[gwnn]

VM1973, on 2011-August-19, 18:14, said:

Well I think that's a strange argument to make considering that the first hand has a reasonable shot at slam opposite the posted hand while the 2nd hand might not even take 10 tricks.

Don't get me wrong... I like having 10s around while playing 3NT just as much as anyone else. I just don't see them being that helpful when I'm trying to make 6NT.

I think you developed a healthy habit of not reading all the lines of my posts. It is probably good for your time management, but makes communication less efficient. Like I said, controls are not bad for suit contracts, but 4321 are much better for NT. And when I get a 5332, many times we will be in NT.

It is also a little bit funny that you responded in 1 line about the king that you counted twice - your bad. That was a 4 point difference in Zars!! You wanted to prove that AKxxx Axx xx xxx is not nearly a minimum opener, and were overjoyed when your favourite evaluation method confirmed your view. You weren't even a little suspicious of your findings, 32, doesn't that sound a little too much? No no, that's what Zars told me! So 90% of your previous post was about the arbitrary 1 zar point you included for concentration, but you forgot about your whole point, that AKxxx Axx xx xxx has a lot of extras (1.5 kings!). So we can agree to disagree on Zar and controls, I don't want to argue about it. If you want to use Zars all the time, you're welcome to do so. However, it is very inaccurate for no trump.

[VM1973]

gwnn, on 2011-August-20, 03:32, said:

I think you developed a healthy habit of not reading all the lines of my posts. It is probably good for your time management, but makes communication less efficient. Like I said, controls are not bad for suit contracts, but 4321 are much better for NT. And when I get a 5332, many times we will be in NT.

It is also a little bit funny that you responded in 1 line about the king that you counted twice - your bad. That was a 4 point difference in Zars!! You wanted to prove that AKxxx Axx xx xxx is not nearly a minimum opener, and were overjoyed when your favourite evaluation method confirmed your view. You weren't even a little suspicious of your findings, 32, doesn't that sound a little too much? No no, that's what Zars told me! So 90% of your previous post was about the arbitrary 1 zar point you included for concentration, but you forgot about your whole point, that AKxxx Axx xx xxx has a lot of extras (1.5 kings!). So we can agree to disagree on Zar and controls, I don't want to argue about it. If you want to use Zars all the time, you're welcome to do so. However, it is very inaccurate for no trump.

No, actually, I've just discovered them and although they seem to work extremely well I am having problems applying them. For example, we all "know" that if your partner makes a negative double and you have 10+ that you're supposed to jump... but what's that in ZPs? Again, the takeout doubler shouldn't bid again unless he has extras... let's say 17+ considering that his partner should have jumped if he had 10+ so what's that in ZPs? Assuming that you want to play ZPs for NT then how much is 15-17 in ZPs?

As for thinking that AKxxx Axx xx xxx that I added up and came up to a number that was bad and I should have known that was too much... that's what comes from being unfamiliar with something. Previously I would never have thought of opening holding:
Axxxx Axxx xxx x as I always used the rule of 20 (8+9=17 ... not even close). Nevertheless ZPs says it's an opener (12 for aces, 13 for shape, 1 for concentration...and even if you don't count concentration you can still add 1 for having the spade suit).

Anyway I read an article written by the author of Binky points in which he basically admitted that ZPs were statistically within 1 trick of the actual trick taking power of the hands (using double dummy play) but that his system was within 0.79 (statistically better) so Binky points are definitely on my reading list, but until then I'm experimenting with ZPs but I feel encouraged that even the detractors of the system (Binky adherents) have to admit that it does work.

Still, I can't help but think that YOUR posts are the ones lacking. You said ZPs overvalue controls. As soon as you said that I opened a second tab and Googled it to try to confirm what you said and I came up with nothing. How do you know ZPs overvalue controls? A deep feeling in the pit of your gut or can you actually link me to a webpage or post that indicates how, when, and why they fail under certain circumstances?

Additionally, I can't help but think that you're the one who isn't reading my posts. If you read back you'll see that my original post was: "I really don't think slam is there unless the opener has extras and I think that just asking for aces is a bad idea because you can always set the spade suit as trump first before asking and find out about suit quality." From there we're off on English teachers living in Peru often can't do math even with a calculator in their hand. Well, if I were good at math, I would be an engineer not an English teacher.

Finally, even if 5-3-3-2 often plays in NT I'm thinking it's not going to play in NT opposite the hand in question... just a theory of mine but you're welcome to play it in NT as often as you'd like.

[VM1973]

MrAce, on 2011-August-20, 00:17, said:

You obviously haven't played with Haspel as partner. When he puts you in a slam, you bet you will need every single T and 9 out there

I'm certain Haspel cannot be any worse than the normal partners one gets when they go on this website and sit down with a random partner.

[gwnn]

I wasn't posting about this particular hand, I was posting about Zar points in general and the Zar evaluation of AKxxx Axx xx xxx in particular. That is why I only replied to your second post, which I disagreed with and not your first post, that I agreed with. I agree with you that 4NT is too much on this hand.

Quote

How do you know ZPs overvalue controls? A deep feeling in the pit of your gut or can you actually link me to a webpage or post that indicates how, when, and why they fail under certain circumstances?

I know that Zar points overvalue controls because in NT 4321 is very close to reality and the most common game in bridge is 3NT. This is the third time I wrote this down, this is why I said Zar points overvalue controls. Let me write this down a fourth time: Zar points overvalue controls because controls are not as valuable in NT as in suit contracts and you do not know when you open 1♠ that you will play 4♠.

[Zelandakh]

xxhong, on 2011-August-19, 15:04, said:

I think it's simple to just play 1S 3NT as RKC to solve this problem. The 4-3-3-3 shape can usually be bid in a slower way. 3NT as RKC may land you at many 4 Ms when you miss two KC, which is very valuable IMO.

I agree with you when comparing RKCB with a natural 3NT response. However, whether RKCB is as useful here as, for example, a void splinter is another question entirely. If you are going to look at conventional responses then you need to compare with other possible conventions and not only against "natural".

[whereagles]

lol.. ZAR wars.

@VM: my issue with 2NT is that unless you have your own souped-up version of the convention, you're going to get absolutely nothing useful out of it

[ArtK78]

VM1973, on 2011-August-20, 07:06, said:

Previously I would never have thought of opening holding:
Axxxx Axxx xxx x as I always used the rule of 20 (8+9=17 ... not even close). Nevertheless ZPs says it's an opener (12 for aces, 13 for shape, 1 for concentration...and even if you don't count concentration you can still add 1 for having the spade suit).

I want to preface this post that I know next to nothing about ZAR points.

However, if it is true that ZAR points evaluate Axxxx Axxx xxx x as an opening one bid then that is all the argument that I need for the statement that ZAR points overvalue controls.

[Zelandakh]

VM, Zars advocates opening very light so Rule of 19 is closer than Rule of 20. Secondly, if you are convinced of the 3-2-1 ratio then the equivalent in Milton is 4.5 - 3 - 1.5. That makes Axxxx/Axxx/xxx/x 1 point short of an opening, as indeed it is in Zars until you adjust for spades. I think if you think in this way wrt Milton it will make your comparisons with Zars better/simpler. What is more, this "adjusted Milton" method is generaly better at handling NT hands than pure Zar points. My understanding is that the few top pairs that use Zar points do so in combination with other evaluation straegies and not the pure evaluation scheme from the website.

[JLOGIC]

FrancesHinden, on 2011-August-18, 13:20, said:

In the last 6 months I've had two hands that responded 4NT to 1M to ask for aces (one of them was written up in the Daily Telegraph later, we gained a big swing on it for slightly complicated reasons).
I've had no hand where the auction started 1M P and I wanted to ask for keycards directly.

No sarcasm intended in this question, are you trying to argue that straight ace ask hands are more common than keycard hands after 1M p, or are you just saying it as an interesting side note?

[jdeegan]

If I have a laydown slam opposite the right minimum, then I should make a strong jump shift. So wtf is the problem?

[FrancesHinden]

JLOGIC, on 2011-August-21, 00:01, said:

No sarcasm intended in this question, are you trying to argue that straight ace ask hands are more common than keycard hands after 1M p, or are you just saying it as an interesting side note?

It was an interesting side note. Both hand types are (IMO) so rare that I have no useful frequency analysis.
Hands where opener wants to keycard at his second bid are not so uncommon, but hands that want to keycard directly in response to a 1M opener are very seldom seen. Particularly if you disagree with doing it on a hand like this.

[VM1973]

gwnn, on 2011-August-20, 07:35, said:

I know that Zar points overvalue controls because in NT 4321 is very close to reality and the most common game in bridge is 3NT. This is the third time I wrote this down, this is why I said Zar points overvalue controls. Let me write this down a fourth time: Zar points overvalue controls because controls are not as valuable in NT as in suit contracts and you do not know when you open 1♠ that you will play 4♠.

As I asked before, can you link me to something supporting your position? I guess the answer is no. So I link you to http://groups.google...fe90fb84687c5e9

According to this link the average error rating of systems are:
HCPs: 1.23
Bergen: 1.08
HCPs plus 3-2-1 distribution: 1.07
ZP: 1.05
BUM RAP +321: 1.03
Binky Points: 0.99

Accordingly I conclude that the simple 4-3-2-1 system is bad, ZPs are superior and Evolved Binky Points may be the best of all, though I'm not sure yet as I still haven't found a coherent description of how the system works.

[gwnn]

Like I said, helene_t made a logistic regression analysis some time before, but I guess you skipped through that part of my post. This is for suit contracts: http://www.bridgebas...s-investigated/

and this is the post on 3NT: http://www.bridgebas...dpost__p__94283

but VM, this is all slightly futile. Bridge will never be completely scientific, and no evaluation method will ever be very accurate. And what I conclude about your approach to the game from labelling AKxxx Axx xx xxx 1.5 kings above minimum is that you're willing, or striving, to suspend all your judgement in favour of the Zar point results. Similarly with accepting Axxxx Axxx x xxx without a problem. Your judgement in bridge is an asset, not a cancer. You will always run into problems where you need your judgement, for example if you sohuld double them or bid on or pass, etc.

Anyway, since when are you interested in scientific formulae for the game, I thought every hand was different and inductive reasoning was faulty?

This post has been edited by gwnn: 2011-August-23, 01:39

[Zelandakh]

gwnn, go easy! Most bridge players who are interested in innovative stuff go through phases where they think they have found the new way. Whether it be Zar points, forcing pass systems, combine carding, whatever. This is a natural part of the development for this type of bridge player. I am sure VM will move on after a while to develop other aspects of his game. In the meantime the lessons he learns with ZP will probably be beneficial, if for no other reason than seeing that yes, distribution is very important, but not the be all and end all that ZP would suggest. But nonetheless much more than simple 3/2/1 with a fit.

ZP are fairly close to 4.5 - 3 - 1.5 - 1 adjusted Milton with 5 - 3 - 1 for distribution. Hence they work extremely well when there is a fit. Unfortunately, when we have to play in NT they stop working. In most bidding systems it is easier to add unexpected extras than to suddenly say you have less than before. This is the inherent advantage of the above-mentioned adjusted Milton approach over ZP. You reach the same evaluation but you do it without overvaluing the misfits. You also do not reduce your credibility by saying that AKxxx Axx xx xxx is 1.5 kings above minimum. Perhaps it is close when partner has a good major suit fit, when they do not you need to re-evaluate.

While doing the re-evaluation think a little about what gwnn said VM. The point made here is extremely good. Use ZP by all means, always making sure to analyse the pros and cons later against alternative evaluation strategies so as to see the strengths and weaknesses of each. But above all work on your bridge judgement. This will pay dividends back to you irrespective of which evaluation system you eventually use. Listening closely to what some of the most experienced posters on BBF say can help but mostly this will come from experience. You cannot always rely on an evaluation system or generic rule for such decisions!

[VM1973]

gwnn, on 2011-August-23, 01:30, said:

Like I said, helene_t made a logistic regression analysis some time before, but I guess you skipped through that part of my post. This is for suit contracts: http://www.bridgebas...s-investigated/

and this is the post on 3NT: http://www.bridgebas...dpost__p__94283

but VM, this is all slightly futile. Bridge will never be completely scientific, and no evaluation method will ever be very accurate. And what I conclude about your approach to the game from labelling AKxxx Axx xx xxx 1.5 kings above minimum is that you're willing, or striving, to suspend all your judgement in favour of the Zar point results. Similarly with accepting Axxxx Axxx x xxx without a problem. Your judgement in bridge is an asset, not a cancer. You will always run into problems where you need your judgement, for example if you sohuld double them or bid on or pass, etc.

Anyway, since when are you interested in scientific formulae for the game, I thought every hand was different and inductive reasoning was faulty?

You're right that bridge will never be scientific and considering that science is empirical and inductive that is a good thing. If you understood much about scientific philosophy, you'd realize that what I propose is falsificationism, which is an invention of Karl Popper in which theories are advanced and attempts are made to falsify them. As such, this method is logically consistent and avoids all the problems that plague induction.

First let me propose that we examine these two hands:

♠Kx
♥Kxx
♦Kxx
♣AJ10xx

♠QJxxxx
♥xx
♦AJ10x
♣Qx

Let us suppose that due to an auction best forgotten South is declaring 4♠ and the defense starts with three hearts, the 3rd ruffed. East wins the ♠A and returns a spade and you're left to make the rest of the tricks. How do you proceed? The percentage play for picking up the rest of the tricks is to cash ♦AK catering for the possibility that the ♦Q will drop and fall back on the club finesse. HOWEVER if you are using double dummy declarer play the declarer will magically know the right way to finesse against ♦Q. Needless to say this is patently unfair.

Now this doesn't mean that such evaluations are completely useless - rather that you must be careful how you apply them. I don't personally think that you can look at such a database and say "A 10 should be worth 0.4" because 10s are worth a lot more when you can look at your opponents' hands and know if you should hook the 10. Accordingly an 0.4 would be the maximum amount a 10 would be worth and its real value would be lower.

Additionally, let's look at this perfect hand:

♠AKQJ
♥AKQ
♦AKQ
♣AKQ

37 HCPs = 13 tricks so a trick is, on average, 2.846 HCPs. Even if we generously assume that 10s are worth 0.4 points that still means you need more than 7 tens to make a one trick difference... and there are only four 10s in the pack.

I also couldn't help noticing that the post said: "Interestingly, queens got a weight of 0.8 rather than the traditional 1.0...." In what system do queens traditionally have a weight of 1.0? Is he really implying that a queen is worth only twice as much as a ten? Or did he misspeak and mean the Jack? And if that's true, why did he continue, "It could be argued that DD-simulation understimates the values of queens because declarer often has to guess how to catch the queen: Kings can only be finesed in one way (except for sec Kings, throw-ins etc)."

[gwnn]

Yes I suppose she meant 1.6 instead of 2. I am not sure, but she is around, she can tell us.

Looking at AKQJ AKQ AKQ AKQ will not give you an accurate picture of how many HCPs are worth how many tricks. For example, it would mean that you need 28.5 hcp's for 10 tricks, a conclusions that nobody accepts. The reason why AKQJ AKQ AKQ AKQ gives you a false picture is twofold: first of all it is a hand that offers you certainty for 13 tricks, second of all because in real contracts you often take tricks with jacks, tens, or even twos, because of length.

I know about the pitfalls of double dummy evaluations, you do not need to give an example hand to tell me about it. However, you cannot dismiss such investigations altogether. Declarer plays much better when he sees all four hands, but defenders also make much better opening leads. There have been other investigations (sorry, no links) that came to the conclusion that double dummy results give better results than reality for declarer at grand and small slam level and worse results at smaller contracts, and that somewhere around the 3NT/4M level double dummy results actually come pretty close to reality. I will not look for this link, it is somewhere here on the forums, I think it was a bridge browser study. I suppose you can disregard this as my anecdotal non-evidence, but at least answer me this: how is it that Helene's numbers give much higher importance to controls in suit contracts and the coefficients are actually very close to 4321 for no trump? How can you explain that purely on double dummy errors? Surely if you want to talk only about double dummy errors favouring declarer, they favour declarer in NT and suit contracts as well?

Anyway, it is funny that you link to Tysen's investigations, where Zars perform only slightly better than HCP+321 and in his own words

Tysen said:

I'm surprised at Zar's poor performance.

Surely you don't mean that people who use HCP's never think about their shortnesses? That we, the obsolete HCP addicts will reject invitations after, say

1♠-1NT
2♥-3♥

with all 11 counts, whether 5422 or 6430?

This post has been edited by gwnn: 2011-August-23, 07:57

[mikeh]

VM1973, on 2011-August-23, 07:36, said:

You're right that bridge will never be scientific and considering that science is empirical and inductive that is a good thing. If you understood much about scientific philosophy, you'd realize that what I propose is falsificationism, which is an invention of Karl Popper in which theories are advanced and attempts are made to falsify them. As such, this method is logically consistent and avoids all the problems that plague induction.

VM: why do you write as if the person you are addressing is ignorant of such ideas as falsification? Which, as you presumably know, is merely one aspect of the scientific method and, as such, predates Popper by more than a few years.

You may be as intelligent as you seem to want us to believe, but the empirical evidence is so far not supportive of that notion.

There is a fundamental difference between ignorance and stupidity. So it may be that ignorance is to blame for such faux pas as posting that, when holding AKQxxxx, partner will hold 3+ support more than 60% of the time, but that post did raise some eyebrows.

Ignorance is nothing to be embarrassed about. 99.999999% of the world population lacks the knowledge to be classed as an expert bridge player, and most of them don't care

More importantly, ignorance can be cured, by the simple task of learning. However, assumptions of superior intellect and disdain for those who already possess such knowledge is likely to impede and possibly prevent your progress.

So, while I am not trying to stop you from posting.....engaging in robust discussion of the ideas that are new and appealing to you is a great way to learn....I am suggesting that you approach the forums as if those posting in response to you are (1) your intellectual equals , and (2) where warranted, possessed of greater knowledge than you.

That doesn't mean that they are 'right'. There are a number of highly skilled, highly accomplished players who post here and arguments amongst them are common. Which is, of course, one reason the game is so appealing to most of us who are hooked on it.

[gwnn]

Mikeh so you believe that there are about 0.7 bridge experts alive today?

[mikeh]

gwnn, on 2011-August-23, 09:50, said:

Mikeh so you believe that there are about 0.7 bridge experts alive today?

Unless I am seriously in error, and typed a 9 (or 99) too many, I think I suggested about 70....and that is an error....I'd go with 700. Of course, if we self-rated as on BBO, I'd have to extrapolate to 700,000.

Anyway, if I were VM, I'd make some caustic reference to your obvious failure to understand hyperbole, invoking a factoid picked up from wikipedia

[VM1973]

gwnn, on 2011-August-23, 07:55, said:

Yes I suppose she meant 1.6 instead of 2. I am not sure, but she is around, she can tell us.

Looking at AKQJ AKQ AKQ AKQ will not give you an accurate picture of how many HCPs are worth how many tricks. For example, it would mean that you need 28.5 hcp's for 10 tricks, a conclusions that nobody accepts. The reason why AKQJ AKQ AKQ AKQ gives you a false picture is twofold: first of all it is a hand that offers you certainty for 13 tricks, second of all because in real contracts you often take tricks with jacks, tens, or even twos, because of length.

I know about the pitfalls of double dummy evaluations, you do not need to give an example hand to tell me about it. However, you cannot dismiss such investigations altogether. Declarer plays much better when he sees all four hands, but defenders also make much better opening leads. There have been other investigations (sorry, no links) that came to the conclusion that double dummy results give better results than reality for declarer at grand and small slam level and worse results at smaller contracts, and that somewhere around the 3NT/4M level double dummy results actually come pretty close to reality. I will not look for this link, it is somewhere here on the forums, I think it was a bridge browser study. I suppose you can disregard this as my anecdotal non-evidence, but at least answer me this: how is it that Helene's numbers give much higher importance to controls in suit contracts and the coefficients are actually very close to 4321 for no trump? How can you explain that purely on double dummy errors? Surely if you want to talk only about double dummy errors favouring declarer, they favour declarer in NT and suit contracts as well?

Anyway, it is funny that you link to Tysen's investigations, where Zars perform only slightly better than HCP+321 and in his own words



Surely you don't mean that people who use HCP's never think about their shortnesses? That we, the obsolete HCP addicts will reject invitations after, say

1♠-1NT
2♥-3♥

with all 11 counts, whether 5422 or 6430?

I had assumed we were speaking about NT contracts. Suddenly now I find reference to shapes 5-4-2-2 and 6-4-3-0. Do you often bid NT with those while counting your 0.4 for tens? Somehow I doubt it.

As for Tysen's investigations, I refer you to http://www.bridgebas...7560#entry17560 in which it was suggested that ZPs misevaluate 5-3-3-2 and 5-4-2-2 hands, among others. I've already written to ask for more information but in any event the solution could be quite simple, for example:
Subtract 1 point for 5-3-3-2, 6-3-2-2, 5-4-2-2, and add 1 point for 4-4-4-1. Surely this will improve the performance of ZPs if tysen's research is correct.

As for the idea that 2.846 points per trick is off, somehow I think that's unlikely as using those figures results in 25.6 HCPs for 9 tricks in 3NT, which is not far off from the people who figure 25 or 26 points. In case you've lost my train of thought, I should mention that I'm simply talking about NT, not 6-4-3-0 shapes.

And yes, to answer your question, I do think that those who use 4-3-2-1 to evaluate their hands for strictly no trump purposes do not include points for shortness.