negative doubles(2)

[ggwhiz]

to focus a bidding system to win the highly competitive part score is a false dichotomy..lets get to the best games or best slams first at the cost of a losing partscore

 

What on earth does focusing on games/slams instead of everything have to do with playing good bridge?

 

A well rounded system addresses competitive partscore bidding too. I would surmise that the JEC match results swinging on game/slam hands is a bit of a red herring as all the participants are good enough in the partscore battles to cancel each other out. Not at all true in the stratified swiss or regional KO's.

[case_no_6]

With all of your example holdings, I think you have to Double. You are just too strong to Pass. If you do, and it goes Pass by LHO and partner dutifully reopens with a double, just what bid will you be able to make then?

 

For these reasons, there is a school of thought that handles this auction by bidding 1S with exactly 4 spades and Doubles otherwise. Double is still negative for takeout.

[jogs]

 

 

 

to focus a bidding system to win the highly competitive part score is a false dichotomy..lets get to the best games or best slams first at the cost of a losing partscore

 

 

Just scanned the first 9 rounds of play by USAI in the Bermuda Bowl. More than ten times one table bid slam while the other didn't. Sometimes making and other times going down. Only once did both tables bid slam. It had no play. Both tables went down.

 

Can only conclude that even the best players in the world have no idea how tricks are generated.

[Zelandakh]

Can only conclude that even the best players in the world have no idea how tricks are generated.

You have demonstrated time and again that you think you know more than them in this particular area.

[jogs]

You have demonstrated time and again that you think you know more than them in this particular area.

At least I think about it. While the experts are in fighting that their methods are better than all others without providing any proof.

[mike777]

again I use fourms vs jec as method of measure....

 

 

partscores do not mean we lose....

 

 

rest of your comments=close to zero

[mike777]

What on earth does focusing on games/slams instead of everything have to do with playing good bridge?

 

A well rounded system addresses competitive partscore bidding too. I would surmise that the JEC match results swinging on game/slam hands is a bit of a red herring as all the participants are good enough in the partscore battles to cancel each other out. Not at all true in the stratified swiss or regional KO's.

 

I ANSWER YOUR QUESTION....IN MY MEASURE.... A STANDARD OF MEASURE....YOU DO NOT.

 

iN FACT YOU DO NOT MENTION STANDARD OF MEASURE

 

yOU DO NOT MENTION PROBABILIY GIVEN STANDARD OF MEASURE

 

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

 

 

yOUR forecast may be better but at this point you do not show it.

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

 

 

see teblock/gardner

[jogs]

You have demonstrated time and again that you think you know more than them in this particular area.

Yes, I am opinionated. I have been studying effects on tricks for about ten years and thousands of hours.

Does anyone have any curiosity?

A five year old may ask, "Where do tricks come from?". How are tricks created? What conditions create tricks? How are tricks generated? An upper division statistic student would ask, "What is the moment generating function for tricks?" Is anyone investigating these questions. Has anyone attempted to identify the vectors which have the greatest effects on tricks. Uncover and quantify the effects of each vector. Recognize that the influence of each is dynamic, not static. Some vectors have more effect on lower level contracts than slams, while other have the reverse effect. Some vectors are independent and other vectors are interdependent. Vectors don't count tricks, they only estimate tricks, meaning there’s margin for error. Variance of the estimates is unavoidable. One should try to minimize the error.

High card points is a variable for estimating tricks. Bridge players need to clear their minds and think estimating partnership tricks instead of calculating adjusted point counts. The goal is a better method to estimate partnership tricks, not improve the point count.

 

Analysis of variance(ANOVA) is used to estimate tricks.

Tricks is a function of high card points.

Tricks is a function of trumps and high card points.

E(tricks) = trumps + (HCP-20)/3 + e

Each time an additional variable is added to the formula, our estimates improve.

E(tricks) = trumps + (HCP-20)/3 + SST + e

SST is an adjustment for the shorter holding of the partnership for each of the side suits.

E(tricks) = trumps + (HCP-20)/3 + SST + SF + e

SF is second suit fit.

E(tricks) = trumps + (HCP-20)/3 + SST + SF + C + e

C is for controls.

e is for the error. We want to minimize the variance of that error.

Hand types reside in multi-dimensional space. Bidding in a one dimensional space. Don't expect anyone to solve this bidding dilemma soon.

Dr. Bill Chen was able to change poker strategy by using game theory. I have been on a one man crusade to persuade bridge players to use ANOVA and partnership tricks to evaluate partnership assets. Alas I have received only negative feedback. Indeed it has been a lonely journey.

 

jogs

[Zelandakh]

The problems are simple - in all of the many hours of posts you have made on the subject you have never provided a shred of evidence that the formulae are accurate and certainly not more accurate than expert evaluation. In that time you have also not managed to give any practical value for the formulae even under the assumption that one or the other are an accurate reflection of reality.

 

Every bridge player would be interested in a practical evaluation method that made their bidding more accurate. But simply posting formulae and telling everyone that this is an improvement is pointless - you need to go away and compare the accuracy of your method against the alternatives given the information available from the bidding. If you can do this and show a significant advantage you might find more interest.

[helene_t]

I like the "standard" meaning of X is 4 spades and 1♠ is 5+. I think separating 4 from 5 is incredibly important.

Wouldn't it be better to play x as 5+ and 1♠ as 4? Less risk of wrongsiding the contract, and opener can accept the transfer to show doubleton support.

[mgoetze]

You have demonstrated time and again that you think you know more than them in this particular area.

"This particular area" = "Bridge"? ;)

[jogs]

The problems are simple - in all of the many hours of posts you have made on the subject you have never provided a shred of evidence that the formulae are accurate and certainly not more accurate than expert evaluation. In that time you have also not managed to give any practical value for the formulae even under the assumption that one or the other are an accurate reflection of reality.

 

Every bridge player would be interested in a practical evaluation method that made their bidding more accurate. But simply posting formulae and telling everyone that this is an improvement is pointless - you need to go away and compare the accuracy of your method against the alternatives given the information available from the bidding. If you can do this and show a significant advantage you might find more interest.

I have posted it on RGB.

Everyone accepts the correlation between high card points and

tricks. They also believe the correlation between trumps and

tricks. Since HCP and trumps are stochastically independent

the two are combined into a formula for tricks. These two

features are usually known by the 3rd or 4th bid of the

auction. Partners can exchange info on other features later

in the auction.

 

The formula is reposted

 

E(tricks) = trumps + (HCP-20)/3 + e

 

This is the formula for the general case. The sum of the HCP

is known, but not the exact location of the honors within the

hands. The e is for the error of the estimates. This is an

estimate of tricks. On specific boards the actual number of

tricks can be much more and much less.

Tricks on average equal trumps. This relationship breaks

down at 10 trumps. I have already posted a thread explaining

that the relationship between tricks and trumps is a

parabola.

13 trumps obviously can't average 13 tricks. It can be as

few as 7 tricks.

Let's start with a pure board. Both sides have 20 HCP. Pure

means all our points are in our two long suits.

AKxxxxx -- xxx QJx // QJxxxx -- xxx AKxx

13 trumps and only 10 tricks.

We have no points outside of trumps.

AKxxxxx -- xxx xxx // QJxxxx -- xxx xxxx

13 trumps and only 7 tricks.

 

E(tricks) = trumps + (HCP-20)/3 + e

 

When the expected tricks is greater than 10, one must inspect

the effects of other variables. There must be controls in

order to make slams.

Therefore this formula is a reliable guideline whenever the

our trumps is less or equal to ten and the expected tricks is

no greater than ten. Notice that our expected tricks

fluctuate wildly depending on the suit designated as trumps.

The standard deviation of the estimates is between 1 and 1.5

tricks/board. When the hands are flat, meaning no singletons

or voids in either partnership hand, the std dev drops to 1

to 1.25. Also flat hands reduce the number of expected

tricks. With a 5-4 fit and 20 HCP the expected tricks is 8

2/3. These std dev's are for the general case. On any

specific board the location of the honors are fixed and the

std dev goes down.

[jogs]

"This particular area" = "Bridge"? ;)

 

A much smaller space than the entire game. How to estimate tricks and how it affects judgment.

[Zelandakh]

I have posted it on RGB.

Everyone accepts the correlation between high card points and tricks.

What you post on RGB is neither here nor there. Let us start with your first assertion. Is this really true except for NT contracts? You will find plenty of agreement for a correlation between adjusted hcp + distrubtional values and tricks but, I suspect, little agreement for the direct link between hcp and tricks. I have plenty of issues with the rest but this seems a decent place to start with.