 card major, Opener promises Hxxx or better

2C = Constructive opening bid

6+ Clubs single suited or
6+ clubs and a bad 4 card major
(denies Hxxx or better in a major)

1N = Balanced hand (4432, 5332, 4443 shape)

11+ - 14 HCP
denies 5 card major
denies 4432 shape with 4 Hearts and 13-14 HCP
denies 4432 shape with 4 Spades and 11-12 HCP

1S = Constructive opening bid

Two suited hands with both Minors or
Sinlge suited hands with Diamonds or
6+ Diamonds and a 4 card major
denies Hxxx in a major

1H = Constructive Opening bid

4+ Spades, could have a longer minor
Open 1D with 4-4 in the majors
Open 1H with 5-5 in the majors

1D = 4+ Hearts, could have a longer minor
Open 1D wth 4-4 in the majors
Open 1H with 5-5 in the majors

1C = strong, artifical and forcing

[Jugoslav Dujic]

OK, here's mine (Panonia). Standard point-count, subject to
re-evaluation (read: "11" below can be excellent 10, strong 1D
also doesn't really require 18. We may pass awful 12 balanced
but open good 11 5332 or 4432).

Pass = 0-10
1C = 15-17 BAL (no 5cM) or
11-17 exactly 4cM + longer minor or any 4441
15-17 m 1-suiter or 2-suiter (same as 2C, 2D below but with 15-17)
responses: 1D = F1 (8+ or 0-7BAL), asking for clarification
anything else = 0-7 (mostly NAT)
1D = 18+ anything
responses: 1H = 0-6
1S = 10+
anything else = 6+ - 9 distribution relay response
1H, 1S = 11-17 5+c (even with longer minor)
responses: 1NT = 10+ HDP, relay
2x = 6-9 NAT NF
2NT+higher = various raises including Bergen
1NT = 12-14 BAL (no 5cM)
2C = 11-14, UNBAL, no 4cM, long clubs or 5+/5+ m (also exactly 5C-4D)
2D = 11-14, UNBAL, no 4cM, 5+ diamonds
2H, 2S = undisciplined weak-two, 5+c, may contain 5m.
2NT = garbage preempt in either minor if allowed; if not, 5+/5+m 6-7 losers
3C, 3D = "nice" preempts (if 2NT garbage allowed), otherwise, ranges from
"garbage" to "nice"
3H, 3S = preempt
3NT = gambling
4C, 4D = Namyats

--
Jugoslav
___________
www.geocities.com/jdujic

Please reply to the newsgroup.
You can find my real e-mail on my home page above.

[Jon Levell]

tys...@yahoo.com wrote in message

> The short answer is yes. Did you wade through Ginsberg's explanation
> of how this works? Basically it's a function of the final contract
> you want to arrive at and how much bidding space you have to reach it.

Which is why this thread should be called "tell us your favorite
forcing pass system" instead of claiming to evaluate system design.

I still think the ideal system for me is aggressive ACOL when the
opponents are not-vulnerable (12-14 NT) and agressive Precision when
the opponents are vulnerable (10-12 NT). I'm sure using up all that
bidding space on opening very weak notrumps would scoor poorly on your
evaluations, though.

Jon

[tys...@yahoo.com]

> Am I right in thinking that bids with multiple definitions are likely to
> fare less well than they would at the table because the opponents
> bidding over them will be modelled as 100% efficient?

It should fare as well as it would against humans, but humans who are
prepared against your system. My effeciency is set at 18%, much lower
than absolute perfect information transmission.

>
> Is TSP ok for hand evaluation?
>

Sure. =)

> Cheers
> Mike

[Kent Feiler]

It sounds like you're talking about a system to win bidding contests
rather than one that would win at the table with opponents and where
you have to play out the hands after the bidding is over. In bidding
contests the Ultimate Club was...pretty ultimate, but it didn't do
that well in actual play and wasn't really adopted by anyone.

The characteristics of a system that I think would win tournaments
are:

(1) It has to be very pre-emptive.
(2) It should be as "pre-empt-proof" as possible.
(3) It should concentrate on reaching the correct contract in as few
bids as possible.

Regards (1), you should have as many pre-emptive bids available as
possible. Firing the first shot in the auction is a big advantage.
Also, the kinds of pre-empts that work best are the either/or style
where the suits aren't initially known. These, of course, are the
ones many bridge organizations bar.

What this breaks down to is, don't waste your 2-level openers on
opening hands. Wheel out Multi-2C, Multi-2D, Ekrens, RCO, Muiderberg,
etc. And if there was no local bridge organization to fight, I'm sure
I could come up with a set of 2-bids that were wonderous strange and
allowed you to open any weak 1 or 2-suiter. Opponents of these systems
sometimes feel that they never get to open the bidding!

----------------------------------

Regards (2), I thought a system where the opening one-bids had two or
three completely different meanings might help. Something like the
convention used over opponents 1NT (forgot the name):

1C = Either:
1-suiter in diamonds,
major 2-suiter
12-14 balanced
etc, etc

...with the idea being that opponents pre-empt might help you figure
out which hand opener had.

------------------------------

Regards (3), one of the big problems with the Ultimate Club was that
it gave the opponents to much information -- they defended too well.
The best auction to reach 4H is a momma-poppa 1H-3H;4H, not something
with 11 asking bids and 11 answering bids that ended in the same spot.

The main reason I like Acol is that it concentrates on the majors and
doesn't bid minors unless they're a significant part of the hand --
unlike Standard American systems that are constantly bidding minors
that have almost nothing to do with the final contract.

A corrollary to (3) is that you want a system where most of the time
you can make the bid you want to make. If you want to raise partner's
suit, you should be able to do so immediatly, not go through some
temporizing bids. If you want to bid NT, there should be a NT bid
available that shows your point range. The temporizing auctions help
opponents defend

--------------------------

So...a system like this may not win many bidding contests, but you
wouldn't want to play against it at the table.

Regards,

Kent Feiler
www.KentFeiler.com

[Sandy E. Barnes]

<tys...@yahoo.com> wrote in message
news:a1bce703.0406...@posting.google.com...

> Every two months or so, someone on rgb asks "what is the best system?"
> The same flurry of answers usually follows starting with "whatever
> you can really remember and use effectively" and "it's the skill of
> the pair and not the system." Then the thread usually spins into
> theory and if is there an optimal system for two computers to use.
> Maybe there is...
>

> Let's have a contest to see who can design the best system (at least
> opening bids)!
>

> I'm not sure this can be answered, but I'm going to attempt to come
> close. If you make a simplified model of bridge bidding you can use
> information theory to quantitatively rate how good a system is. The
> trick is to make this model as close to "real" bridge as possible and
> not to make it too simplified of a model. Matt Ginsberg first
> proposed doing this back in 1996 and I think his theory is mostly
> sound. For details of how Matt's information theory system works,
> look here:
>
> http://tinyurl.com/27rql
>
> Be warned that this is not for the faint of heart or the
> non-mathematically inclined. For those who do not need the details,
> it talks about a way to take definitions for bids and quantitatively
> state how good of a definition it is. It takes into consideration
> several factors of bidding including:
>
> 1. how the bid helps us find the optimal contract
> 2. how the bid helps refine the definitions of *other* bids
> 3. how the bid hinders our opponents from finding their optimal
> contract
>
> It does have some limitations, and so it does not address:
>
> 1. right-siding the contract (all contracts are assumed to be
> right-sided)
> 2. lead directing inferences
>
> If you feel that these last two points are minor compared to what you
> get with the first three, then this contest might be right for you.
>
> So I propose the following system generation contest to be conducted
> here on rgb. There is no prize for winning except for the pride.
> After it's over, everyone here on the list can look at which systems
> were rated the highest and I think it will generate some healthy
> debate. We can discuss what points the program feels are the best and
> decide if they would really apply at the table. I think it can be a
> good learning experience for everyone. Plus I think it will be a lot
> of fun.
>
> Anyone who wants to participate in this contest should post their
> system on this thread (don't send me an email with it, only posted
> systems will be judged). Only post definitions for OPENING bids
> (anything more would be too much to handle in this format). Keep the
> following things in mind:
>
> 1. Any definitions are allowed; you can be as artificial as you want.
> Just be sure the definitions are accurate enough that I can program a
> computer with them.
> 2. Clearly specify all meanings for all bids (don't forget pass).
> Include ways to sort out which bid will be used when more than one
> definition can apply. For example if you have one bid that shows 5+
> spades and one that shows 5+ hearts, make sure I know which one to use
> if you are 55, 56, 65, etc.
> 3. You can specify meanings for 3C and higher, but *you don't have
> to*. Just list definitions from Pass to 2NT. Higher bids will be
> given to the "unassigned" hand definitions and the computer will
> intelligently assign definitions to them. Your system will preempt
> more aggressively if you have more unassigned hands. Just to make
> things clear, you can specify the hands that you will open 3C+ so I
> make sure to capture them, but they won't be inputted in specifically.
> These higher bids will really have no effect on your overall score
> since they will likely be rare. Also since these bids often take
> combinations of suit quality, exact shape, and other things it's hard
> to program them effectively.
> 4. Scoring is IMPs and vulnerability is none.
> 5. Only 1 system per person
> 6. All entries must be posted by June 20, 2004.
>
>
> Here's an example of how a Standard American system might look (and
> the level of detail I need):
>
> Count HCP and add 3/2/1 points for void/singleton/doubleton. Points
> include distribution unless specified otherwise.
> Pass = all 0-12 point hands with 0-5 in spades, hearts, and diamonds,
> and 0-6 clubs. All hands with 0-4 HCP and a 7-card suit.
> 1C = 13-22 points, 3+ C
> 1D = 13-22 points, 3+ D
> 1H = 13-22 points, 5+ H
> 1S = 13-22 points, 5+ S
> 1NT = 15-17 HCP any 5332, 4432, or 4333 shape
> 2C = 23+ points, any shape
> 2D = 7-12 points, 6 D, no 4cM
> 2H = 7-12 points, 6 H, no 4cS
> 2S = 7-12 points, 6 S, no 4cH
> 2NT = 20-21 HCP any 5332, 4432, or 4333 shape
> *For hands that have more than one definition: 1NT & 2NT take highest
> priority. Otherwise longest suit in the hand takes priority. If
> there is a tie for longest suit, then higher ranking suit takes
> priority except in the case of 33 in the minors, then bid 1C.
> *Note that the unassigned meanings for weak hands with 7+ suits will
> be intelligently assigned to bids of 3C+.
>
>
> I'm sure everyone here can be much more creative than this when you
> are not bounded by memory or system restrictions. Let's get creative
> here!
>
> I've played around with several definitions, so let me give you some
> hints as to what the computer will rate highly:
> 1. Shape is more important than strength. A very specific shape bid
> with wide strength (within reason) will do better than a nebulous bid
> with narrow strength.
> 2. Open light, especially unbalanced hands

> 3. Bids with multiple definitions will do okay if the meanings are
> very different so that further rounds of bidding can differentiate
> them. If the opponents preempt, will your partner be able to tell
> which definition you have?

> 4. The opponents will always have the best defense prepared, so you
> can't rely on unfamiliarity with the system!
>
> I think that's it. Email me only if you have private questions,
> otherwise let's try to keep everything posted publicly so that
> everyone can learn.

My effort:
OPENING BIDS:
1C/1D: 9+ HCP's, less than 8 cashing tricks in hand with 3 defensive tricks;
weaker hands have 6 card suits with 2 of top 3 honors, 1-1/2 QT
1H/1S: 9+ HCP's, 5 + card suit, less than 8 cashing tricks in hand with 3
defensive
tricks; weaker hands have 6 card suits with 2 of top 3 honors, 1-1/2
QT
1NT: 11-14 when opponents RED, 15-17 when opponents WHITE
2C: Game force, artificial (2D Negative)
2D/2H/2S: 1 round force, 8+ playing tricks, 3+ defensive tricks
2NT: 21-23 HCP's
3NT: 25-27 HCP's, no 4 card major
4C/4D: Namyats (transfer to hearts or spades), 8-9 tricks in hand
4H/4S: Natural, less than a 4C/4D call
4NT: Blackwood

However, one needs to look beyond the opening bids to judge system design.
I would maintain that the response structure is almost of greater
importance, and that structure may be way to great to examine in this forum.

Sandy Barnes

[Sandy E. Barnes]

"Sandy E. Barnes" <sandyb...@cox.net> wrote in message news:e84wc.47839

I left out the 3 level:
3C/3D = solid suit and out, 7 card length
3H/3S = Normal preempt

I would also add that the response structure is directed at game
investigation, and uses Fit-Jumps and Inverted Raises to quickly identify
fit and level in all auctions. Many negative infrences from Opening bids
and first response.

[tys...@yahoo.com]

jle...@hotmail.com (Jon Levell) wrote in message news:<c23f7f29.04060...@posting.google.com>...

The opponents are the other half of the equation though. It's not
just about constructive auctions. I'm predicting it's not going to be
a runaway victory for a focing pass system. I haven't run all the
bids yet, but I have experimented enough to know what works and what
doesn't.

Same for the weak NT. Sure you take up bidding space for you, but you
take it away from your opps as well.

[tys...@yahoo.com]

Kent Feiler <zz...@zzzz.com> wrote in message news:<ojh1c0liqb9t6loe2...@4ax.com>...

> It sounds like you're talking about a system to win bidding contests
> rather than one that would win at the table with opponents and where
> you have to play out the hands after the bidding is over.

It's not as bad as all that. There *are* opponents and we're trying
to preempt them as well as trying to get constructive auctions.
Resisting preemption is also a key factor in scoring well in this
contest. Many bids will simply not do well if they can't resist
preemption.

Helping the defense because of our descriptions is unfortunately
unavoidable. That is a deviation away from real-world bridge.
There's no way to get real play data for these systems in action, and
everything has to be double-dummy. So it's not perfect, but it's the
best we've got. The extra help we give the defense *is* balanced by
the fact that we're reaching better contracts in the first place.
Who's to say which is a bigger factor?

[Mike Bell]

In message <a1bce703.0406...@posting.google.com>
tys...@yahoo.com wrote:

You could take into account the information given away in the bidding by
creating a number of layouts that are consistent with the auction and
opening leader's hand, see what lead works most often double dummy, and
then force that lead on the actual hand before playing it out double
dummy. This would obviously make the simulation much slower, and still
wouldn't give uninformative bidding its full benefit, as it would only
affect the lead. This could be extended to run throughout the play, but
it gets very slow, very quickly!

Mike

[Mike Bell]

In message <4c0ub0pco1dvpk6t6...@4ax.com>
Michael Clark <clar...@NOSPAMbtopenworld.com> wrote:

> On 2 Jun 2004 15:13:28 -0700, tys...@yahoo.com wrote:
>
> >3. Bids with multiple definitions will do okay if the meanings are
> >very different so that further rounds of bidding can differentiate
> >them. If the opponents preempt, will your partner be able to tell
> >which definition you have?
>
> We don't have to specify the responses or continuations to these
> opening bids, but does there have to be some logic behind it? For
> instance, could I specify:
>
> 1NT = 15-17 balanced OR weak two in clubs
>
> These two meanings are nicely differentiated and if partner knows
> which one you have you'll be in a happy position. However, it's
> obviously impossible to have a sensible set of responses to this.

Actually the more I think about it, the more I think this is genius...

1NT = 15-17 balanced or 3-8, 6+clubs.

Responses:

With < game invite opposite 15-17, bid 2D/H/S/3C weak takeout, or pass.
If opener has the weak 2 the opps should have game on, so it is ok to
play 1N or a 5-1 fit down a few.

With GI+ values opposite 15-17 bal, but not opposite a weak 2,
responder bids 2C, non-forcing stayman :)

With GI+ values opposite a weak 2, 2N feature enquiry or similar.

Mike

[Chris Ryall]

Mike Bell wrote on "Bidding System Design Contest"

>> 1NT = 15-17 balanced OR weak two in clubs
>>
>> These two meanings are nicely differentiated and if partner knows
>> which one you have you'll be in a happy position. However, it's
>> obviously impossible to have a sensible set of responses to this.

Seems fairly trivial to me. Most such combo bids need a forcing and non
forcing enquiry with 2C fairly obvious for the latter. any system above
this would lose precision due to the need to incorporate a weakness
showing call on nearly every branch.

Would it be efficient? Well nv passing the 1NT "happy either way" might
score well, but a vulnerable 1NT on trashy clubs ain't so good for
business.

Would anyone allow it? WBF look to be quite content. EBU certainly
aren't in anything short of Spring Foursomes (HUMs/brown allowed). But
then we are designing outside such constraints aren't we.

--
Chris Ryall Wirral-UK
("cut out" spamtrap to email me)

[Chris Ryall]

"Betty" in an attempted flow diagram order,
simplified to try to help,
and tidied to be similar to others

2C: GF or 16+ 4441 or 5440, or 8-9 tricks in 5+major
or 10 tricks in 6+ minor
2NT: 20-22 balanced, singleton OK (4441 see above)
2D: UK-Multi weak 2 in 6+ major, or 9 tricks in minor
or 23-24 balanced, singleton OK (4441 see above)

3/4 bids: standard preempts, no 4cM side suit
tend to pass 7222 minors or open 2D 7222 majors

1NT: nv) 9-12 balanced 5422, 6cm or 5cM OK (good run outs)
vul) 14-16 with 5422, 6cm or 5cM OK
All suitable hands in above ranges tend to be opened 1NT

2H: 5 - 10 HCP, 4+H 4+S (or longer) majors
2S: 5 - 10 HCP, 5+S + 4+minor

1S: 11-19 HCP 5+ suit (usually open 1S with 5-5 and this suit)
1H: 11-19 HCP 4+ suit open longer major, or 1H with 4-4 majors
1D: 11-19 HCP 4+ suit
1C: 11-19 HCP 4+, or 3+ suit if 4333 with 4 spades

longer suit first, then priority to bidding majors if we have one

pass: in extremis :))

Dynamics ARE important. eg 3 of suit opposite 1NT is a preempt and with
< 11 opposite opening we make a fit jump. Like Kent, we aspire to hit
games in only 3-4 bids when there is shape about using LOTT.

[Kent Feiler]

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Well...that's my problem with this since I think the answer to "which
is the bigger factor" is completely obvious. The two things you're
comparing are: (1) how often does a better bidding system reach a
better contract, and (2) how often can the opponents take advantage of
having more information when they're defending? What's the answer to
(1)? Hard to say, but I'd guess that the best system in the world
against an ordinary pretty good system still wouldn't pick up more
than one hand in twenty. What's the answer to (2)? Almost every hand!
In the end, a system that doesn't allow opponents to mis-guess or
mis-defend isn't going to be successful.

Also, one way to produce a system that has better constructive bidding
is to give up pre-emptive bids and replace them with constructive
ones, but that's actually another way of giving the opponents
information. Now they get extra information by having a free rein in
their own constructive auctions. In addition to your reaching better
contracts with your constructive bidding, they're reaching better ones
with their's because you're not interfering with them!

Regards,

Kent Feiler
www.KentFeiler.com

[Michael Clark]

On Sat, 5 Jun 2004 10:03:32 +0100, Chris Ryall <groups2@[127.0.0.1]>
wrote:

Ok, I stand corrected! I'm sure there must be better examples though
of the point I was trying unsuccessfully to make. A convention which
might appear at first glance to work ok but where formulating proper
responses turns out to be impossible. It seems that the program used
in this contest can properly assess this, though, which I'm quite
impressed with.

My example isn't such a failure, really, because you're going to lose
a fair bit of accuracy in the bidding when you hold the 15-17 type -
try determining that your 8 card major fit will play better in 3NT
now!

So Chris, I know this convention isn't strictly a weak 'two' but
perhaps it can gain honorary status and a footnote in your great
archive? :D

Michael

[Chris Ryall]

Michael Clark wrote on "Bidding System Design Contest"

>>>> 1NT = 15-17 balanced OR weak two in clubs

>So Chris, I know this convention isn't strictly a weak 'two' but

>perhaps it can gain honorary status and a footnote in your great
>archive? :D

I'd be happy to include it as a nice combo version of weak 2C, but
there are (honestly) some ground rules for the Archive ..

I try to only include things that have been played somewhere. eg
Velociraptor 2S=4S & 5+minor was spotted by Girben playing against
the Norwegians (and interestingly is now in his pet system here )

If anyone sees 1NT = 1NT | {weak clubs} anywhere *at the table* please
let me know. Otherwise congratulations to Michael on a shrewd idea.
I can't see EBU letting him use it, as with a forced penalty double
against both the weak two OR strong 1NT it would be a nuisance :))

btw - saw yet another 1NT defence playing Hacket/Mold last week
and will send into DS's archive of those.

[Larry Lowell]

> <tys...@yahoo.com> wrote in message
[snip]

> > Every two months or so, someone on rgb asks "what is the best system?"
> > The same flurry of answers usually follows starting with "whatever
> > you can really remember and use effectively" and "it's the skill of
> > the pair and not the system." Then the thread usually spins into
> > theory and if is there an optimal system for two computers to use.
> > Maybe there is...
> >
> > Let's have a contest to see who can design the best system (at least
> > opening bids)!
> >

[snip]

Here is my favorite that I play with one partner: Canape 'Precision'

1C = 16+ hcp, any distribution
1D = 11-15 hcp, 4+D, may have a 5-card major or 5+C, with good 5M always open 1D
1H = 11-15 hcp, 4+H (Qxxx+)may have a longer 5+ card minor, 44xx/45xx open 1H
1S = 11-15 hcp, 4+S, may have a longer 5+ card minor, or 4+H
1NT = 11-13 nV, 14-16 V with transfers and DONT escapes, no 4M (Qxxx+)
2C = 11-15 hcp, 5+C, no second suit of Qxxx
2D = 11-15 hcp, 6+C, no 4M, maybe 4C (3 controls minimum for 11 hcp)
2H = 11-14 hcp, 5H332 distribution, lebensohl responses
2S = 11-14 hcp, 5S332 distribution, lebensohl responses
2NT = 11-14 hcp, unusual for the minors, 55
3C = 11-14 hcp, 6-7 playing tricks with semi-solid+ clubs

Larry Lowell
Knoxville, TN, USA

[Larry Lowell]

Kent Feiler <zz...@zzzz.com> wrote in message [snip]

>
> It sounds like you're talking about a system to win bidding contests
> rather than one that would win at the table with opponents and where
> you have to play out the hands after the bidding is over. In bidding
> contests the Ultimate Club was...pretty ultimate, but it didn't do
> that well in actual play and wasn't really adopted by anyone.
>

[snip]

>
> Regards (3), one of the big problems with the Ultimate Club was that
> it gave the opponents to much information -- they defended too well.
> The best auction to reach 4H is a momma-poppa 1H-3H;4H, not something
> with 11 asking bids and 11 answering bids that ended in the same spot.
>

[snip]
>
> Kent Feiler
> www.KentFeiler.com

Well, I disagree about too much information. I played a simplified
version of Ultimate combining with sweep q-bids and distributional
asks somewhat easier than Ultimate, and we did not have long auctions
unless slam potential was identified with the control ask relay.
Usually, we just closed up shop and bid game. 'Low Information'
without implying the weak spot control wise in the hand. I have the
Ultimate Club book and have studied it and Symmetric Relay.

[Chris Ryall]

Larry Lowell wrote on "Bidding System Design Contest"

>
>Here is my favorite that I play with one partner: Canape 'Precision'

A strongly polarised selection that more or less abandons pre-emption
below the 3D level in favour of well defined semi-constructive bids,
while it's 1C on stronger hands is maximally vulnerable to counter
measures.

Provided the test engine is successful in following these auctions
through this will be an interesting horse to watch against say 3 ranks
of weak twos, and something slightly evil over the strong club.

However it should score well on hands where it can open a neat 11-14

[John Crinnion]

tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...

> Every two months or so, someone on rgb asks "what is the best system?"
> The same flurry of answers usually follows starting with "whatever
> you can really remember and use effectively" and "it's the skill of
> the pair and not the system." Then the thread usually spins into
> theory and if is there an optimal system for two computers to use.
> Maybe there is...
>
> Let's have a contest to see who can design the best system (at least
> opening bids)!

Might we have two categories: natural and artificial?

<massive SNIP>

[tys...@yahoo.com]

> Well...that's my problem with this since I think the answer to "which
> is the bigger factor" is completely obvious. The two things you're
> comparing are: (1) how often does a better bidding system reach a
> better contract, and (2) how often can the opponents take advantage of
> having more information when they're defending? What's the answer to
> (1)? Hard to say, but I'd guess that the best system in the world
> against an ordinary pretty good system still wouldn't pick up more
> than one hand in twenty. What's the answer to (2)? Almost every hand!
> In the end, a system that doesn't allow opponents to mis-guess or
> mis-defend isn't going to be successful.
>
> Also, one way to produce a system that has better constructive bidding
> is to give up pre-emptive bids and replace them with constructive
> ones, but that's actually another way of giving the opponents
> information. Now they get extra information by having a free rein in
> their own constructive auctions. In addition to your reaching better
> contracts with your constructive bidding, they're reaching better ones
> with their's because you're not interfering with them!

I think someone already covered this, but the resulting system will
not simply reveal everything to the opponents. Most of the time
you'll have game or less and there simply won't be enough room to
exchange meaningful information. Each hand will not be mapped out.
The model always assumes that our continuations *won't* be perfect and
that the opponents could come in at any time. You won't get a system
that bids very slowly and perfectly. The opponents will come in and
you won't be able to fully explain yourself. Aiding the opponents
constructive auction is also something the simulation will pick up and
it will affect your score. If you reveal spade shortness at a low
level, the opponents will have that information and make more
intelligent decisions.

The system that wins will have the greatest balance of preemption,
construction, and the ability to handle ambiguity.

[tys...@yahoo.com]

Please everyone remember to define what your Pass bid means.
Sometimes I can't tell from your descriptions which hands will pass
vs. preempt. I've privately emailed many of you with this, but
haven't received responses from everyone. For those who haven't
submitted an entry yet, please remember to be specific and don't
assume anything.

Also, with respect to entires like "good suit" or "good 11": I'm going
to ignore this unless you can give me some fixed rules on how to make
this judgment. "Two of top 3 honors" "Qxxx or better" "At least 3
controls" would all be adequate descriptions.

Thanks

[MartinRJCarpenter]

In message <p1er3DEgSuwAFwco@[127.0.0.1]>

Chris Ryall <groups2@[127.0.0.1]> wrote:

> Michael Clark wrote on "Bidding System Design Contest"
>
> >>>> 1NT = 15-17 balanced OR weak two in clubs
>
> >So Chris, I know this convention isn't strictly a weak 'two' but
> >perhaps it can gain honorary status and a footnote in your great
> >archive? :D
>

> If anyone sees 1NT = 1NT | {weak clubs} anywhere *at the table* please
> let me know. Otherwise congratulations to Michael on a shrewd idea.
> I can't see EBU letting him use it, as with a forced penalty double
> against both the weak two OR strong 1NT it would be a nuisance :))

It depends: as long as you restrict the hand shapes on the weak 2 in clubs
to the 'semi - balanced' types defined in the orange book then you might well
be OK at level 3.
(iirc These include 6322,7222,5422 etc).
You could even throw weak 2's in diamonds in too. No conventional responses
to this opening allowed of course!

Only really of any potential 3rd in at green etc I fancy.... And a little
dodgy due to the heavy treatment of the weaker part of the 1NT range.
--
****__. Martin Carpenter .__****
*****@/ \@*****
*****-- --*****
***\ /***

[Nancy Lowell]

lnlo...@flash.net (Larry Lowell) wrote in message [snip]

>
> Here is my favorite that I play with one partner: Canape 'Precision'
>
> 1C = 16+ hcp, any distribution
> 1D = 11-15 hcp, 4+D, may have a 5-card major or 5+C, with good 5M always open 1D
> 1H = 11-15 hcp, 4+H (Qxxx+)may have a longer 5+ card minor, 44xx/45xx open 1H
> 1S = 11-15 hcp, 4+S, may have a longer 5+ card minor, or 4+H
> 1NT = 11-13 nV, 14-16 V with transfers and DONT escapes, no 4M (Qxxx+)
> 2C = 11-15 hcp, 5+C, no second suit of Qxxx
> 2D = 11-15 hcp, 6+C, no 4M, maybe 4C (3 controls minimum for 11 hcp)
> 2H = 11-14 hcp, 5H332 distribution, lebensohl responses
> 2S = 11-14 hcp, 5S332 distribution, lebensohl responses
> 2NT = 11-14 hcp, unusual for the minors, 55
> 3C = 11-14 hcp, 6-7 playing tricks with semi-solid+ clubs
>
> Larry Lowell
> Knoxville, TN, USA

Not with me you don't! Sweetheart, I prefer to make the 1 club
opening less vulnerable to interference:

1C = 15+ hcp,(a)Any balanced hand 15+ with 3 clubs, (b) A one suiter
with 6+ clubs, (c) a two suiter with 5+ clubs and 4-cards maximum of
2nd suit, (d) All strong hands of 4 losers or less. Responses are
one-under transfers 0-8 hcp, 2 clubs = G.F. any distribution. 2D & 2H
= 1-under transfers & 6-7 hcp & 6-cards.
1D = 11-19 hcp, 4+D, (a) One suited with diamonds, (b) 2-suited with
weak 4M,(c) 3-suited: 4441, 4144, 1444 with 4 diamonds
1H = 11-19 hcp, 4+H (Qxxx+) may have weak 5 diamonds, 44xx/45+xx open
1H
1S = 11-19 hcp, 4+S (Qxxx+) may have weak 5 diamonds, 54xx/5+5xx open
1S
1NT = 11-14 nV, With transfers and DONT escapes, no 4M (Qxxx+)
2C = 11-14 hcp, 5+C, no second suit of Qxxx
2D = 11-14 hcp, 6+D, no 4M, maybe 4C (3 controls minimum for 11 hcp)
2H = 11-14 hcp, 5H & 4C
2S = 11-14 hcp, 5S & 4C

2NT = 11-14 hcp, unusual for the minors, 55
3C = 11-14 hcp, 6-7 playing tricks with semi-solid+ clubs

3D+ = 3-9 hcp, 6-7 playing tricks (nV, V)
3NT = Gambling, 7-8 tricks
4X = 3-9 hcp, 7-8 playing tricks (nV, V)

Nancy

P.S. I think 'my' system will beat your system in this contest!!

[Gerben Dirksen]

jcri...@yahoo.com (John Crinnion) wrote in message news:<2f389bab.0406...@posting.google.com>...

Then we get to fight about what is natural. Cool! Polish Club is as
natural as Std. American or maybe even more so. Both have one forcing
opening bid, and in both cases it is in clubs. In my posted system the
forcing "opening" is pass. Maybe that is even more natural since now
even 1C is natural! I don't know...

Gerben

[tys...@yahoo.com]

> > Might we have two categories: natural and artificial?
>
> Then we get to fight about what is natural. Cool! Polish Club is as
> natural as Std. American or maybe even more so. Both have one forcing
> opening bid, and in both cases it is in clubs. In my posted system the
> forcing "opening" is pass. Maybe that is even more natural since now
> even 1C is natural! I don't know...

Yeah I was thinking the same thing. Who is to decide what is natural?
I'll list all the systems and their scores. If you submit a natural
system, wind up ranking 3rd but the top two are highly artificial,
then call yourself a winner. I'm not being sarcastic. There is
something to be said for having a good but not perfect system, but a
system you can actually remember at the table. But I think there is
no need to have two separate contests.

[tys...@yahoo.com]

We have a few unique systems here, but I was hoping to see more people
submit something "strange." Most people are submitting the stuff they
like to use now. That's fine but this is a great opportunity to see
what works and what doesn't. I'm sure a lot of you out there have
something that you thought was cool in your head, but you've never had
the time to work out the full responses, or maybe you haven't found a
willing partner to practice. This is the perfect forum to test it
out! Go super crazy, or just a little. Maybe something like:

1C = any balanced
1D = any 3-suiter or any 6+ card single suiter
(these are easy to differentiate)
1H = 2-suiter (54+), same color
1S = 2-suiter (54+), same rank
1N = 2-suiter (54+), same shape
2C = strong
2D = multi
2H+ = whatever

Will this work? I don't know. Now that I'm thinking about it, maybe
this is better:

1C = any balanced
1D = any 3-suiter or any 6+ card single suiter
1H = 2-suiter (54+), always has hearts
1S = 2-suiter (54+), always has spades
1N = strong
2C = both minors (54+)

Neither of these is super crazy. I'm just brainstorming here. Just
writing this gets the thoughts flowing. Who can come up with
something creative that they've never tried before?

[b...@moscito.org]

A system that I wrote as a lark and that someone actually picked up to play.

Pass: 14+, semi-balanced (4432, 4333, 5332, 5422), or unbalanced with a
single (6+) suit or 10+ cards in two suits.
-> 1C semi-positive 6-9 hcp
-> 1D negative 0-5 hcp
-> 1M+ like symmetric relay, opener breaks to show the unbalanced hands.

Over intervention, use transfer lebensohl solutions assuming opener has
a 14-to-17- no-trump, opener will bid freely with 18+ and unBAL.

1C: 8+ hcp up, 5431, 4441, 5440, 6430.
-> 1D relays, everything else is pass or correct.

1D: 0-8 hcp (if 8 then balanced)
-> 2C is the only force, shows a strong two-bid.

1H: 9-13 hcp, 4-5 H, semiBAL or 5H+5m
1S: 9-13 hcp, 4-5 S, semiBAL or 5S+5x
1N: 9-13 hcp, no 4cM, semiBAL
-> 2C is an artificial relay
-> 2D over 1S is a transfer to H
-> 2M-1 is an artificial strong raise.
-> 2N is a signoff in a minor or strong.

2x: 8+-13 hcp, 6+x
-> next step relays

2N: 8+-13 hcp, 5-5 in the two suits.
-> 3H relays

3x: preempts, wild

The version that was played in tournaments looks like this:

1C: 16+, not semi-3-suited
1D: 10+, semi-3-suited.
1M: 11-15 semiBAL with 4-5M or 2-suited.
1N: 11-15 semiBAL without 4-5M
2x: 10-+15 6+ cards
2N: 10-15 minors

I might have my notes stuck away somewhere ...

[Kent Feiler]

On 8 Jun 2004 09:26:59 -0700, tys...@yahoo.com wrote:

>We have a few unique systems here, but I was hoping to see more people
>submit something "strange." Most people are submitting the stuff they
>like to use now. That's fine but this is a great opportunity to see
>what works and what doesn't. I'm sure a lot of you out there have
>something that you thought was cool in your head, but you've never had
>the time to work out the full responses, or maybe you haven't found a
>willing partner to practice. This is the perfect forum to test it
>out!

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Ok, you got me. This system is now called:

Standard Suction

1C = Either:
(1) Diamond 1-suiter,
(2) Hearts and a minor 2-suiter,
(3) 3-suiter short in spades,
(4) 13-15 flat.

1D = Either:
(1) Heart 1-suiter,
(2) Spades and a minor 2-suiter,
(3) 3-suiter short in clubs,
(4) 16-18 flat.

1H = Either:
(1) Spade 1-suiter,
(2) Club-Diamond 2-suiter,
(3) 3-suiter short on diamonds,
(4) 19-21 flat.

1S = Either:
(1) Club 1-suiter,
(2) Diamond-Heart 2 suiter.
(3) 3-suiter short in hearts
(4) 22-24 flat.

1NT = 10-12 flat

2C = Multi. Weak two in any suit.

2D = Three-suiter. 6-10.

2H = Two suiter. 6-10. Hearts and another.

2S = Two suiter. 6-10. Spades and another.

2NT = Minors. 6-10.

3X = Pre-empt

With distributional hands, this system should be able to open the
bidding with any 6+ HCP hand. With flat hands, any 10+ HCP hand.

All one-of-a-suit bids are forcing. With a bad hand, reponder bids
the next suit up. If the opponents don't interfere, you should be able
to easily sort out the four possibilities for each opening bid. If
they do interfere, their interference may help responder figure out
which hand his partner has.

There's no strong opening bid. With a strong hand, opener makes the
normal opening bid for his distribution and then jumps at his next
turn to force one round.

...or something like that. Might need a little work, but I'm willing
to play it in the local Regional Open Pairs as is.

Regards,

Kent Feiler
www.KentFeiler.com

[Bert Beentjes]

tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...

> Let's have a contest to see who can design the best system (at least
> opening bids)!

What I'm currently working with, called B Sharp:

1C - 13+ other or 12+ any XS/XT
1D - 13+ 5+H, no longer second suit
1H - 13+ 5+S, no longer second suit
1S - a) 8-12 SV/SIX 4+C
b) 10-12 SB 5+C or any 4333 or any 4432 (no 4D333 or 4D4M32)
1NT- a) 8-12 SV/SIX 4+D
b) 10-12 SB 5+D or 4D333 or 4D4M32
2C - a) 8-12 SV 5H
b) 10-12 SB 5+H
2D - a) 8-12 SV 5S
b) 10-12 SB 5+S
2H - 8-12 SIX H
2S - 8-12 SIX S
2NT- 8-11 XT S
3C - 8-11 XS C
3D - 8-11 XS D
3H - 8-11 XS H
3S - 8-11 XS S
3NT- 8-11 XT H
4C - 8-11 XT C
4D - 8-11 XT D

SB = 4333 4432 5332 5422 6322 7222
SV = 4441 5440 5431 5521 5530
SIX= 6331 6421 6430 7321 7330
XS = 8+ or 74
XT = 6+/5+

for SV opening bids: 55 preference for the Major suit, 55 in the
Majors start with hearts, 55 in the minors start with clubs
for XT opening bids: longest suit or lowest bid

---------------
Bert Beentjes
Nijkerk (Gld.)
The Netherlands

[Bruce...@nospam.akamail.com]

tys...@yahoo.com wrote:
: Every two months or so, someone on rgb asks "what is the best system?"

: The same flurry of answers usually follows starting with "whatever
: you can really remember and use effectively" and "it's the skill of
: the pair and not the system." Then the thread usually spins into
: theory and if is there an optimal system for two computers to use.
: Maybe there is...

: Let's have a contest to see who can design the best system (at least
: opening bids)!

Greetings.

One hopes that the entries need not be original. I was waiting
for someone to enter Regres, a standard FP system. Since noone
else has, here goes:

Openings (w/ exceptions, all openings are 8-12)
===============================================
P 13+, any.
1D 0-7, any.
1C Most hands w/ singleton or void.
1H 3-4H, <=4S, H>=S. any 4=4=x=x, 3=3=x=x better H.
1S 3-4S, <=3H, S>H. 3=3=5=2 w/ poor diamonds, better S.
1N 5-7 cards in a major.
2C 5+C, if 5332, then good clubs.
2D 5+D, if 5332, then good diamonds.
2H 6+C/4+H, 5C/5H, 6+D/4+S, OR 5D/5S
2S 6+C/4+S, 5C/5S, 6+D/4+H, OR 5D/5H
2N 6+D/4+C, 5D/5C, 6+S/4+H, OR 5S/5H
3C 6+C/4+D, 6+H/4+S

Tie breaking: The alternate two suiters are always used when
appropriate. Any opening between 1H and 2D inclusive denies a
singleton or void. I'll interpret good clubs/diamonds to mean
that HCP(minor) > HCP(3 card major). All openings except pass
and 1D show 8-12 HCP.

I'm fairly sure that a seasoned FP practitioner wouldn't play it
exactly as described above, but it should be reasonably
unambiguous.

I look forward to the results and suspect that Regres will be
a contender.

Bruce

[Carl Ritner]

tys...@yahoo.com wrote in message news:<a1bce703.04060...@posting.google.com>...

> We have a few unique systems here, but I was hoping to see more people
> submit something "strange." Most people are submitting the stuff they
> like to use now. That's fine but this is a great opportunity to see
> what works and what doesn't. I'm sure a lot of you out there have
> something that you thought was cool in your head, but you've never had
> the time to work out the full responses, or maybe you haven't found a
> willing partner to practice. This is the perfect forum to test it
> out! Go super crazy, or just a little.

Here's a concept that's more than a little offbeat, and anyone is free
to modify their system to accommodate it and then submit it as an
alternative. I thought about this while reading Matt's basic approach.

1C = Gerber. Everything else can follow the basic theory of your pet
system.

Both casual and serious players realise the importance of checking for
aces before committing to slam, but this is also a critical issue for
games, since no matter what the combined strength, a partnership
missing the majority of aces will very rarely make game. Determining
the number of aces between the two hands early on sets the pace for
the rest of the bidding, and can frequently be enough information for
the opener to proceed, while keeping his own strength from the opps.

Since this particular form of contest only considers opening bids,
this concept might fare well enough to tweak an otherwise promising
system to victory. Or perhaps not. But it certainly does fulfill the
request for strange...

Cheers,
Carl Ritner
www.carlritner.com <== home of the ACBL library sale!

[Chris Ryall]

Carl Ritner wrote on "Bidding System Design Contest"

>
>1C = Gerber. Everything else can follow the basic theory of your pet
>system.

Nice idea. In essence played by the Italian Blue team in the 60's :))

[Bill Reich]

Chris Ryall <groups2@[127.0.0.1]> wrote:

By one pair, the ones playing Neapolitan. Roman is nothing like the "Blue
Team Club" that was sold to the bridge public. Adding a few Roman conventions
to Neapolitan did not blend them into one system.

I was always fascinated by the articles and conversations selling Big Club
methods on the basis of their success. Only one pair played a Forcing Club
system.
Will in New Haven

--

This hand will raise now.
There is no I to do it;
The cards themselves act.

----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! >100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---

[John Crinnion]

tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...

<SNIP>

> Let's have a contest to see who can design the best system (at least
> opening bids)!

<MASSIVE SNIP>

When are the judges going to announce the result?

[tys...@yahoo.com]

> When are the judges going to announce the result?

I opened the contest through June 20th so new ones can come in until
then. I've been trying to analyze the systems in the mean time since
it's not that easy of a task. Depending on how much time I'm able to
devote to it and how many more new systems come in, it could be a week
or so later that I'll post the results.

Tysen

[Sandy E. Barnes]

<tys...@yahoo.com> wrote in message
news:a1bce703.04061...@posting.google.com...

What is the judging criteria?

It seems to me this is difficult to describe since you can only evaluate the
opening bid action, and that limits you to offensive descriptive actions,
and defensive blocking actions. To really evaluate any system, you need to
understand their complete system structure both un-molested, and after
interference.

As an example, after Pass (13+ forcing), 3H by opponent, how does the
original pass in a Forcing Pass System compare to an opening bid of (say) 1C
in a "natural" system, followed by 3H. Notice I am not selecting 1S which
would favor a 5 card major opening approach, but this is also a
consideration to the complete system structure being compared.

In addition, every system features sequences which allow it to handle some
hands better than other systems may. How is this going to factor in. How
well does the system "under the microscope" handle somewhat normal
featureless hands compared to other systems?

Sandy Barnes

[tys...@yahoo.com]

> What is the judging criteria?
>

The full answer to this is quite complicated. Refer to the link
mentioned in the original post.

> It seems to me this is difficult to describe since you can only evaluate the
> opening bid action, and that limits you to offensive descriptive actions,
> and defensive blocking actions. To really evaluate any system, you need to
> understand their complete system structure both un-molested, and after
> interference.

Part of the premise of doing system evaluation this way is that a
complete analysis is impossible because it depends on the opponents'
methods. You simply can't evaluate the entire system. The opponents'
bids would depend on your meanings, and then your follow on responses
depend on wheir methods, etc. Again refer to the link in the original
post. This method of evaluation is the only one that I've heard of
which can approximate handling this problem. It's so flexible since
you don't even need to know the opps' system or even your own rebids!
It uses information theory and simply says that our rebids are unknown
at this point but are assumed to have a certain effeciency from
perfection (in my case 18%). Our opponents are also assumed to be
using methods which are 18% effecient. So without knowing what the
specific rebids will be, we get an approximation of the result. This
handles rebids with and without interference. If you are better able
to define your hand with your opening bid, your 18% effecient rebids
will have a much easier time coming to the right contract than if your
opening bid is more general.

[Steve Willner]

I'd like to propose two systems based on the same evaluation
criteria. (As a general comment, it will be hard to tell how much a
ranking comes from system design and how much from the evaluation
criteria chosen.)

For both systems, first decide whether the hand is suit-type or
notrump-type as follows:
1. All 4333 and 4432 are notrump-type.
2. 5332 is notrump-type unless the suit is a major headed by Q or better.
3. For 5422 and 6322, apply these rules in order:
suit-type if the long suit is a major;
notrump-type if at least one doubleton has an A or K;
notrump-type if both doubletons have a Q;
otherwise suit-type.
4. Other distributions are suit-type.

Then evaluate:
For notrump-type, AKQJT are 4.1, 3.0, 2.8, 1.8, 0.3 respectively.
(Feel free to change the numbers if you have better values.)

For suit-type, it would be nice if you could use C4. Failing that,
use BUMRAP+321. Ideally there would be adjustment for whether the
honors are in long suits or short ones, but that's probably not
necessary.

Define "a preempt" as: any hand with an 8-card or longer suit, any
with a 7-carder except 7222, also 7222 if no A or K outside the long
suit. These hands will open 3C or above if they don't qualify for
one of the defined openers. (Again you can refine the preempt
definition, but it probably won't matter.)

System 1 is K-S with slight adjustments. (Hope I've got the
definitions about right.)

One-bids have priority over two-bids if qualified for both.

2NT=notrump-type, 20.5 to 22.0 points

2S= spades 5-7 long, no suit longer, <4 hearts, at most one outside A
or K, 6.5 to 11.4 points but do not count honors except A in any
singleton or doubleton suit
2H= hearts 5-7 long, <4 spades, otherwise as above
2D= diamonds 5-7 long, <4 hearts and <4 spades, otherwise as above
except strength 8.5 to 13.4 points

2C= notrump-type >22 points or suit-type too strong for one-level
opener

1NT=notrump type, 11.8-14.8 points

1S= suit-type, 5+ spades, no suit longer, 11.5 to 22 points
1H= suit-type, 5+ hearts, no suit longer, spades shorter than hearts,
11.5 to 22 points

1D= suit-type, 4+ diamonds, diamonds longest or 4441 (not short
diamonds) or clubs=diamonds, 13.5 to 24 points; or notrump-type
14.9 to 20.4 points, doubleton club

1C= suit-type, clubs longest or 4=4=1=4, 13.5 to 24 points; or
notrump-type 3+ clubs 14.9 to 20.4 points

P= not a preempt and not qualified for any of the above openers.

System 2 is sort of a cross between K-S and Nightmare:

1M has priority over 2M if qualified for both, but 2m has priority
over 1m.

2NT=preempt with long minor, no outside A or K, no A or KQ in suit.

2S= spades 5-7 long, no suit longer, <4 hearts, at most one outside A
or K, 6.5 to 11.4 points but do not count honors except A in any
singleton or doubleton suit
2H= hearts 5-7 long, <4 spades, otherwise as above

2D= suit-type 5+ diamonds, <4 hearts and <4 spades, 9-15 points
2C= suit-type 5+ clubs, clubs > diamonds, <4 hearts and <4 spades,
9-15 points

1NT=notrump type, 11.8-14.8 points

1S= suit-type, 5+ spades, spades >= hearts, 11.5 to 21 points
1H= suit-type, 5+ hearts, spades shorter than hearts, 11.5 to 21 points
(note these openings can have a longer minor)

1D= suit-type, not qualified for any of above (can have long clubs,
short diamonds!), 13.5 to 23 points

1C= notrump type >14.8 points; or suit-type too strong for any of
above

P= not a preempt and not qualified for any of the above openers

[tys...@yahoo.com]

wil...@cfa.harvard.edu (Steve Willner) wrote in message news:<37bd12b7.04061...@posting.google.com>...

> I'd like to propose two systems based on the same evaluation
> criteria.

I can use your NT evaluator and BUMRAP+321, but I'm only allowing 1
system per person. Which system would you like to enter?

Tysen

[Mike Bell]

In message <91691fb6.04061...@posting.google.com>
b.bee...@keesing.nl (Bert Beentjes) wrote:

> tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...
>
> > Let's have a contest to see who can design the best system (at least
> > opening bids)!

Evaluation by Fifths count (4.0-2.8-1.8-1.0-0.4) and TSP (Tysen's
evaluation scheme based on 6-4-2-1 honour count).

P Any hand that is <18 TSP or 18 TSP, <10.6 FC, no 5 card major, unless
12-17 TSP with a 6 card suit or suitable for 2H/2S
1C 10.6-13.4 (FC) Bal, or any hand with either 26+ (TSP) or 16.6+ (FC)
1D Min 10.6 FC+19 TSP Max 25 TSP, Not balanced, <6 clubs, <5 major
1H 18-25 TSP, 5+H, H>S
1S 18-25 TSP, 5+S
1N Bal/5422 without 4 spades or 5 hearts; 13.6-16.4 (FC)
2C Min 10.6 FC+19 TSP Max 25 TSP, no 5 card major or diamonds, unbal.
2D 12-17 TSP, 6H/6S
2H 12-17 TSP, 5H/4-5m
2S 12-17 TSP, 5S/4-5m
2N 12-17 TSP, 6C/6D, no 4H/4S, not two top honours in the 6 card suit
unless < 6.8 (FC).

1H, 1S have priority over 1C.
1N has priority over 1D and 2C.
1D has priority over 2C.

Cheers
Mike

[tys...@yahoo.com]

"ENG hank" <e...@ismra.fr> submitted the following system to me by
email since he could not post on rgb. It was submitted on Saturday
before the deadline.

The openings use the 4-3-2-1 Goren count for openings. Singleton K,
Q,
and Js are not counted. Exact distributions are denoted with equal
signs in the order of S=H=D=C.

I'll call it the DHS system for now.

1C Artificial 11-14 points, but denies the ability of making a 1N
(3rd/4th seat) 2C, 2D, or 2N opening (see below).
1D Artificial, forcing opening, 15+. Unbalanced with minor as the
primary suit OR balanced hand without a four card major.
1H Natural, forcing opening, 15+. Hearts are primary suit. Four card
major opening when
* balanced with four hearts (my definition of balanced does
not included a hand like 2=4=2=5 but includes 3=4=2=4).
Holding 44 in the majors, you open 1H.
* holding 4441 distribution
1S Natural, forcing opening, 15+. Spades are primary suit. Four card
major opening when
* balanced with four spades (but not with four hearts)
* holding exactly 4=1=4=4 distribution
1N First or second seat: 8-10 balanced or pseudobalanced. Here are
the
distributions any 3334, 2344, 2245, or 2335. 2236 hands with a
long
minor are opened 1N.
Third or fourth seat: 11-14 balanced hands: any 3334, 2344, 2245
or
2335 distributions.
2C 11-14 with both minors, where both minors are five cards or longer
AND the difference in length between them is at most one card.
So, 2=1=5=5, 2=0=6=5, 2=0=5=6, etc is ok, but 1=0=5=7 is not.
2D 11-14 with both majors, where both majors are five cards or longer
AND the difference in length between them is at most one card.
2H 5-10 six hearts, may have void or side four card suits
2S 5-10 six spades, may have void or side four card suits
2N 9-14 any hand with a seven card suit but
* either a solid suit (AKQxxxx) without a side A or Kx(x)
* or semi-solid suit (at least KQJxxxx) with a side A or Kx(x)
* denies a void or side four card major
* may have side Qs or Js.

PASS all 0-4 hands; 5-7 hands lacking a seven plus card suit OR six
card major; in 1st/2nd seat 8-10 hands lacking a seven plus card
suit, six card major, OR pseudobalanced hand; in 3rd/4th seat
8-10
hands lacking a seven card suit or a six card major.

[tys...@yahoo.com]

Okay, the June 20th deadline has passed. That's all the systems that
are being accepted at this time. Give me another week or so to gather
the results. I have the following systems in my records:

Bert unlim 1-bids, tr open
Bruce strong pass
BY strong pass
Chris nat 4.5cM, aggr 1N-2S
Dave precision
Gerben weak/strong pass
Hank unlim 1-bids
Jugoslav strong diamond
Kent suction openings
Larry canape precision
Marcel strong pass
Mike 1C bal or strong
Nancy 1C nat,bal, or strong
Richard moscito
Ron strong pass
Sandy natural strong 2s
Steve nightmare/KS
Tony natural canape

Looks like 18 systems. Let me know if I've missed a post and your
name isn't on the list.

[akhare1]

tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...

I have an idea that I posted to a different forum. Basically, a
variation on Ron's strong pass that passed intermediate hands (13 -
17) and multiplexes the fert (0 - 8) w/ 18+ hands :):

Pass: 13+ - 17
1C: 0 - 8 OR 18+ (forcing fert :)
1D: 4+H; 9 - 13-
1H: 4+S; 9 - 13-
1S: 4+D; 9 - 13-
1N: Balanced; 9 - 13
2C: 9 - 13-; Clubs
2D: Multi OR Wilcowz OR Frelling 2
2H/2S: Muiderberg OR Frelling 2
2N: Weak 3 in Minor / Weak 2

Over the limited openings Symmetric relays are used w/ inv+ hands.
Over the forcing fert :), all bids except 1N show 3+ cards and are NF.

Atul

[Gerben Dirksen]

tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...

You might want to compare this with some standard systems:

Std. American
Acol
Polish Club
Classical Precision (with 12-15 NT, Mini-Roman and 5-card majors)

Gerben

[Mike Bell]

In message <fa0f26c...@RiscPC02.bellfamily.org.uk>
Mike Bell <mi...@bellfamily.org.uk> wrote:

Oops, I've just realised this is opening 4414 shapes 1D. If possible
could I correct 'Bal' to 'Bal or 4414' for both 1C and 1NT, giving 1C
priority over 1D? If for any reason you would rather not change it now
then I will accept the losses of my carelessness :))

Mike

[tys...@yahoo.com]

Still working on the results. But in the mean time I'd like to turn
this thread into a discussion on theory. That was the second purpose
of this contest: to create topics of discussion.

Anyone want to take a stab at their predictions of what will do well
and more importantly why? If you've looked over a system, did you see
some bids that you thought were very well designed? Did you see some
that were poorly designed?

[akhare1]

tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...

I think that both Moscito and FP will do pretty well with their
aggressive 4CM limited opening bids. I would have liked the FP system
to have a medium pass range, i.e., narrow the pass range from 14 - 40
to something like 14-20 and multiplex the 21+ hands w/ some other bids
(a 1C fert :)!!!

BTW, is this going to an iterative process? Or will the results of
your tests pretty much pinpoint the shortcomings of each system?

Atul

[Tony Warnock]

tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...

> But in the mean time I'd like to turn
> this thread into a discussion on theory.

Just a quick question: what are the pros and cons of the Roman 1C
opening on the weak NT hands vs opening a Weak 1NT?

[p.paulf...@comcast.net]

"Tony Warnock" <t...@lanl.gov> wrote in message
news:33581e3c.04062...@posting.google.com...

Ignoring what happens to the rest of the system ...

Pros: bidding starts lower
much tighter distribution (4333 or 4432; 5332 extremely
rare; 5422 and 6322 almost unheard of)

Cons: Bidding starts lower (it can work either way)
very wide range (12-16)

Leghorn Diamond (an offshoot of Roman) tighten the range a bit to
12-15.

Paul

[tys...@yahoo.com]

Boy it was a close race, but the winner is… Dave!

NAME STYLE CONST DESTR TOTAL
Dave precision -0.054 0.125 0.071
Ron strong pass -0.203 0.273 0.070
Mike 1C bal or strong -0.108 0.145 0.037
Nancy 1C nat,bal, or strong -0.035 0.061 0.026
Marcel strong pass -0.278 0.294 0.016
Sandy natural strong 2s 0.061 -0.048 0.013
Larry canape precision -0.085 0.093 0.008
Tysen standard American 0.000 0.000 0.000
Gerben weak/strong pass -0.216 0.210 -0.006
Steve nightmare/KS -0.051 0.042 -0.009
Tony natural canape -0.014 0.003 -0.010
Bruce strong pass -0.338 0.301 -0.037
Hank unlim 1-bids, 8-10 NT -0.234 0.172 -0.061
BY strong pass -0.424 0.352 -0.072
Kent suction openings -0.383 0.306 -0.077
Bert unlim 1-bids, tr open -0.124 0.018 -0.106
Richard moscito -0.212 0.089 -0.123
Chris nat 4.5cM, aggr 1N-2S -0.325 0.151 -0.174
Jugoslav strong diamond -0.357 0.067 -0.290

A brief description on what the numbers mean. All of these numbers
are IMPs per board in comparison to the Standard American system I
described in the original post. Large positive numbers are good.
CONST is how well we can reach our optimal contract, with and without
interference. DESTR is how well we keep out opponents out of their
optimal contract. DESTR is how well you preempt *effectively*. To
explain what I mean imagine you define a 7S opening as your 0-7 fert.
That's very preemptive, but it's not very *effective* as it is now
very easy for your opponents to find their optimum contract…

All of the systems (except Sandy's) have worse constructive openings
and better destructive than Standard American. That's because almost
all of the systems open lighter than SA, which is usually a good
thing. If you can keep the total positive, then it's worth it.

Let me comment specifically on the top 3 and bottom 3 systems to let
you know what the simulations thought were the key features.

TOP 3

Dave (Precision): Dave also wins the prize for shortest description I
think. His 1D, 1H, 1S, and 2C bids are all well defined and promise
5+ cards. The simulations greatly prefer this over a more nebulous
diamond. No one bid stands out, just overall very well defined.

Ron (strong pass): The simulations hammer the fert pretty hard. It
doesn't think it will do well against prepared opponents. But the
fert does help the rest of the system… The 1C and 1D bids save the
day. Using 1C/1D to describe the majors is huge, because you can use
the extra bidding space to define yourself better. A problem with
"standard" 1C/1D openings is that they are your most common openings,
yet when you open them, you would still most likely prefer to be in a
major contract. So most of time you'd prefer to be somewhere besides
the feature you've named. Since Ron's 1C/1D bids are essentially all
hands with 4cM, the other bids can focus more on non-major contracts.

Mike (1C bal or strong): Similar to Dave's in that no one bid stands
out, but the whole system is pretty good. I'd love to say TSP (my
point-count system) is the reason Mike did well but I don't think it
was a major factor. I guess I was personally hoping that Mike's
system would do well, but not so well that people would accuse me of
favoritism. =) Actually I've found that including distribution
strength in your bid definitions (even shapeless bids like a strong
1C) is better than just HCP. So a definition like 1C = 18+ with
distribution is better than 16+ HCP.

BOTTOM 3

Richard (MOSCITO): The constructive openings are good, but the
simulation thinks that the 2D/2H/2S preempts are a bit too much. A
large percentage of the time you don't have a major fit and it can be
tough to handle the misfits. The opponents are assumed to be
well-prepared for your methods (that doesn't mean perfect) and so they
can often penalize you when it's right. The 2S opening is often 6+
cards, but since it is also often only 4, that can be a nasty guess
for responder.

Chris (natural 4.5cM, aggressive 1N-2S): The 9-12 NT that can include
5cM and also 5422 is asking for trouble according to the simulations.
There are many times where you want to be in a suit partscore and you
just can't find it. The amount of preemption you get is not worth the
loss in constructive accuracy. The simulation has the same problems
with preempts as it does for Richard's.

Jugoslav (strong diamond): The preempts are just too much as they are
currently defined. Since the weak 2 can be made on a 5332 as well as
more extreme shapes, responder is going to have trouble. The
opponents have trouble too, but well-prepared ones will be able to
penalize you too often. A tighter definition on shape and/or suit
quality will help. For example, Steve had a similar weak 2 except
that it was never balanced and never had 4 cards in the other major.

GENERAL COMMENTS:

Dave's system is far from perfect according to the simulations. You
can do some tweaking by opening lighter (when unbalanced) and perhaps
lowering the 1C strength as well. Numerous things can be done to all
the systems. There are also probably some good ideas that are unlike
any others, but were never submitted.

I was hoping for good results from someone with unlimited 1-bids
(Hank, Bert, Kent) because I think those have promise if the shape is
well-defined. The simulations don't like "any shape" bids, either
weak or strong. Unfortunately all 3 of the unlimited 1-bid systems
had problems. Hank's 1N opening gets him in trouble since it can have
a 5cM as well as any 5422 shape. Also the 2C and 2D openings are
practically totally wasted since they are so rare. Bert's 1C opening
was too nebulous and the 1S opening was often club-oriented, but often
not as well making it tough to stop in the right place. Kent's 1-bids
did okay but the 2M bids were overloaded since they could be only 4
cards long with a much longer side suit.

Winner for the best bid (when it comes up):
Mike's 2D = mini-multi
2nd place: Kent's 2C = weak 2 in any suit!

Worst bid*: Bert's 1C = 13+ no 5cM, or 12+ (any 8+c or 74 or 6+/5+)
2nd place: Kent's 1N = 12-15 balanced (since it's wide, weak, and can
have a 5cM)

*Note that these are the worst bids when they come up. That doesn't
mean that they necessarily bad, because they could be helping the rest
of the system.

Okay I think that's enough for one post. I'd love to start some
discussion.

[Jugoslav Dujic]

tys...@yahoo.com wrote:
| Boy it was a close race, but the winner is. Dave!

|
| NAME STYLE CONST DESTR TOTAL

<snip>

| Jugoslav strong diamond -0.357 0.067 -0.290
|
|

| Jugoslav (strong diamond): The preempts are just too much as they are
| currently defined. Since the weak 2 can be made on a 5332 as well as
| more extreme shapes, responder is going to have trouble. The
| opponents have trouble too, but well-prepared ones will be able to
| penalize you too often. A tighter definition on shape and/or suit
| quality will help. For example, Steve had a similar weak 2 except
| that it was never balanced and never had 4 cards in the other major.

Now it's losers' time to groan that your evaluation method is ridiculous,
full of b**t et cetera....

Just kidding, it was fun. Now seriously, I regarded the weak-2 part
just as an "ammendment" to the basic system and didn't pay much attention
to describing it. It's not so undisciplined in practice, and we don't
open it with lousy suits, and 5-baggers have to be at least KQ109x or
better, and vulnerable are avoided. Other 4cM is not allowed either.
My description was just a sketch of style, and I doubt we're more
aggressive w2 openers than at least half of rgb'ers. And I've had
a lot of success (and only occasional bad results) in opening 5cM w2's
in the past so you didn't convince me...

| I was hoping for good results from someone with unlimited 1-bids
| (Hank, Bert, Kent) because I think those have promise if the shape is
| well-defined. The simulations don't like "any shape" bids, either
| weak or strong.

I understand ambiguous 1C and 1D are devalued in the evaluation method,
but we didn't have much problems with it, as strong opening is purposefully
moved to 18+, so our opponents don't interfere much (and we collected
some juicy penalties from those who felt inclined to bid over 1D with 13
cards). 1C is bit tougher case, but since it's either semi-strong (15-17)
or a shapely 2/3 suiter, it's not so tough to get on your feet in competition
when bidding heats up.

I wonder how much of my bad result was due to loose preempt definition,
and how much due to the rest of the system.

<groan>
Hope I'll beat you at the bridge table one day if I couldn't here.
</groan>

Regards

--
Jugoslav
___________
www.geocities.com/jdujic

Please reply to the newsgroup.
You can find my real e-mail on my home page above.

[Bert Beentjes]

First, thank you for all your effort!

I presume the CONST scores are from hands where our side has the
absolute par score, and the DESTR scores are from the same hands, but
with sides switched?

And if possible, I'd like to see the scores per opening bid in a
matrix. This would reveal a little more on how the scores are
composed.

And of course, I'm a little disappointed by my score... But then
again, I could see the bad 1C score coming ;-)

[Richard Willey]

> Richard (MOSCITO): The constructive openings are good, but the
> simulation thinks that the 2D/2H/2S preempts are a bit too much. A
> large percentage of the time you don't have a major fit and it can be
> tough to handle the misfits. The opponents are assumed to be
> well-prepared for your methods (that doesn't mean perfect) and so they
> can often penalize you when it's right. The 2S opening is often 6+
> cards, but since it is also often only 4, that can be a nasty guess
> for responder.

Sniff, sniff, sniff. Looks like I might need to go back to the
drawing board.
Pity, cause I really enjoy the Frelling Twos and they do seem to be
scoring pretty well.

Any chance that I might be able to convince you to run a second
simulation with a different preemptive structure:

In particularly, I'd be interested in seeing the following grafted
onto the core system.

2D = Weak 2 in Hearts or 5S + 5m 4-9
2H = weak 2 in Spades or 5H + 5m 4-9
2S = weak minor preempt or 5/5 Majors 4-9
2N = 5/5 minors, 5 - 10

[HWE]

tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...
> Boy it was a close race, but the winner is? Dave!

>
> NAME STYLE CONST DESTR TOTAL

> Dave precision -0.054 0.125 0.071

>...

> Hank unlim 1-bids, 8-10 NT -0.234 0.172 -0.061

>...
> GENERAL COMMENTS:

> I was hoping for good results from someone with unlimited 1-bids
> (Hank, Bert, Kent) because I think those have promise if the shape is
> well-defined. The simulations don't like "any shape" bids, either

> weak or strong. Unfortunately all 3 of the unlimited 1-bid systems
> had problems. Hank's 1N opening gets him in trouble since it can have
> a 5cM as well as any 5422 shape. Also the 2C and 2D openings are
> practically totally wasted since they are so rare.

>...

I was also hoping for good results with my unlimited one bids. :-) I
guess there's a reason I've never played this system in actual
competition before.
As you noted, my 1C bid (11-14 almost any shape) was probably too
vague, and I was quickly making up 1N, 2C, and 2D bids that I thought
we be interesting to test. If I were to find a crazy partner to play
this system, I'd probably change the bids to show 11-14 hands with
different distributions: 1C (any hand with a four card major), 1N
(balanced without a four card major), and 2C/2D (other minor suited
hands).

Normally I play a Big Club system with either four or five card majors
(depending on the partner), very similar to Dave's proposal.

By the way Tysen, as Gerben suggested and if you have the time, I do
think you should try running simulations with the more commonly known
systems other than Standard American, like Acol, Polish Club,
Classical Precision, Swedish Club, Italian Blue Club, and Romex.

-Hank Eng

tys...@yahoo.com relayed my email when he wrote in message news:<a1bce703.04062...@posting.google.com>...

> "ENG hank" <e...@ismra.fr> submitted the following system to me

...

> 1C Artificial 11-14 points, but denies the ability of making a 1N
> (3rd/4th seat) 2C, 2D, or 2N opening (see below).
> 1D Artificial, forcing opening, 15+. Unbalanced with minor as the
> primary suit OR balanced hand without a four card major.
> 1H Natural, forcing opening, 15+. Hearts are primary suit.

...

> 1S Natural, forcing opening, 15+. Spades are primary suit.

...

> 1N First or second seat: 8-10 balanced or pseudobalanced. Here are
> the distributions any 3334, 2344, 2245, or 2335. 2236 hands
> with a long minor are opened 1N.
> Third or fourth seat: 11-14 balanced hands: any 3334, 2344, 2245
> or 2335 distributions.
> 2C 11-14 with both minors, where both minors are five cards or longer
> AND the difference in length between them is at most one card.
> So, 2=1=5=5, 2=0=6=5, 2=0=5=6, etc is ok, but 1=0=5=7 is not.
> 2D 11-14 with both majors, where both majors are five cards or longer
> AND the difference in length between them is at most one card.

...

[tys...@yahoo.com]

b.bee...@keesing.nl (Bert Beentjes) wrote in message news:<91691fb6.04062...@posting.google.com>...

> First, thank you for all your effort!
>
> I presume the CONST scores are from hands where our side has the
> absolute par score, and the DESTR scores are from the same hands, but
> with sides switched?
>

I'm not sure I've understood you correctly, but I think the answer is
no. In all the hands, we're opening in 1st seat. Each side is trying
to reach their optimum contract. For example, if we can take 9 tricks
in spades and they can take 10 tricks in hearts, our optimum contract
is 4S while they want 4H. Our par is negative, but if we don't push
to 4S and sell out at 4H, then we lose points. If they let us have 3S
or overbid to 5H then they lose points. Make sense?

> And if possible, I'd like to see the scores per opening bid in a
> matrix. This would reveal a little more on how the scores are
> composed.

I'll post the matrix, but don't expect to get a lot out of it:

NAME PASS 1C 1D 1H 1S 1N 2C+ TOTAL
Dave 0.134 -0.029 0.003 0.006 0.012 -0.070 0.015 0.071
Ron 0.059 0.028 -0.006 -0.040 -0.030 0.005 0.054 0.070
Mike 0.096 -0.036 -0.011 0.009 0.037 -0.052 -0.006 0.037
Nancy 0.140 -0.049 -0.005 0.000 -0.009 -0.029 -0.023 0.026
Marcel 0.059 -0.002 -0.086 0.028 -0.004 0.013 0.008 0.016
Sandy 0.084 -0.021 0.001 0.020 0.007 -0.056 -0.023 0.013
Larry 0.134 -0.029 -0.004 0.000 -0.044 -0.026 -0.024 0.008
Tysen 0.046 -0.028 0.011 0.017 0.004 -0.056 0.006 0.000
Gerben 0.056 -0.007 -0.022 0.004 0.027 -0.036 -0.027 -0.006
Steve 0.071 -0.053 -0.024 0.018 0.028 -0.062 0.014 -0.009
Tony 0.127 -0.028 -0.017 -0.006 -0.039 -0.043 -0.006 -0.010
Bruce 0.059 -0.002 -0.086 0.010 -0.009 0.007 -0.017 -0.037
Hank 0.041 0.005 -0.033 -0.033 -0.020 -0.067 0.047 -0.061
BY 0.026 -0.032 -0.060 0.001 0.003 -0.041 0.031 -0.072
Kent 0.112 0.012 -0.001 -0.021 -0.005 -0.090 -0.083 -0.077
Bert 0.076 -0.098 0.003 -0.007 -0.029 -0.003 -0.048 -0.106
Richard 0.071 -0.055 -0.009 -0.018 -0.017 -0.040 -0.054 -0.123
Chris -0.005 -0.017 -0.011 0.002 -0.002 -0.055 -0.087 -0.174
Jugoslav -0.049 -0.022 -0.038 0.008 0.000 -0.061 -0.128 -0.290

The reason this isn't that valuable is that these scores are all
interrelated, so it's the total score that really matters. These
scores are affected by the definitions of other bids, and each bid's
score doesn't reflect its effect on the rest of the system. For
example, Dave's 1C bid is given a minus score. This doesn't mean that
his 1C is a bad bid because his 1C definition is helping out all of
the other bids. Pass is also a strange customer. Pass is naturally
affected by all the other definitions and it helps refine all of the
other bids as well. It's really hard for me to put my finger on what
makes a good or bad Pass.

It *is* possible for me to come up with how good each bid's definition
is, including the benefits it gives to other bids, but that takes at
least twice as much work as this already took. Plus I'm getting a lot
of requests both here and by email by people wanting me to refine
their bids and see what their scores are. I'm not sure when I'm going
to be able to get to those requests.

Tysen

[tys...@yahoo.com]

"Jugoslav Dujic" <jdu...@yahoo.com> wrote in message news:<2kcrct...@uni-berlin.de>...

> Just kidding, it was fun. Now seriously, I regarded the weak-2 part
> just as an "ammendment" to the basic system and didn't pay much attention
> to describing it. It's not so undisciplined in practice, and we don't
> open it with lousy suits, and 5-baggers have to be at least KQ109x or
> better, and vulnerable are avoided. Other 4cM is not allowed either.
> My description was just a sketch of style, and I doubt we're more
> aggressive w2 openers than at least half of rgb'ers. And I've had
> a lot of success (and only occasional bad results) in opening 5cM w2's
> in the past so you didn't convince me...

I think a tighter description would increase your score tremendously.
It's probably because you didn't word it the way you really wanted to.
Your original description would open 2H on

xxx
xxxxx
xxx
AQ

>
> | I was hoping for good results from someone with unlimited 1-bids
> | (Hank, Bert, Kent) because I think those have promise if the shape is
> | well-defined. The simulations don't like "any shape" bids, either
> | weak or strong.
>
> I understand ambiguous 1C and 1D are devalued in the evaluation method,
> but we didn't have much problems with it, as strong opening is purposefully
> moved to 18+, so our opponents don't interfere much (and we collected
> some juicy penalties from those who felt inclined to bid over 1D with 13
> cards). 1C is bit tougher case, but since it's either semi-strong (15-17)
> or a shapely 2/3 suiter, it's not so tough to get on your feet in competition
> when bidding heats up.

I've done tests on strong openings. If it makes you feel any better,
the simulations feel that 1D (18+) is much better than 1C (16+).

>
> I wonder how much of my bad result was due to loose preempt definition,
> and how much due to the rest of the system.

I think most of it the the loose preempt definition. Take a look at
the specific bid matrix I posted. Yours it pretty clearly coming from
the preempts.

[Mike Bell]

In message <a1bce703.0406...@posting.google.com>
tys...@yahoo.com wrote:

[Big snip on bidding system evaluation]

Cheers Tysen.

Am surprised the mini-multi did well, everyone always says that a multi
is much easier to defend against than natural weak 2s. Not surprised
Richard's 2S (4S5C or 6S weak) did badly, while it often allows you to
make the right guess, when you do not the consequences can be painful!
But I would have expected his 2D and 2H to do better.

What made Nancy's pass so great?! Was it opening all 11 counts? Yet
Dave's is good too... I expected Nightmare-style systems to do quite
well, with a concentration on shape rather than strength, but still
getting across the strength of strong NTs in one bid. I'd like to see
how this would do:

1C - 15+ C/Bal (not including any GF, this makes 1C-(1H)-P-(4H), ?
nicer!)
1D - 4D5C or unbal primary D
1H, 1S - 5 card majors
1N - 12-14, bal or 22(54), no 5M
2C - 11-14 6C or 5C4M
2D - minimulti
2M - lucas
2N - ?

How did the precision style 2C openers do? If my 2D was the best bid,
what did I do to get a negative on 2C and above? :)

How much do systems vary from 18% efficiency in follow-ups? I have a
feeling relays may be better than this, as long as they start low
enough. Is it possible to design 100% efficient follow-ups?

Have you noticed much difference from fiddling around with evaluation
methods? Have you tried any systems that have come out more efficient
than those submitted?

Cheers
Mike

[Bert Beentjes]

tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...

> b.bee...@keesing.nl (Bert Beentjes) wrote in message news:<91691fb6.04062...@posting.google.com>...
> > First, thank you for all your effort!
> >
> > I presume the CONST scores are from hands where our side has the
> > absolute par score, and the DESTR scores are from the same hands, but
> > with sides switched?
> >
> I'm not sure I've understood you correctly, but I think the answer is
> no. In all the hands, we're opening in 1st seat. Each side is trying
> to reach their optimum contract. For example, if we can take 9 tricks
> in spades and they can take 10 tricks in hearts, our optimum contract
> is 4S while they want 4H. Our par is negative, but if we don't push
> to 4S and sell out at 4H, then we lose points. If they let us have 3S
> or overbid to 5H then they lose points. Make sense?

So this example would add to the const scores? And in an example where
our par is 4H and theirs is 4S it adds to the destr scores?

> > And if possible, I'd like to see the scores per opening bid in a
> > matrix. This would reveal a little more on how the scores are
> > composed.
>
> I'll post the matrix, but don't expect to get a lot out of it:

Thanks.

> NAME PASS 1C 1D 1H 1S 1N 2C+ TOTAL
> Dave 0.134 -0.029 0.003 0.006 0.012 -0.070 0.015 0.071

It looks like the total is the sum of the parts. But shouldn't the
parts be weighed for frequency? It seems to me that if you pass 90% of
the time and score -1 imps on average, and bid 10% of the time and
score +9 imp on average, the total should be 0 imps, not +4 imps.

> The reason this isn't that valuable is that these scores are all
> interrelated, so it's the total score that really matters.

Of course, but there are parts of systems that (almost) equal in
content, but that differ in the distribution over bids. And I assume
that there is a gradient, so that small changes in bid definitions
result in small changes in score.

> It *is* possible for me to come up with how good each bid's definition
> is, including the benefits it gives to other bids, but that takes at
> least twice as much work as this already took. Plus I'm getting a lot
> of requests both here and by email by people wanting me to refine
> their bids and see what their scores are. I'm not sure when I'm going
> to be able to get to those requests.

The good thing is that people are interested :-) Thanks for your work
so far. If there is any way I can help you with this, please let me
know. I think the general approach has a bright future.

[Bert Beentjes]

> NAME PASS 1C 1D 1H 1S 1N 2C+ TOTAL
> Dave 0.134 -0.029 0.003 0.006 0.012 -0.070 0.015 0.071

I think I know the answer to my question about the total already. The
scores per bid are already averages on all deals, they are not
averages on the deals where the specific bid is used. This makes
sense.

Bert

[Peter Leighton]

Congratulations to Dave Flower on your winning system!

Where do you think this diabolically original system should
be permitted to be played, in the interest of fairness?

Peter

[DavJFlower]

Obviously against other perpretators of equally diabolical systems!

Dave Flower

PS I know that it's confusing that 1D means diamonds, 1H promises hearts, 1S
shows spades, and 2C indicates clubs!

[tys...@yahoo.com]

b.bee...@keesing.nl (Bert Beentjes) wrote in message news:<91691fb6.04062...@posting.google.com>...
> tys...@yahoo.com wrote in message news:<a1bce703.04062...@posting.google.com>...
> > b.bee...@keesing.nl (Bert Beentjes) wrote in message news:<91691fb6.04062...@posting.google.com>...
> > > First, thank you for all your effort!
> > >
> > > I presume the CONST scores are from hands where our side has the
> > > absolute par score, and the DESTR scores are from the same hands, but
> > > with sides switched?
> > >
> > I'm not sure I've understood you correctly, but I think the answer is
> > no. In all the hands, we're opening in 1st seat. Each side is trying
> > to reach their optimum contract. For example, if we can take 9 tricks
> > in spades and they can take 10 tricks in hearts, our optimum contract
> > is 4S while they want 4H. Our par is negative, but if we don't push
> > to 4S and sell out at 4H, then we lose points. If they let us have 3S
> > or overbid to 5H then they lose points. Make sense?
>
> So this example would add to the const scores? And in an example where
> our par is 4H and theirs is 4S it adds to the destr scores?

The scores are always an "IMPs from perfection." So on this example,
if they let us have 3S, our CONST is 0 IMPs (perfect) and we get a
DESTR bonus of 11 IMPs. If we sell out at 4H we lose 10 IMPs on CONST
since our optimal contract is now 4S and DESTR is 0 (they bid
perfectly). If we bid on to 4S and they let us have it, CONST and
DESTR are both 0 since both sides bid perfectly.

>
> > > And if possible, I'd like to see the scores per opening bid in a
> > > matrix. This would reveal a little more on how the scores are
> > > composed.
> >
> > I'll post the matrix, but don't expect to get a lot out of it:
>
> Thanks.
>
> > NAME PASS 1C 1D 1H 1S 1N 2C+ TOTAL
> > Dave 0.134 -0.029 0.003 0.006 0.012 -0.070 0.015 0.071
>
> It looks like the total is the sum of the parts. But shouldn't the
> parts be weighed for frequency? It seems to me that if you pass 90% of
> the time and score -1 imps on average, and bid 10% of the time and
> score +9 imp on average, the total should be 0 imps, not +4 imps.
>

You answered your own question in another post. Yes, the scores for
each bid are already weighted by frequency.

Tysen

[tys...@yahoo.com]

Mike Bell <mi...@bellfamily.org.uk> wrote in message news:<282fd7c...@RiscPC02.bellfamily.org.uk>...

> In message <a1bce703.0406...@posting.google.com>
> tys...@yahoo.com wrote:
> What made Nancy's pass so great?! Was it opening all 11 counts? Yet
> Dave's is good too...

Like I said, each bids score is really influenced by the definitions
of the other bids. A high pass score can just be from well-defined
other bids. That's why I was reluctant to post the individual bid
scores, it's the total of the system as a whole that really matters.

>
> How did the precision style 2C openers do? If my 2D was the best bid,
> what did I do to get a negative on 2C and above? :)

I've looked at Precision style 2C openers before. I know the
simulations prefer a 6-card club suit that denys a 4cM compared to
allowing 5 w/ 4cM.

>
> How much do systems vary from 18% efficiency in follow-ups? I have a
> feeling relays may be better than this, as long as they start low
> enough. Is it possible to design 100% efficient follow-ups?

Part of the reason the follow ups are less than 100% is that the
opponents sometimes actually say something! So you'll never get to
100%. On Ginsberg's original test of this method, he set the
effeciency to 23%. The system he generated as the best ended up
passing in 1st seat 80% of the time! It always assumed the follow ups
would be so much better than what it could think of so it *never*
wanted to eat up any bidding space if possible!

>
> Have you noticed much difference from fiddling around with evaluation
> methods? Have you tried any systems that have come out more efficient
> than those submitted?
>

Of course. =) But I decided from the beginning that this was a
contest for you guys. I have an unfair advantage since I can test and
retest and see what works.

[Daniel Morgan]

Too late for the contest, I know, but ... Have you tried using this
method to look at minor changes to a particular system rather than to
compare entirely different systems?

For example, suppose you took Standard American as described by you
but instead of

> 1NT = 15-17 HCP any 5332, 4432, or 4333 shape

you specified

1NT = 15-17 HCP any 5332, 4432, or 4333 shape but no 5-card major.

How would that score?

Is this a useful way of thinking about this type of old chestnut?

[akhare1]

tys...@yahoo.com wrote in message news:<a1bce703.04063...@posting.google.com>...
<snip>

> Of course. =) But I decided from the beginning that this was a
> contest for you guys. I have an unfair advantage since I can test and
> retest and see what works.

For curiosity's sake, since you have run these tests hundreds of
times, what does your dream system look like?

Atul

[HWE]

davjf...@aol.com (DavJFlower) wrote in message news:<20040630070941...@mb-m13.aol.com>...

You should try some country that lets you play those strange things.
Maybe Taiwan or Italy, but surely not in the US.

-Hank

[Gerben Dirksen]

tys...@yahoo.com wrote in message news:<a1bce703.0406...@posting.google.com>...
> Boy it was a close race, but the winner is? Dave!

>
> NAME STYLE CONST DESTR TOTAL
> Dave precision -0.054 0.125 0.071
> Ron strong pass -0.203 0.273 0.070
> Mike 1C bal or strong -0.108 0.145 0.037
> Nancy 1C nat,bal, or strong -0.035 0.061 0.026
> Marcel strong pass -0.278 0.294 0.016
> Sandy natural strong 2s 0.061 -0.048 0.013
> Larry canape precision -0.085 0.093 0.008
> Tysen standard American 0.000 0.000 0.000
> Gerben weak/strong pass -0.216 0.210 -0.006
> Steve nightmare/KS -0.051 0.042 -0.009
> Tony natural canape -0.014 0.003 -0.010
> Bruce strong pass -0.338 0.301 -0.037
> Hank unlim 1-bids, 8-10 NT -0.234 0.172 -0.061
> BY strong pass -0.424 0.352 -0.072
> Kent suction openings -0.383 0.306 -0.077
> Bert unlim 1-bids, tr open -0.124 0.018 -0.106
> Richard moscito -0.212 0.089 -0.123
> Chris nat 4.5cM, aggr 1N-2S -0.325 0.151 -0.174
> Jugoslav strong diamond -0.357 0.067 -0.290

Great work!

I expected to do well in the "destructive" department but I would've
expected more from:

a) 1NT (12 - 14 V and 9 - 11 NV)
b) 1D (Natural and 4+cards, but 8 - 16 instead of 11 - 19)

I wonder how my 2-bids did. If Mini-Multi did well, what about my
Major Flash (2H = weak two in H or S) and Wilkosz?

Gerben

[Tysen Streib]

daniel....@gmail.com (Daniel Morgan) wrote in message news:<ce068bec.04063...@posting.google.com>...

Absolutely. Changing this will change 3 bids: 1H, 1S, and 1N. You
can (1) keep the rest of the system the same and see how it scores.
You can also try to see how a bid's definition is rated in general by
(2) effectively defining just the one bid and leaving the rest of the
system "undefined" and the simulations fill in the blanks at 18%
effeciency. So if one believes the results of the simulation
absolutely (questionable!) then one could settle the argument of which
is better: strong or weak NT, 5cM or not, etc.

I've done some work on this and found that while the second way
answers the question of what type of NT is better "in general" you
really need to do it the first way and fit it into your particular
system since you often get different results. I've found that in
general, the simulations think 1N w/ 5cM is better mainly because it
increases the frequency of the 1N opening. But when you look at a
specific system (say the SA system I posted) it sometimes says that
it's better to *not* have the NT contain 5cM. I guess it depends on
the definitions of your major bids. In the SA case, it wants as many
hands as possible to open 1M since those are wonderful bids in its
eyes. So puting those hands into 1N is a mistake.

I will say that in general, the simulations don't see that much
differnence between strong and weak NT (maybe that's why no one can
agree on this!) - it does say strong is *slightly* better. One thing
the simulations do say is that the weaker your NT becomes, the worse
it is to include a 5cM or even a 4cM. This makes sense from a theory
point of view in my eyes. The weaker the 1NT opening, the more likely
you miss a good major partscore.

[Mike Bell]

In message <a1bce703.04070...@posting.google.com>
tys...@yahoo.com (Tysen Streib) wrote:

> So if one believes the results of the simulation
> absolutely (questionable!) then one could settle the argument of which
> is better: strong or weak NT, 5cM or not, etc.

My understanding of your methods isn't good enough to see what might
cause the simulation to not be perfect. If it won't take too long,
could you explain where might errors come from? You mentioned Matt
Ginsberg's attempts to create a system using this method, I found his
work on RGB but only as far as the 2C opener (which showed something
like 23-24 points 3334 shape IIRC)! Do you know what his 1 level bids
came out as?

> I will say that in general, the simulations don't see that much
> differnence between strong and weak NT (maybe that's why no one can
> agree on this!) - it does say strong is *slightly* better. One thing
> the simulations do say is that the weaker your NT becomes, the worse
> it is to include a 5cM or even a 4cM. This makes sense from a theory
> point of view in my eyes. The weaker the 1NT opening, the more likely
> you miss a good major partscore.

Very interesting. Is there a noticeable difference between allowing 4/5
card heart suits and 4/5 card spade suits?

Mike

[Tysen Streib]

akh...@my-deja.com (akhare1) wrote in message news:<20bb4527.04063...@posting.google.com>...

Most of my work has been focused on developing a system that is:
(1) GCC legal so I can actually play it
(2) easy to remember

I've experimented with some non-GCC stuff, but most of it has been
GCC. The system I currently play with my regular partner is a
variation on matchpoints precision:

In general, balanced hands are opened with 11 HCP, unbalanced hands
are opened with 8 HCP.

1C = 17+ Total Points (with distribution)
1D = 10 HCP-16 TP, artificial - promises a 4cM
1M = 8 HCP-16 TP, 5+ cards, longer minor OK
1N = 11-15 balanced, no 4cM
2m = 8 HCP-16 TP, 6+ cards, no 4cM
2M = weak
2N = 8 HCP-16 TP, 54+ minors (either way), no 4cM

My current focus is on developing something that is Mid-Chart legal.
It's a work in progress that I'm calling the "Shape System" for lack
of a better name. It's a system that emphasises shape over strength.
Like I said, it's a WIP, but this is the current version:

This system uses TSP evaluation. HP = Honor Points (6-4-2-1 for AKQJ
plus 1 point for every suit that has 2+ honors including the Ten).
TSP = HP + 5/3/1 for void/singleton/doubleton + 1 for every card over
4 in length. Average TSP score is ~1.5 times as much as "normal"
points.

In deciding whether to open or pass, hands should look at TSP but
*triple* the distribution component. Hands that are 20 or more with
tripled distribution should open. Notice that passing could include a
normal 14 HCP 4333 hand.
In deciding between a preemptive and constructive open, in general
hands that have up to 12 HP should be preempts, 13+ should be
constructive.

1C = all balanced hands up to 31 HP
1D = const, up to 29 TSP, either 44+ in both M or no 4cM
1M = const, up to 29 TSP, 4+ in M, 0-3 in OM
(1D and 1M bids are always unbalanced)

1N = 30+ TSP any shape but always unbalanced
2C = preempt, 2-suiter w/ S or 32-34 HP balanced
2D = 6-card preempt in a M, or 35+ HP balanced
2H = preempt, hearts + minor
2S = preempt in a minor
2N = good preempt in hearts
3C = preempt, both minors
3D = good preempt in spades
3M = poor preempt
3N = minor preempt
4m = Namyats
4M = poor preempt

I've never tried this at the table, but I think this would be fun.
Simulations predict the opponents are going to have a tough time
overcalling the 1D bid!
Also the 1C bid is fairly resistant to preemption. Since opener is by
definition defensively shaped, responder can pull out a penalty double
more often and also use negative free bids (transfers anyone?) knowing
there will be at least 2+ support.

Tysen

[KWSchneider]

On 1 Jul 2004 11:13:31 -0700, tys...@yahoo.com (Tysen Streib) wrote:

>akh...@my-deja.com (akhare1) wrote in message news:<20bb4527.04063...@posting.google.com>...
>> tys...@yahoo.com wrote in message news:<a1bce703.04063...@posting.google.com>...
>> <snip>
>> > Of course. =) But I decided from the beginning that this was a
>> > contest for you guys. I have an unfair advantage since I can test and
>> > retest and see what works.
>>
>> For curiosity's sake, since you have run these tests hundreds of
>> times, what does your dream system look like?
>>
>
>Most of my work has been focused on developing a system that is:
>(1) GCC legal so I can actually play it
>(2) easy to remember
>
>I've experimented with some non-GCC stuff, but most of it has been
>GCC. The system I currently play with my regular partner is a
>variation on matchpoints precision:
>
>In general, balanced hands are opened with 11 HCP, unbalanced hands
>are opened with 8 HCP.
>
>1C = 17+ Total Points (with distribution)
>1D = 10 HCP-16 TP, artificial - promises a 4cM
>1M = 8 HCP-16 TP, 5+ cards, longer minor OK
>1N = 11-15 balanced, no 4cM
>2m = 8 HCP-16 TP, 6+ cards, no 4cM
>2M = weak
>2N = 8 HCP-16 TP, 54+ minors (either way), no 4cM
>
>

Tysen

I play a very similar system, eerily similar - couple of questions:

1) How do you define TP?
2) How do you handle 5m4m31 hands of 10-16TP?
3) How do you handle major 8-16TP 2-suiters - 54 or 64 either
way?
4) Do you roll all 5M332 10-16TP into 1N?
5) What are your NT ranges? ie 1C-1D-1N=16-19? How do you handle
the 22-23 range taken by 2N weak?

I can't tell how remarkably similar the system is including 11-15
Notrumps without a 4card major. We include 5carders [weak only - with
14-15 we open 1M followed by 2N].

KurtSchneider
NH, USA

[Frans Buijsen]

tys...@yahoo.com wrote:

It's interesting to see that almost all systems (except Sandy's "never
preempt" system) score worse than Standard American on constructive
bidding.
The rest of your post seems to indicate that a balanced choice of how to
preempt is actually the most important in getting results.

It would be interesting to see how specific bids score. The few scores
I've seen seemed to put a very high penalty on ambiguity, where I think
it may well be crucial in real life
Examples: Polish club, or a strong 2C which can also be any weak
varitation. My experience is with the latter, and I regard it as nearly
essential to stop pre-empting over your strong openings.

Now I realize that doing something specifically to stop the opps from
doing something else is quite complex. Is any of that sort of reasoning
included in your simulations?

--
Frans Buijsen (Amsterdam, The Netherlands)

.. Mail to the address above is not read. Try this address instead: ..
.. fab dot usenet at unetmail.nl ..

[Bert Beentjes]

tys...@yahoo.com wrote in message

> Part of the reason the follow ups are less than 100% is that the
> opponents sometimes actually say something! So you'll never get to
> 100%. On Ginsberg's original test of this method, he set the
> effeciency to 23%. The system he generated as the best ended up
> passing in 1st seat 80% of the time! It always assumed the follow ups
> would be so much better than what it could think of so it *never*
> wanted to eat up any bidding space if possible!

How did you arrive at 18%? Was it common sense, the result of testing,
or just something that seemed to work?

It seems that this percentage decides where the balance between
constructive and preemptive bidding should be. And that makes it very
important for the evaluation method.

Bert

[Bert Beentjes]

> It's a work in progress that I'm calling the "Shape System" for lack
> of a better name. It's a system that emphasises shape over strength.

I've developed this:

1C - 15+ HCP SB or SIX or 13+ XS
1D - 9+ HCP SV or 13+ XT
1H - 9-11 HCP SB
1S - 12-13 HCP SB
1NT - 14 HCP SB
2 in a suit - 9-14 HCP SIX
2NT and further 8-12 XS/XT

It works quite well in simulations. Even though sometimes there are
quite large casualties. At least it outperforms my more mundane
systems in the simulation setup.

SB = 4333 4432 5332 5422 6322 7222
SV = 4441 5440 5431 5521 5530
SIX= 6331 6421 6430 7321 7330
XS = 8+ or 74
XT = 6+/5+

It will probably evaluate badly with your method, since there are no
known anchor suits.

The advantage of the system is that by showing a general distribution
type early in the bidding, partner can quickly decide how to
(re)valuate his high card strength. Upgrade for notrumpish deals or
downgrade for distributional deals. The fact that there is no suit
information available is sometimes a problem, but also makes it more
difficult for opponents to counter effectively. There are no bids in
the enemy suit available.

Bert

[Tysen Streib]

> >1C = 17+ Total Points (with distribution)
> >1D = 10 HCP-16 TP, artificial - promises a 4cM
> >1M = 8 HCP-16 TP, 5+ cards, longer minor OK
> >1N = 11-15 balanced, no 4cM
> >2m = 8 HCP-16 TP, 6+ cards, no 4cM
> >2M = weak
> >2N = 8 HCP-16 TP, 54+ minors (either way), no 4cM
> >
> >
> Tysen
>
> I play a very similar system, eerily similar - couple of questions:
>
> 1) How do you define TP?

For this system just HCP + 321 for shortness

> 2) How do you handle 5m4m31 hands of 10-16TP?

2N. Did you mean one of the "m" to be a "M"? 5M4m opens 1M, 5m4M
opens 1D.

> 3) How do you handle major 8-16TP 2-suiters - 54 or 64 either
> way?

Open the longer major.

> 4) Do you roll all 5M332 10-16TP into 1N?

1N can't have a 4cM (or 5cM). 5M332 opens 1M, but you need 11 HCP to
open a balanced hand.

> 5) What are your NT ranges? ie 1C-1D-1N=16-19? How do you handle
> the 22-23 range taken by 2N weak?

I use a method slightly modified from Rigal's "Precision in the 90's"
where after 1C-1D, all balanced hands start with 1H. Responder has a
second negative of 1S, after which the NT ranges are slightly bigger
and not perfect because of the lack of a 2N opening bid. It has not
been bad at all. Our theory is it's better to guess between 1N and 3N
and use 2N invitational as rarely as possible.

>
> I can't tell how remarkably similar the system is including 11-15
> Notrumps without a 4card major. We include 5carders [weak only - with
> 14-15 we open 1M followed by 2N].
>

No 5cM for me.

[Tysen Streib]

> It's interesting to see that almost all systems (except Sandy's "never
> preempt" system) score worse than Standard American on constructive
> bidding.
> The rest of your post seems to indicate that a balanced choice of how to
> preempt is actually the most important in getting results.
>

I think it mostly has to do with light/more frequent openers. If you
pass more often, you can bid slower and it might help your
constructive bidding. However, letting the opponents in is the price.
It's all a balance. If the opponents hurt more than you do by
opening light then it's worth it.

> It would be interesting to see how specific bids score. The few scores
> I've seen seemed to put a very high penalty on ambiguity, where I think
> it may well be crucial in real life
> Examples: Polish club, or a strong 2C which can also be any weak
> varitation. My experience is with the latter, and I regard it as nearly
> essential to stop pre-empting over your strong openings.
>
> Now I realize that doing something specifically to stop the opps from
> doing something else is quite complex. Is any of that sort of reasoning
> included in your simulations?

Yes. Bids like this make it tougher on the opponents to choose
between bidding their own contract and trying to penalize you.
Sometimes they get it wrong. However, depending on the meanings of
the weak/strong option, if the opening bid is overcalled, responder
may not want to speak up with a moderate hand. Because of this,
opener often feels compelled to speak up with the strong type even if
partner passed. Sometimes this works sometimes it doesn't.

[Tysen Streib]

Mike Bell <mi...@bellfamily.org.uk> wrote in message news:<112cc6c...@RiscPC02.bellfamily.org.uk>...

> In message <a1bce703.04070...@posting.google.com>
> tys...@yahoo.com (Tysen Streib) wrote:
>
> > So if one believes the results of the simulation
> > absolutely (questionable!) then one could settle the argument of which
> > is better: strong or weak NT, 5cM or not, etc.
>
> My understanding of your methods isn't good enough to see what might
> cause the simulation to not be perfect. If it won't take too long,
> could you explain where might errors come from? You mentioned Matt
> Ginsberg's attempts to create a system using this method, I found his
> work on RGB but only as far as the 2C opener (which showed something
> like 23-24 points 3334 shape IIRC)! Do you know what his 1 level bids
> came out as?

There are a few assumptions that make these simulations a "model of
reality" rather than a reality itself. Things like right-siding, DD
play and defense, and the assumption that bidding can be modeled as an
information theory form of communication. I think Matt's method had
promise (which is why I've done all this work!) but I felt his assumed
effeciency was too high. The opening bids his system generated
illustrated this, wanting to pass about 80% of all hands! I know I've
seen the whole structure somewhere on rgb, but I can't find it
anymore. Hmmm...

>
> > I will say that in general, the simulations don't see that much
> > differnence between strong and weak NT (maybe that's why no one can
> > agree on this!) - it does say strong is *slightly* better. One thing
> > the simulations do say is that the weaker your NT becomes, the worse
> > it is to include a 5cM or even a 4cM. This makes sense from a theory
> > point of view in my eyes. The weaker the 1NT opening, the more likely
> > you miss a good major partscore.
>
> Very interesting. Is there a noticeable difference between allowing 4/5
> card heart suits and 4/5 card spade suits?
>

Haen't looked at that yet.

[Tysen Streib]

b.bee...@keesing.nl (Bert Beentjes) wrote in message news:<91691fb6.04070...@posting.google.com>...

> tys...@yahoo.com wrote in message
> > Part of the reason the follow ups are less than 100% is that the
> > opponents sometimes actually say something! So you'll never get to
> > 100%. On Ginsberg's original test of this method, he set the
> > effeciency to 23%. The system he generated as the best ended up
> > passing in 1st seat 80% of the time! It always assumed the follow ups
> > would be so much better than what it could think of so it *never*
> > wanted to eat up any bidding space if possible!
>
> How did you arrive at 18%? Was it common sense, the result of testing,
> or just something that seemed to work?

The effeciency you pick goes hand-in-hand with an "IMPs loss from
perfection." I did a small study of results to see if I could
approximate what this loss is in real life. It suggested about 18%.
I then tried that out testing some meanings and it seemed to be okay.
Ginsberg tried to double-check his number by saying that after he
generated his system, he would look at how the loss of his opening
bids compared to the assumed loss at the beginning. Unfortunately the
flaw in this is that it ignores the "dealer's advantage." We should
be better off than our opponents since we get in the first punch.

Hey, I have an idea. Does anyone havethe CPU time and access to GIB
or Jack that can run a long test for me? Have the computer
automatically bid and play a large number of boards, always having
North be the dealer. Let's keep the vulnerability set to none
throughout as well. Now replay the same boards always having East be
dealer. Score this up as an IMPs match so that one team always has
the dealer's advantage and the other doesn't. How much does that team
win by? I think that would be cool to see the results.