offense/defense ratio

[vsp...@hotmail.com]

Never heard of this concept til I read Robson/Segel. Seems like this is the most important bidding concept which is never mentioned. Can't remember ever seeing any authors referring to this concept in any articles on bidding.

jogs

[Emily Gumperz]

On May 21, 12:40 pm, vspo...@hotmail.com wrote:
> Never heard of this concept til I read Robson/Segel.  Seems like this is the most important bidding concept which is never mentioned.  Can't remember ever seeing any authors referring to this concept in any articles on bidding.
>
> jogs

Jeff Rubens did in Secrets of Winning Bridge and he may have got it
from Culbertson, since he used some other ideas that came from him
"Plastic Valuation" being one.

[Martin Ambuhl]

vsp...@hotmail.com wrote on Monday, May 21, 2012 03:40 PM
(rec.games.bridge):

Suppose you have 1/2 trick on offense and 0 tricks on defense. Your
offense/defense ration is infinite, and if able should bid over anything the
opponents might do. Robson/Segal love technical-sounding language without a
clue what the words mean.

[dak...@aol.com]

Suppose you have 1/2 trick on offense and 0 tricks on defense.  Your
offense/defense ration is infinite, and if able should bid over
anything the
opponents might do.  Robson/Segal love technical-sounding language
without a
clue what the words mean.

***
Using this logic 'ratios' do not exist as 1 / 0 in undefined infinite.
I did studt that in my math courses.

[Charles Brenner]

On May 22, 4:49 pm, Martin Ambuhl <mamb...@earthlink.net> wrote:
> vspo...@hotmail.com wrote on Monday, May 21, 2012 03:40 PM

It a team knockout event, it happened that we played a 3-way match
with one team going through, and the rule being that if all three
teams win one match, the winner is decided by "quotient." Can you
guess (or do you know) what "quotient" is referred to?

Charles

[Martin Ambuhl]

Not quite right. The "definition" of n/0 for n>0 follows from this:

For x>0 and y>0 or x<0 and y<0, the limit of x/y as t approaches 0 is
(positive) infinity.

When x and y are of opposite sign, the limit is negative infinity.

What is undefined as 0/0, for which _any_ value works (since z * 0 = 0 for
all z).

Now suppose you have 6 offensive tricks and 0 defensive tricks. A
Robson/Segal offense/defense ratio is, according to you, undefined. So I
suppose that means you throw up you hands in despair because you depend on a
concept which, according to you, gives no guidance at all.

[vsp...@hotmail.com]

This ratio is more of a concept or a way of
thinking of tricks. It isn't a rigorous math
approach. Certain patterns are better suited
for offense and queens and jacks in long suits
are better suited for offense, while queens and
jacks in short suits are better suited for defense.

[vsp...@hotmail.com]

On Monday, May 21, 2012 12:40:54 PM UTC-7, vsp...@hotmail.com wrote:
> Never heard of this concept til I read Robson/Segel. Seems like this is the most important bidding concept which is never mentioned. Can't remember ever seeing any authors referring to this concept in any articles on bidding.
>
> jogs

In point count an ace is assigned a value of 4. Many authors think 4 is too low and continually search for a more exact value of an ace. There is no exact fixed constant value for an ace. The valuation of an ace is a variable dependent on how it relates and interacts with the rest of the hand and rest of the board.
S A10972 H A852 D A32 C A
Those four aces aren't of equal value. Aces in longer suits are worth more. The ace of clubs may be worth only one trick. The ace of spades potentially creates many more spade tricks.
Queens and jacks are more valuable in long suits than short suits. In the secondary side suits queens and jacks are often worth no tricks for offense. They sometimes have more value for defense.
High point count is a random variable. Each of the components(aces, kings, queens, and jacks) are also random variables. The precise value is less important than the knowledge of the direction of the bias.
Hopefully I can tie this together. How variable values and ODR affects our bidding.

There isn't an actual ratio in a rigorous mathematical sense. Robson/Segel have taken literary license to use the term ratio loosely. This ratio is the relative offense/defense nature of the hand. There is no numerical value associated with this ratio.

[dak...@aol.com]

***
What is undefined is 0/0 UNTIL L'Hospital.
Continue obtusely refusing to recognize (just because it isn't
rigorous)
"offense/defense ratio" as a concept.

[Charles Brenner]

On May 23, 5:58 am, vspo...@hotmail.com wrote:
> On Tuesday, May 22, 2012 5:54:55 PM UTC-7, Charles Brenner wrote:
> > On May 22, 4:49 pm, Martin Ambuhl <mamb...@earthlink.net> wrote:
> > > vspo...@hotmail.com wrote on Monday, May 21, 2012 03:40 PM
> > > (rec.games.bridge):
>
> > > > Never heard of this concept til I read Robson/Segel.  Seems like this is
> > > > the most important bidding concept which is never mentioned.  Can't
> > > > remember ever seeing any authors referring to this concept in any articles
> > > > on bidding.
>
> > > Suppose you have 1/2 trick on offense and 0 tricks on defense.  Your
> > > offense/defense ration is infinite, and if able should bid over anything the
> > > opponents might do.  Robson/Segal love technical-sounding language without a
> > > clue what the words mean.
>
> > It a team knockout event, it happened that we played a 3-way match
> > with one team going through, and the rule being that if all three
> > teams win one match, the winner is decided by "quotient." Can you
> > guess (or do you know) what "quotient" is referred to?
>
> > Charles
>
> This ratio is more of a concept or a way of
> thinking of tricks.  It isn't a rigorous math
> approach.

I don't really see how that is a reply to what I said, but you're
making the case that it's senseless pedantry to be concerned about
misuse of words, right? The Humpty-Dumpty philosophy.

Charles

[Charles Brenner]

I think Martin is suggesting that you're confusing 0/0 versus n/0.

By the way l'Hopital bought "l'Hopital's rule" from Bernoulli.

Charles

[vsp...@hotmail.com]

It is on page 23. I've reread it many times.
There is no math type ratio. It is not a misuse
of words. It is taking liberties with words.
Doesn't matter. Robson/Segel has shown us an
other way to view bidding.
Some hands are better suited for offense. Other
hand are better suited for defense. And many
hands are better suited for passing.

---
New path for thinking.

Pattern: 4333, 4432, 5332, etc.

Some authors refer to this as distribution. Others use shape. I will use pattern. There are 39 combinations of patterns. Each of these patterns have a offense/defense type features. Skewed patterns tend to be better for offense. Flat patterns tend to be better for defense. Sometimes flat patterns are good for neither. Just bid less with them and stay out of trouble.
Thinking in ODR terms will assists players in contested auctions. When should one bid 3 over 2? When should one bid 3 over 3? Cohen/Bergen says bid to the level of your trumps. ODR suggests with flat patterns produce fewer tricks than total trumps. All patterns without a singleton or void are flat.

Pattern pair.

This is the patterns of the partnership. When patterns mesh it tends to be correct to bid more. 5=4=3=1 facing 4=5=1=3. When patterns clash it tends to be correct to bid less. 5=4=2=2 facing 2=2=4=5. Flat patterns suggest bidding less. 5=3=3=2 facing 3=3=3=4.

If I have not lost the interest of the audience. I will post how this
changes the way certain conventions should be played(according to me).

[Adam Beneschan]

On Wednesday, May 23, 2012 1:46:14 PM UTC-7, Charles Brenner wrote:

> > This ratio is more of a concept or a way of
> > thinking of tricks.  It isn't a rigorous math
> > approach.
>
> I don't really see how that is a reply to what I said, but you're
> making the case that it's senseless pedantry to be concerned about
> misuse of words, right? The Humpty-Dumpty philosophy.

Actually, here's what m-w.com says about "ratio":

1a : the indicated quotient of two mathematical expressions

b : the relationship in quantity, amount, or size between two or more things : PROPORTION

The first entry is clearly about dividing numbers, but the second definition is a lot more fuzzy and could refer to any relationship (bigger than, how much bigger, etc.) that isn't strictly a quotient. (The entry for PROPORTION has a similarly fuzzy definition.) So you could say that R-S aren't really misusing the term, although I rarely see it used in this kind of fuzzy sense. Of course, M-W is an American English dictionary and R-S aren't Americans, so maybe this doesn't apply.

-- Adam

[Charles Brenner]

As an aside Adam, I don't agree that the second definition is a lot
more fuzzy. In view of the indication that it means "proportion" it is
at most, if at all, slightly more fuzzy. A proportion is the ratio of
two things though I agree we use the word even for things that aren't
numerically quantifiable ("a reaction that was out of proportion to
the insult"). Notwithstanding what Meriam claims Webster thought
though, I've never before now heard "ratio" employed so loosely.

That's an aside because, though inspired by a complaint about the word
"ratio", my comment was about the word "quotient" and I (perhaps
stubbornly) stuck to that idea in the face of vspo's apparent non
sequitur. The answer to my riddle about deciding a bridge winner by
"quotient" turned out to be that "quotient" in fact meant
"difference." I suppose we all, to some extent, accept the principle
of using words figuratively and bending their meanings, but differ in
our tastes as to how far to go in specific cases. Substituting a wrong
word for an obvious right choice that is not more obscure or technical
falls, in my view, in a different realm, the realm of the ludicrous,
sponsored by the same mentality that thinks saying "million" for
"billion" is inconsequential.

Charles

[Mark Brader]

Charles Brenner (copyedited):
>>>> In a team knockout event, it happened that we played a 3-way match

>>>> with one team going through, and the rule being that if all three
>>>> teams win one match, the winner is decided by "quotient." Can you
>>>> guess (or do you know) what "quotient" is referred to?

It's the quotient of the IMPs scored by and against each team.
If A beats B by 15 IMPs to 9, and B beats C 6-5, and C beats A
20-10, then the quotients are computed:

A (15+10)/(9+20) = 0.86
B (9+6)/(5+15) = 0.75
C (5+20)/(6+10) = 1.56

so C survives; if there were two survivors, A would also survive.

> The answer to my riddle about deciding a bridge winner by
> "quotient" turned out to be that "quotient" in fact meant
> "difference."

There was no riddle. IMP difference is *another* way to score
3-way matches when each team wins. It usually produces the same
outcome as the quotient, and is easier to compute mentally:

A (15+10)-(9+20) = -4
B (9+6)-(5+15) = -5
C (5+20)-(6+10) = +9

Of course, this has nothing to do with "offense/defense ratios".
--
Mark Brader "I wasn't the one who misplaced the entire
Toronto Deltivid asteroid belt!"
m...@vex.net "Deja Q", ST:TNG, Richard Danus

My text in this article is in the public domain.

[Lorne]

"Martin Ambuhl" wrote in message
news:8u-dnedn_5qcuyHS...@earthlink.com...

...........

It is defined on page 28 of the book (p23 if you have the electronic
version) but not in the strict mathematical way you are imposing on it.
They are comparing the size of offensive potential to the size of defensive
potential and saying high ODR means offense greater than defense and low ODR
is defense greater than offense. ie subtracting one from the other rather
than dividing. It may not be the correct mathematical term but no other
term seems to be better at describing the idea.

[Charles Brenner]

On May 24, 12:41 am, m...@vex.net (Mark Brader) wrote:
> Charles Brenner (copyedited):
>
> >>>> In a team knockout event, it happened that we played a 3-way match
> >>>> with one team going through, and the rule being that if all three
> >>>> teams win one match, the winner is decided by "quotient." Can you
> >>>> guess (or do you know) what "quotient" is referred to?
>
> It's the quotient of the IMPs scored by and against each team.
> If A beats B by 15 IMPs to 9, and B beats C 6-5, and C beats A
> 20-10, then the quotients are computed:
>
>     A  (15+10)/(9+20) = 0.86
>     B  (9+6)/(5+15)   = 0.75
>     C  (5+20)/(6+10)  = 1.56
>
> so C survives; if there were two survivors, A would also survive.
>
> > The answer to my riddle about deciding a bridge winner by
> > "quotient" turned out to be that "quotient" in fact meant
> > "difference."
>
> There was no riddle.  IMP difference is *another* way to score
> 3-way matches when each team wins.  It usually produces the same
> outcome as the quotient, and is easier to compute mentally:

It was a riddle in that one of my team members of the day used the
word "quotient" but when I asked what it meant he (rather snootily)
explained the method which you have more reasonably called
"difference." But I'm relieved to hear that quotient can also mean
quotient.

Charles

[KWSchneider]

On May 21, 3:40 pm, vspo...@hotmail.com wrote:
> Never heard of this concept til I read Robson/Segel.  Seems like this is the most important bidding concept which is never mentioned.  Can't remember ever seeing any authors referring to this concept in any articles on bidding.
>
> jogs

Ratio is not useful. Trick differential is. I've been pushing offense/
defense differential for 2 years now with my PUSH concept [renamed
"Unified Point Count"]. And as you know [although you disagree], I
have developed a method to calculate it.

Kurt

[KWSchneider]

On May 23, 9:28 am, vspo...@hotmail.com wrote:
> On Monday, May 21, 2012 12:40:54 PM UTC-7, vsp...@hotmail.com wrote:
> > Never heard of this concept til I read Robson/Segel.  Seems like this is the most important bidding concept which is never mentioned.  Can't remember ever seeing any authors referring to this concept in any articles on bidding.
>
> > jogs
>
> In point count an ace is assigned a value of 4.  Many authors think 4 is too low and continually search for a more exact value of an ace.  There is no exact fixed constant value for an ace.  The valuation of an ace is a variable dependent on how it relates and interacts with the rest of the hand and rest of the board.

The point value of an Ace is mainly a function of your tricks/points
ratio. If you assume, for instance, that it takes, on average, a
combined 25 points to make 10 tricks, then a trick is 2.5 points.
Hence 4 points represents 1.6 tricks. My work has proven [at least to
me] that 4 points is about right for a TRUMP Ace, 3 Points for a non-
trump Ace, and 2 Points for an Ace in a suit that you and you partner
have 9 or more points - and you are defending. This, of course,
presumes a 25 point count for a 10 trick game.

And before you go into statistical epilepsy, I like to use the term
"potential". Nothing is exact, these are relative values ON AVERAGE.

> S A10972  H A852  D A32  C A
> Those four aces aren't of equal value.

Agreed.

> Aces in longer suits are worth more.

This is BS in trump contracts - non-trump Aces diminish in value as
the suit gets longer [or partner has shown length]. In your example,
assuming that Spades are trump, the SA is worth 4points [1.6 tricks],
the HA 3points [1.2 tricks] unless your partner has bid the suit. If
you are declaring, then the HA is worth full value, if you are
defending then the HA is only worth 2 points [0.8 tricks - it can be
ruffed potentially, since you have 9plus hearts between you].

> The ace of clubs may be worth only one trick.

More than 1 trick, since it potentially will be hidden in declarer's
hand and third hand will feel the need to play high.

> The ace of spades potentially creates many more spade tricks.

If trump...

> Queens and jacks are more valuable in long suits than short suits.  In the secondary side suits queens and jacks are often worth no tricks for offense.  They sometimes have more value for defense.

Q and J are essentially worthless in suits that are neither trump or
part of a double fit.

> High point count is a random variable.  Each of the components(aces, kings, queens, and jacks) are also random variables.  The precise value is less important than the knowledge of the direction of the bias.

Again this is BS. They are not random - an Ace is worth more a King,
etc. I have no idea what your "point" is here.


Let me give you an example of ODD [offensive-defensive differential].
Neither side vulnerable, Partner opens 3S, RHO doubles. You have xxxx
xx Axxx xxx, What is your plan? And on what basis do you decide? Using
LoTT, you would probably decide that the total trumps [=tricks] is 21
[11+10] and decide to pre-emptively sacrifice over 6H with 6S.

However, if you could calculate the offense and defense [and hence the
differential] of your combined assets, you could predict the total
tricks since TT = ODD + 13 tricks. In this case, Unified Point Count
tells us that our combined ODD of 6 tricks [partner has 6 offense and
0 defense, while you have 1 offense and 1 defense] yields our exact
expectations - 1 defensive trick and 7 offensive tricks, and 19 total
tricks. So they are likely to make 6H and if they do, we can only make
1S - NOT a profitable sacrifice.

Kurt

[Douglas Newlands]

if you have no offensive tricks, you are not bidding.
If you have no defensive tricks you are not doubling.
So treat off/def as a calculation only when off neq 0
and def neq 0

0/0 => do nothing
off/0 => consider bidding more
0/def => consider doubling

off/def > 0 =>consider bidding to be better than doubling

off/def < 0 => consider doubling to be better than bidding

This seems to be just bridge oriented rule for dealing
with zeroes rather than the standard mathematical one.

doug

[vsp...@hotmail.com]

> Kurt

http://en.wikipedia.org/wiki/Random_variable

Kurt, your stat background is weak.
Random variable is a stat term. Does not
mean an ace can be worth less than a king.
Only means the value of an ace is not fixed.

[vsp...@hotmail.com]

Support doubles revisited. Also only applies to majors.

1) 1C - p - 1H - 1S, X.
2) 1C - p - 1H - 1S, 2H

When playing support doubles, this double shows three card heart support. The direct raise shows four card heart support.

Maybe this is not the best way to differentiate the two calls.
ODR. Offense/defense ratio. It's best for us to think in terms of ODR. The direct raise suggests we want partner to

compete to 3H with five hearts and a minimum. The double suggests less offense. With five hearts responder may wish to

back off and allow opponents to declare 2S. The direct raise is still usually four. The double is still usually three. Just not

always.

a) S 5 H K76 D KQ93 C A9753
b) S QJ5 H Q765 D K93 C A97

With a) raise with 3 we probably wish to compete against 2S. With b) double we don't wish to disturb 2S. These two calls

distinguish between the offensive nature of the hands.

c) S QJ5 H Q76 D K93 C A973

Pass. With this bidding this is no longer an opening hand.

3) 1C - p - 1H - 2D, X.
4) 1C - p - 1H - 2D, 2H.

d) S KQ9 H K976 D 964 C A95
e) S KQ95 H K76 D 4 C A9754
f) S KQ95 H K76 D 94 C A975
g) S KQ95 H K976 D 94 C A95

Bid 2H with e) and g). Double with d) and f). The support double is used to assist partner in competing onto the three level.

It is not used to tell partner the number of hearts in your hand. Raise with offensive oriented support. Raise with hands which have a high worth with partner's suit as trumps. Double with flattish support or general strength. After a raise partner should compete three over three with five trumps. After a double partner should pass 3/3 with only four trumps. That doesn't mean these actions will always be right. It does mean we are playing the percentages.

[Chris xxxxx]

I believe the ACBL used to use quotient in the situation you describe
and changed to net IMPs a few years ago. Some folks still call the
tie-break "quotient" either because they don't know the rule changed
or because it had become a fixed phrase for them.

Christopher Monsour

[vsp...@hotmail.com]

I'm suggesting it should be c). It was never a ratio.
It is more about offense/defense relativity. Some hands
are better suited for offense, while others are better
suited for defense.

[Mark Brader]

Chris Monsour:

> I believe the ACBL used to use quotient in the situation you describe
> and changed to net IMPs a few years ago.

Thanks, Chris. I'd always assumed that the choice was up to the body
running the event, but it is in fact covered in the ACBL Conditions of
Contest document. The current one

http://web2.acbl.org/documentlibrary/play/coc/koEvents.pdf

is dated 2005 and requires the use of net IMPs (difference between IMPs
won and IMPs lost), on page 4.

But I found an old version, dated 1999, on one of the districts' web sites,

http://www.d16acbl.org/COC/COC_KO.htm

and it requires IMP quotient.
--
Mark Brader "A moment's thought would have shown him,
Toronto but a moment is a long time and thought
m...@vex.net is a painful process." --A.E. Housman (alleged)

[vsp...@hotmail.com]

"What it boils down to is this: we must consider our offence to defence ratio (henceforth abbreviated as ODR)
when choosing what level to raise partner to. The higher that ratio is (the more offence we have relative to defence),
the higher we should preempt."

Here's a quote from page 23. It was always offense
relative to defense. There was never a math defined
ratio.

[KWSchneider]

I have a degree in Engineering [albeit 35 years ago] so I would argue
with you on that.

I'm not familiar with the term "random" in that sense. I grew up with
the term "stochastic". I agree that each variable is stochastic, but
the value of the mean [and the other statistical descriptors, like
moment/skewness/kurtosis] are definable and have ranges. And the mean
value of the Ace is greater than that of a King.

For example, using the analogy of dice. Getting a seven is not
predictable - you could get zero of these in 10 rolls. But it is
statistically more likely than getting a five which is more likely
than a three...

Although they are stochastic variables [random, as you call it], you
can calculate the probability of a seven, a five, and a three. Just
like you can calculate the "probable value" of an Ace, a King, etc.

Kurt

[Steve Willner]

On 2012-05-23 8:58 AM, vsp...@hotmail.com wrote:
> This ratio is more of a concept or a way of
> thinking of tricks. It isn't a rigorous math
> approach.

We can see that. A better word would have been 'difference' or Kurt's
'differential'. Then it would be pretty close to rigorous, though
impossible to evaluate perfectly. Offense-defense difference is not the
only parameter one needs for a bidding a decision, but it's an important
one.

Using ratio instead of difference is the same as Charles' example of
using quotient instead of difference. Using words that are related but
not exact is probably OK in informal speech, but it was probably a bad
idea in a serious book.

--
Help keep our newsgroup healthy; please don't feed the trolls.
Steve Willner Phone 617-495-7123 swil...@nhcc.net
Cambridge, MA 02138 USA

[vsp...@hotmail.com]

The definition of a random variable hasn't
changed in the last fifty years. The public
is now just more aware of statistics.
Statisticians calculate the mean value of
a statistic and its variance. Clearly the
mean value of an ace is highly than that of
a king.

[Fred.]

Without quibbling over usage, it is a term which, unfortunately,
misleads people into, at least momentarily, looking for the
numerator and denominator of the alleged ratio. Having said that,
I don't have a better term for the same concept in my pocket and,
so, will continue to use it.

Fred.

[Fred.]

I don't have access to the OED at the moment, but I would be shocked
if

I don't have access to the OED at the moment, but would be shocked
if it failed to include a use at least as nebulous as 1b.

Nevertheless, I think the use of "ratio" sends a lot of people
looking for a number or two, and for that reason is less than
ideal. I try to slur the letters a little bit when I say "ODR"
because it helps me avoid wincing. :-)

Fred.

[vsp...@hotmail.com]

Replace the word ratio with relativity.
Offense/defense relativity.
No more division by zero.
On page 23 Robson/Segel does mention
offense relative to defense.

[Fred.]

ODR is the term which is generally understood
and, therefore, best to use. My dislike of
it doesn't change that.

Fred.