BBO Discussion Forums: Zar points, useful or waste of energy - BBO Discussion Forums

Jump to content

  • 19 Pages +
  • « First
  • 7
  • 8
  • 9
  • 10
  • 11
  • Last »
  • You cannot start a new topic
  • You cannot reply to this topic

Zar points, useful or waste of energy New to the concept, does it help...

#161 User is offline   Echognome 

  • Deipnosophist
  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,386
  • Joined: 2005-March-22

Posted 2005-August-31, 03:06

For starters, all of these hand evaluation schemes are discrete. That is to say, they only take on certain integer values. (Some take on fractions, but they are not allowing for say any real number.) So the distributions themselves will be discrete distributions. However, we can often approximate a discrete distribution by a continuous one if we have a large enough sample. Obviously, Zar's example is large enough.

Whether the normal distribution is a good approximation or not can be tested and I'm sure he does (whether or not he reports it in his book which I have not read). I would not worry too much about that as many different distributions approach the normal distribution in their limit. E.g. the height of a person is often approximated by a normal distribution and that will be limited to say between 1 foot (for a small newborn) to around 8 feet (for the tallest person in the world). Yet, the normal distribution can take on any real value. The main thing is that you can make the tails of the distribution so small that they are insignificant.

I say that the biggest problem with the empirical tests is that they are designed to test machine versus machine. A better test would be human versus human, especially with players of roughly the same ability.
"Half the people you know are below average." - Steven Wright
0

#162 User is offline   coyot 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 487
  • Joined: 2005-July-09

Posted 2005-August-31, 03:12

Par result is the one that "should" be achieved on a particular board, meaning that if either side bids, they're gonna get worse result than they would get by passing.

We're not taking into account "miraculous" declarer or defense play, we're just assuming "good" play on both sides, in the sense that most of the "good" players in the field will make it.

If 4 makes for 620 and 4 goes down for 500, par of the board is 4 doubled down three. If 4 makes +1 for 650, par of the same board would be 5 just made, because overbidding 4 gives better result for the heart line - and the spade line should pass because they'd go down 4 for 800 in 5.

At least this would be my definition of par: best practically reachable result in which good pairs should end (barring exotic bidding systems, psyches or strange overcalls).

Par need not be the best theoretical result. If you bid 4M and go down one when defense gets a ruff that you could not anticipate in the bidding, 4M down one should still be the par of the board, although 3NT will make on the some board due to some lucky breaks or blocked suit...
0

#163 User is offline   mike777 

  • PipPipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 16,826
  • Joined: 2003-October-07
  • Gender:Male

Posted 2005-August-31, 03:17

Echognome, on Aug 31 2005, 04:06 AM, said:

For starters, all of these hand evaluation schemes are discrete.  That is to say, they only take on certain integer values.  (Some take on fractions, but they are not allowing for say any real number.)  So the distributions themselves will be discrete distributions.  However, we can often approximate a discrete distribution by a continuous one if we have a large enough sample.  Obviously, Zar's example is large enough. 

Whether the normal distribution is a good approximation or not can be tested and I'm sure he does (whether or not he reports it in his book which I have not read).  I would not worry too much about that as many different distributions approach the normal distribution in their limit.  E.g. the height of a person is often approximated by a normal distribution and that will be limited to say between 1 foot (for a small newborn) to around 8 feet (for the tallest person in the world).  Yet, the normal distribution can take on any real value.  The main thing is that you can make the tails of the distribution so small that they are insignificant.

I say that the biggest problem with the empirical tests is that they are designed to test machine versus machine.  A better test would be human versus human, especially with players of roughly the same ability.

I am old!

If I need to pull out my old "ARCH" auto regressive, cond. heteroskadicity, or what- ever books I am in trouble. Unit roots are confusing enough for us non math majors...if i got to pull out my time series stuff....good grief. This was kindergarten level....have no idea what phd do except confuse the heck out of me.

no wonder i cannot play bridge or spell,
0

#164 User is offline   Chamaco 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,908
  • Joined: 2003-December-02
  • Gender:Male
  • Location:Rimini-Bologna (Italy)
  • Interests:Chess, Bridge, Jazz, European Cinema, Motorbiking, Tango dancing

Posted 2005-August-31, 03:23

It seems to me that the original ZAR method reached a hand evaluation system which outputs results VERY similar to the Losing Trick Count (as Misho did point out once), with some adjustment for reevaluating fitting honors etc etc.

Now the MISFIT points seem to handle also negative adjustments.

I have tried to use ZAR points in some borderline decisions, and found out that using LTC with some commonsense would lead to more or less the same.

I believe that even the drawback of ZAR points is the same of LTC: the offensive power is well represented *if we find a fit*, but:
- the defensive power in terms of defensive tricks is not well represented and
- there is a high risk of ending in 3NT baased solely on distributional bidding and not hcp.
"Bridge is like dance: technique's important but what really matters is not to step on partner's feet !"
0

#165 User is offline   jikl 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 558
  • Joined: 2004-October-08
  • Location:Victoria, Australia

Posted 2005-August-31, 04:54

The biggest drawback of the system would be that of full disclosure.

By giving full disclosure in a real match (with time limits) would be impossible. The opponents don't need to know your system, but you have to be able to explain it in terms that can be understood.

Here is what I mean:

The players are NZ1 (non-Zar player), NZ2, Z1 (Zar player), Z2. The other you will recognise.

Z1: Alert 1NT
NZ1: OK, what's it mean?
Z1: OK, it is 26-30 Zar poi-
NZ1: What the ^&*^ is a Zar point?
Z1: I'll get to that in a second, 26-30 Zar points, no 4 card major, could contain a singleton...
NZ1: I repeat, WT# is a Zar point?
Z1: OK, a Zar point is.......................
NZ1: DIRECTOR!!!!
TD: Yes?
NZ1: This person won't explain his bids to me properly.
TD: What do you mean?
Z1: I explained...
TD: (to whole table) Why are you still on the first board? We are 10 minutes into play and only one bid has been made!
TD: What does your bid mean?
Z1: Well... (lengthy explanation)
TD: I will be right back
TD: (after discussion with fellow TDs) OK, simple, you may not play this system as you cannot explain what the bids mean. Furthermore, since we are now 5 boards into the movement, you are fined 3 IMPs per board for those 5 boards. You are now playing SAYC whether you like it or not. Here are the appeal forms.

Some of you will hate this, some of you will like it, but I can see it happening.

Sean
0

#166 User is offline   ochinko 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 647
  • Joined: 2004-May-27
  • Gender:Male
  • Interests:Cooking

Posted 2005-August-31, 05:53

I remember a post from rgb from a player playing against a well known star. The poster partner asked:
'How many points does your partner have for that bid?'
and got the answer from the star:
'You're in the big leagues now, sonny!, we don't use points to evaluate hands'

Ok, this was obvoiusly somewhat rude, but you get the point.

In terms of distribution, and assuming you play the purer version with 4 card majors, Zar's 1NT excludes:
1. a 4 card major
2. a 6 card minor
3. two 5 cards minors

You won't have #3 in a regular 1NT opening, but you could well have #1 or even #2. Obviously you can tell your opponents more about your partner's hand than you will be able if he opened a SAYC, 2/1, or Precision 1NT.

In terms of HCP you won't be too vague either. If I have understood the bid correctly your partner will have 13-22 HCP + points for high card controls (A=2, K=1). So you can tell that he can't have 4 Aces, and if he has 3 Aces he won't have more than a King besides and nothing else. Again, you are able to supply your opps with more info regarding your partner's opening than they would be able to do if they had opened 1NT. All that assuming they won't be bothered with Zar points.

Petko
0

#167 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2005-August-31, 06:41

Couple of issues from the last stream of post...

Echo... Zar's books are totally FREE (he is not "selling" anything but ideaas".

Awm.. your hand with the ACE of clubs removed is not 47 ZAR points. It was 47 ZAR points BEFORE the superfit was found. Now, the hand with the removed club ACE adds one point for the spade queen (48). Due to the superfit, the ZAR misfti points can be addes as well. So that is difference of 2 in clubs, and 1 each in hearts and diamonds for a total of 4 Points. From a superfit point, Zar's text says to add the larger of the MISFIT points (here 4) or the "super trump points" (here two). Since 4 is bigger than 2, the new total on this hand becomes 52. Zar would say to bid this "game" even without the club ACE. So the par result is reached mathematically by default as it was.

Jikl - Zar 1NT opening bid is perfectly explainable in terms of HCP and distribution. ZAR is just one way of expressing values, you can obviously convert back and forth between units when talking to people unfamilar with ZAR (although the sound of your explaination often will seem odd to the un-informed). .

Z1 - 1NT
Z2 - ALERT
NZ1 - EXPLAIN PLEASE
Z2 - BAL or SEMI-BAL, no 4+Major, no 6+ minor, 9-18 hcp

This "wide range" might violate some local rules, but it describes the 1NT opening bid.
--Ben--

#168 User is offline   hrothgar 

  • PipPipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 15,488
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Natick, MA
  • Interests:Travel
    Cooking
    Brewing
    Hiking

Posted 2005-August-31, 06:55

awm, on Aug 30 2005, 11:56 PM, said:

There's a lot of different stuff on the ZAR website. The simulation results on different types of hand evaluation are actually pretty convincing. This seems like a fairly straightforward thing to figure out -- supposing that we make game (or slam) decisions primarily on the basis of some evaluation method (i.e. high card points + distribution, losing trick count, ZAR) how often will we get it right? He's careful to measure from both sides (both good games missed, and bad games overbid to). It's a fairly clear case that ZAR is better than any comparably "simple" evaluation system out there.

Tysen has done a lot of very good studying the statistical merit of different hand evaluation metrics. There following URL will take you to a summary where he presents some of his findings. http://groups.google.com/group/rec.games.b...a3f343e4ef4c2b6

However, I think that its worthwhile to include the following results. Tysen's work clearly suggests that Zar points are less accurate that "easier" methods. Most notably BUMRAP + a 5/3/1 scale for shortage is easier to calculate and more accurate.

I've never seen an adequate refutation for this result.

COMPARING EVALUATORS

I'll extend my previously posted table of evaluator comparisons to see
how much improvement you can expect by using this method:

ERROR SCORE
HCP 1.23 -0.49
HCP+321 1.07 0.00
HCP+531 1.05 0.07
Zar 1.05 0.08
BUMRAP+321 1.03 0.14
BUMRAP+531 1.02 0.21
TSP 1.02 0.21
Binky 0.99 0.32

ERROR is the average # of tricks there is in difference between how
many tricks we think we can take and how many we actually take.

SCORE is an estimation of the IMPs/board we expect to gain against a
team that uses a simple HCP+321 evaluation method. It's a measure of
how much payoff there is for using a better evaluation system.

HCP is A=4, K=3, Q=2, J=1

HCP+321 is HCP + 3 per void + 2 per singleton + 1 per doubleton

HCP+531 is the same with more points assigned to shortness

Zar is HCP + Controls + twice the length of longest suit + once the
length of second-longest suit minus length of shortest suit.
http://public.aci.on.ca/~zpetk ov/

BUMRAP is a substitute for HCP: A=4.5, K=3, Q=1.5, J=0.75, T=0.25

TSP is the method described in this article. It's an attempt to find
the best evaluator using simple whole numbers.
Alderaan delenda est
0

#169 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2005-August-31, 07:08

hrothgar, on Aug 31 2005, 08:55 AM, said:

awm, on Aug 30 2005, 11:56 PM, said:

There's a lot of different stuff on the ZAR website. The simulation results on different types of hand evaluation are actually pretty convincing. This seems like a fairly straightforward thing to figure out -- supposing that we make game (or slam) decisions primarily on the basis of some evaluation method (i.e. high card points + distribution, losing trick count, ZAR) how often will we get it right? He's careful to measure from both sides (both good games missed, and bad games overbid to). It's a fairly clear case that ZAR is better than any comparably "simple" evaluation system out there.

Tysen has done a lot of very good studying the statistical merit of different hand evaluation metrics. There following URL will take you to a summary where he presents some of his findings. http://groups.google.com/group/rec.games.b...a3f343e4ef4c2b6

However, I think that its worthwhile to include the following results. Tysen's work clearly suggests that Zar points are less accurate that "easier" methods. Most notably BUMRAP + a 5/3/1 scale for shortage is easier to calculate and more accurate.

I've never seen an adequate refutation for this result.

COMPARING EVALUATORS

I'll extend my previously posted table of evaluator comparisons to see
how much improvement you can expect by using this method:

ERROR SCORE
HCP 1.23 -0.49
HCP+321 1.07 0.00
HCP+531 1.05 0.07
Zar 1.05 0.08
BUMRAP+321 1.03 0.14
BUMRAP+531 1.02 0.21
TSP 1.02 0.21
Binky 0.99 0.32

ERROR is the average # of tricks there is in difference between how
many tricks we think we can take and how many we actually take.

SCORE is an estimation of the IMPs/board we expect to gain against a
team that uses a simple HCP+321 evaluation method. It's a measure of
how much payoff there is for using a better evaluation system.

HCP is A=4, K=3, Q=2, J=1

HCP+321 is HCP + 3 per void + 2 per singleton + 1 per doubleton

HCP+531 is the same with more points assigned to shortness

Zar is HCP + Controls + twice the length of longest suit + once the
length of second-longest suit minus length of shortest suit.
http://public.aci.on.ca/~zpetk ov/

BUMRAP is a substitute for HCP: A=4.5, K=3, Q=1.5, J=0.75, T=0.25

TSP is the method described in this article. It's an attempt to find
the best evaluator using simple whole numbers.

For what it is worth, I will REPEAT here again for you richard (well for others, since I have pointed this out to you and tysen before). Tysen used ZAR points, period. No correction for ZAR FIT points, no correction for ZAR SUPERFIT, and certainly no correction for ZAR MISFIT points.

Even ZAR's largest study, left out ZAR FIT points if I recall. I have looked at a lot of the "Bad example" hands in the dataset and found that by apply the FIT/MISFIT points to ZAR, as good as the results were, it gets much better.

But then, to each his own.... use whatever works for you.
--Ben--

#170 User is offline   mikestar 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 913
  • Joined: 2003-August-18
  • Location:California, USA

Posted 2005-August-31, 07:17

I have read thru Zar's new book but I don't quite get how to count misfit points at the table. Could someone give me some examples?
0

#171 User is offline   hrothgar 

  • PipPipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 15,488
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Natick, MA
  • Interests:Travel
    Cooking
    Brewing
    Hiking

Posted 2005-August-31, 07:35

inquiry, on Aug 31 2005, 04:08 PM, said:

For what it is worth, I will REPEAT here again for you richard (well for others, since I have pointed this out to you and tysen before). Tysen used ZAR points, period. No correction for ZAR FIT points, no correction for ZAR SUPERFIT, and certainly no correction for ZAR MISFIT points.

And I'll repeat to you "So What"... You're comparing apples to oranges.

In order to apply "Zar Fit points", "Zar Super Fit points", and "Zar Misfit" points you need a fair amount of information. You need a round or two of bidding to start applying all of the necessary adjustments. At the very least, you need partner to make an opening bid where you happen to have either length or shortage. Contrasting a system where you've had the opportunity to exchange additional data with one that has not made any kind of dynamic adjustments is very much a straw man comparison. If you couldn't out-score a competing structure with all this additional data you'd have to be doing something VERY wrong. Personally, I suspect that a system is only as good as its foundation. If BUMRAP +5/3/1 is more accurate that Zar points, I suspect that BUMRAP + fit/misfit adjustments is going to be more accurate that Zar + fit/misfit adjustments.

I'd argue that comparing Zar points to multiple systems leveraging the Milton Work 4-3-2-1 point count is equally problematic, particularly when you're simultaneously excluding any one of a number of systems using modified versions of the "Four Aces" 6-4-2-1 scale. (For example, BUMRAP preserves this ratio). It seems disingenuous to exclude these systems from your comparisons.
Alderaan delenda est
0

#172 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2005-August-31, 07:47

mikestar, on Aug 31 2005, 09:17 AM, said:

I have read thru Zar's new book but I don't quite get how to count misfit points at the table. Could someone give me some examples?

Well, this is the "point" of ZAR bidding system. I believe he has designed it to not only help get focused on ZAR points (in little bite-size ranges), but also to help get across distribution.

But let me see if I can help you. I think the first rule, is the "MIS FIT" points should only be applied by the person who gets teh most information from his partner. Say the captain. Obvioiusly if there are 8 misfit points, and both partners apply them, they will overbid every time.

Zar "estimated" that there are roughly five misfit points, on average, when a relatively balance hand faces a balanced hand. Five points is onr level of bidding. Using this metric, on hands without superfit, you subtract five. So with balanced hand opposite balanced hand, 52 ZAR points is only worht playing at the three level. Luckily that is 3NT. :-)

When you have a FIT (preferably a superfit), then the more "misfit" points you have the better. To estimate the MISFIT points, one partner has to describe his hand. Let's take a very simple example...

1S - 4C where 4C is a splinter, and promises four card spade support.

If you hold...

6-1-2-4 hand, you know right off the bat that you have three misfit points in clubs, estimating the misfit points in the red suits is more statistical than anything ealse. Partner has, on average five cards in the black suits, so he has 8 red cards (maybe only 7 reds, but that give partner a void in clubs or extra long spade). So you will have five more misfit points in the red suit. For a total of 9 misfit points.

This is not exactly rocket science, but on some auctions it gets easier. For instance after partne makes an unsual 2NT or michaels cue-bid, or leapoing michaels. Of course in those cases, counting his normal ZAR points becomes problematic.
--Ben--

#173 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2005-August-31, 07:54

hrothgar, on Aug 31 2005, 09:35 AM, said:

inquiry, on Aug 31 2005, 04:08 PM, said:

For what it is worth, I will REPEAT here again for you richard (well for others, since I have pointed this out to you and tysen before). Tysen used ZAR points, period. No correction for ZAR FIT points, no correction for ZAR SUPERFIT, and certainly no correction for ZAR MISFIT points.

And I'll repeat to you "So What"... You're comparing apples to oranges.

In order to apply "Zar Fit points", "Zar Super Fit points", and "Zar Misfit" points you need a fair amount of information. You need a round or two of bidding to start applying all of the necessary adjustments. At the very least, you need partner to make an opening bid where you happen to have either length or shortage. Contrasting a system where you've had the opportunity to exchange additional data with one that has not made any kind of dynamic adjustments is very much a straw man comparison. If you couldn't out-score a competing structure with all this additional data you'd have to be doing something VERY wrong. Personally, I suspect that a system is only as good as its foundation. If BUMRAP +5/3/1 is more accurate that Zar points, I suspect that BUMRAP + fit/misfit adjustments is going to be more accurate that Zar + fit/misfit adjustments.

I'd argue that comparing Zar points to multiple systems leveraging the Milton Work 4-3-2-1 point count is equally problematic, particularly when you're simultaneously excluding any one of a number of systems using modified versions of the "Four Aces" 6-4-2-1 scale. (For example, BUMRAP preserves this ratio). It seems disingenuous to exclude these systems from your comparisons.

But, as you well know, Tysen made fit adjustments in his "this is better than ZAR" calculations for his systems. And, the other systems make adjustments. take LTC for instance.

And, a final word. The "Final evaluation" was the number of "points" (by what ever criteria), and the level of contract. How many ZAR points are in hand A + B and did they reach makable something or over reach. It wasn't do you open, or do you instanly leap to game/slam. Also, many of the evil ZAR slams found based on lots of ZAR points are off two ACES, sorry, I don't buy that. I have heard of, and have become good at using, BLACKWOOD. I would never chaulk those up to bad system... becasue I would not leap blindly to my death on those just because of the "math says" to do so.
--Ben--

#174 User is offline   Zar 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 153
  • Joined: 2004-April-03

Posted 2005-August-31, 10:57

>
Now as to the bidding system, I think some things may have been ignored. On many competitive hands, it is essential to evaluate the degree of fit early in the auction. This is actually more important than figuring out the combined level.
<

Completely agree on the importance of the FIT (and the MISFIT for that matter), and completely disagree on stating that those “have been ignored”.

Zar Misfit Points and Zar Ruffing Power address EXACTLY this important issue. Please see the first 15 pages of the Backbone Book for the Misfit and the last chapters of the first Zar Points Hand Evaluation book for the Ruffing Power calculations (the first and second download items in the Download Section of the website).

>
It's often the case that you can figure out that you want to bid 4♠ on a hand, without actually knowing whether the contract will make or be a good sacrifice.
<

Agree again. This is usually the case in competitive bidding (and you find this out a bit too late :-)

>
This is where the ZAR bidding system seems to have a lot of holes. Artificial bids like the 1♣ and 1♦ calls tend to make it hard to find the best fit right away. Four card majors are also weak in this regard. The tight limits on the ZAR strength of opener's hand help you figure out what you can make, but in competitive auctions the goal is the par spot, not necessarily the making spot.
<

Amazingly enough, I agree again (on the importance of the par). I fail to see though HOW knowing your play level would hamper your ability to compete (including your sac decisions).

>
I'd argue that precision style openings (accurate to within two levels) are sufficiently precise in terms of playing strength to reach a good spot in most constructive auctions. This is also the reasoning behind the preemptive openings that ZAR bidding seems to ignore. How can partner know whether to raise to game if I open 3♠ on a wide range of hands? In general he cannot know whether game will make. But whether bidding 4♠ is right, is not necessarily the same thing as whether 4♠ will make!
<

So you are trying to project that when your PD opens 3S and you bid 4S, you have no clue whether you are going to make it +1 if not +2 or you are going to go -3 doubled. Either way it’s gonna be a good result :-) This reminds me of a VERY SIMILAR issue discussed on another Zar Points threads here on this forum (there are some 15 threads about Zar Points here – the one we are currently on is just 1 of these 15). The thread is called “One for Zar Points fans”. Your PD opens 4S and you hold:

♠ AJ9
♥ AK932
◆ JT
♣ A95

What do you do? ... It’s gonna take some ”inferencing” on the information he communicated so clearly. Take your time - I’ll not call the Director :-)

Cheers:

ZAR
0

#175 User is offline   tysen2k 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 406
  • Joined: 2004-March-25

Posted 2005-August-31, 11:45

Okay, everyone knows that I will have to eventually chime in here...

Let me just start off by saying that I've spent a couple hours reading all of Zar's stuff. He has put a lot of effort into this.

First let me comment on what Ben and Richard were saying about my comparisons of Zar vs. other systems. The one Richard posted here in this thread was my initial evaluation where it simply compares initial evaluations. Zar doesn't have fit points but neither does any other system. I later posted another one that compares all of them with fit points. Zar did much better, but was still behind BUMRAP+531+2 per extra trump (and behind TSP).

                 R2     Ave Error   Score 
HCP             0.65     1.21     -0.33 
Bergen          0.71     1.12     -0.03 
HCP + 321       0.71     1.11      0.00 
Zar             0.72     1.10      0.04 
HCP + fit       0.73     1.07      0.14 
BUM RAP + 321   0.73     1.07      0.14
TSP             0.74     1.06      0.20 
Zar + fit       0.74     1.05      0.22 
BUM RAP + fit   0.75     1.03      0.32 
Binky           0.75     1.02      0.33 
TSP + fit       0.76     0.99      0.44
Evolved Binky   0.78     0.97      0.54 


I also did another study that went into a lot more detail and also took care of the "missing 2 aces" problem. That study was here.


Okay, on to Zar's new stuff. I might comment on his bidding system later, but for now I'll just talk about the points since they are seperable. Zar's MISFIT points seemed like an interesting idea, but I did notice something was missing from Zar's work. Zar, where is your data that shows how Zar+Fit/Misfit performs better than simply Zar+Fit? The thing is, when I did a quick study of it's performance, the Zar+Fit/Misfit was actually worse than the Zar+Fit.

So I decided to look into the Misfit points. I looked at a bunch of hands that have no 8-card fit. Then I compared M4 to (Unadjusted Zar) - (Tricks * 5) - 2. This number is the amount of points we "should" adjust down for the misfit. For example if the unadjusted Zar total was 60, but we can only take 10 tricks, then we should adjust down by 8 points (60-10*5-2) to get down to 52 points. Zar's Misfit theory says that these two numbers should be about the same since he wants you to subtract the Misfit points if you have no fit. Here is a graph of the two plotted against each other:
Posted Image

It doesn't take a PhD in statistics to know what an R2 of 0.105 means. Misfit points as they are calculated don't have anything to do with how many points you need to downgrade by.

Tysen
A bit of blatant self-pimping - I've got a new poker book that's getting good reviews.
0

#176 User is offline   Zar 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 153
  • Joined: 2004-April-03

Posted 2005-August-31, 15:02

>
I remember a post from rgb from a player playing against a well known star. The poster partner asked:
'How many points does your partner have for that bid?'
and got the answer from the star:
'You're in the big leagues now, sonny!, we don't use points to evaluate hands'
Ok, this was obviously somewhat rude, but you get the point.
<

If you give me a description of this method, I’ll run it through the database also :-) And post the results.

>

In terms of distribution, and assuming you play the purer version with 4 card majors, Zar's 1NT excludes:
1. a 4 card major
2. a 6 card minor
3. two 5 cards minors

You won't have #3 in a regular 1NT opening, but you could well have #1 or even #2. Obviously you can tell your opponents more about your partner's hand than you will be able if he opened a SAYC, 2/1, or Precision 1NT.
<

Unfortunately though, it has the frequency of a Strong NT though rather than a Weak NT – but you can not have it both ways.

>
In terms of HCP you won't be too vague either. If I have understood the bid correctly your partner will have 13-22 HCP + points for high card controls (A=2, K=1). So you can tell that he can't have 4 Aces, and if he has 3 Aces he won't have more than a King besides and nothing else. Again, you are able to supply your opps with more info regarding your partner's opening than they would be able to do if they had opened 1NT. All that assuming they won't be bothered with Zar points.
<

I tend to believe that the important thing to communicate to your opponents is the INFORMATION that you have received rather than the meaning of each and every bid (excluding the deductions you have made looking at your own cards).

There is nothing wrong in having a SINGLE LINE at the top of your CC stating that:

ZP = HCP + CTRL + (a + B) + (a – d), Opening 26+ points, Game 52+ points.

and explain the boundaries of your partner’s bid AND the promised length of the suits.

Don’t you think it’s OK?

ZAR


P.S. I am answering a bit off-tact but I have lots of emails too, so I hope you’ll tolerate the delay.

ZAR
0

#177 User is offline   Zar 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 153
  • Joined: 2004-April-03

Posted 2005-August-31, 16:42

I’ll follow Ben’s approach of “mix-n-merge” to reduce the number of posts.

>
I have tried to use ZAR points in some borderline decisions, and found out that using LTC with some commonsense would lead to more or less the same.
<

Common sense differs, Chamaco :-) That’s actually why we are doing all these “artificial” gymnastics. You can even use commonsense ONLY and determine the value of the hand better than Zar Points – I even know one such guy who does this all the time (I can’t believe he’s doing it constantly better than me – just makes me mad). His name is Garozzo.

>
The thing is, when I did a quick study of it's performance, the Zar+Fit/Misfit was actually worse than the Zar+Fit. So I decided to look into the Misfit points.
<

I see – you first did the study about the Misfit points and then decided to look at the subject :-) There is an explicit note that misfit points (AND other things) are kept out to make the “battle” on as equal footing as possible, but when you have “spent a couple hours reading all of Zar's stuff” it’s hard to see everything, I guess :-)

>
I believe that even the drawback of ZAR points is the same of LTC: the offensive power is well represented *if we find a fit*, but:
- the defensive power in terms of defensive tricks is not well represented and
- there is a high risk of ending in 3NT based solely on distributional bidding and not hcp.
<

I tend agree with both, but your partner will be AWARE of that also, right? He wouldn’t expect you to have 4 defensive tricks just because you have opened and Zar Points value the availability of Controls, when you are limited with 30 Zar Points.

Having said that, to say that Zar Points are “just like” LTC (and I am not projecting that the LTC is truly bad or something) you must have an imagination bigger than mine :-) Which is possible, after all :-)

>
For starters, all of these hand evaluation schemes are discrete. That is to say, they only take on certain integer values.
<

You can take (for example) an Ace for its “face value” in Zar Points (which is 6.18 points).

How is that going to help you AT THE TABLE though?

I believe we have to make some compromises and come up with something good enough for “at-the-table” use rather than presenting you with the opportunity to call the Director as ask if you can use your calculator just for the opening bid.

>
The key point here is that the distribution of the hands will determine the par spot almost regardless of the actual values. This is something ZAR (and the bidding backbone) don't seem to take into account. It might be interesting to try designing a hand evaluation method around the par spot instead of the making spot.
<

That would be nice – it constitutes the art of bidding, Mike, since it definitely takes in consideration the EVOLUTION of the bidding as it progresses (meaning the information you get from your opponents bidding). Unfortunately it is impossible. Here is why. I’ll just take 3 cards from West’s Spades suit and exchange it for 3 cards of East’s Hearts suit. You know the rest of the story.

Cheers:

ZAR
0

#178 User is offline   tysen2k 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 406
  • Joined: 2004-March-25

Posted 2005-August-31, 16:48

Zar, on Aug 31 2005, 05:42 PM, said:

>
The thing is, when I did a quick study of it's performance, the Zar+Fit/Misfit was actually worse than the Zar+Fit. So I decided to look into the Misfit points.
<

I see – you first did the study about the Misfit points and then decided to look at the subject :-) There is an explicit note that misfit points (AND other things) are kept out to make the “battle” on as equal footing as possible, but when you have “spent a couple hours reading all of Zar's stuff” it’s hard to see everything, I guess :-)

So are you saying that you've never actually tested misfit points to see if they work?
A bit of blatant self-pimping - I've got a new poker book that's getting good reviews.
0

#179 User is offline   awm 

  • PipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 8,375
  • Joined: 2005-February-09
  • Gender:Male
  • Location:Zurich, Switzerland

Posted 2005-August-31, 17:53

Okay, ignore my examples, since obviously I'm not up on how to compute Fit/Misfit points. But the major points are these:

Any scheme for computing the making spot will fail when asked to compute the par spot, and vice versa, simply because these spots are frequently not the same. To relate this to ZAR points, an ace is worth something like 6 ZAR (4 for hcp + 2 controls). If you take a pair of hands and change them by replacing an ace with a small card, you will get a pair of hands with 6 ZAR fewer. This would seem to be true regardless of Fit/Misfit points (which are based upon the distribution of the hands and not how big the cards are). Thus ZAR would predict that you should be about one level lower (actually slightly more) on the hand pair without the ace. Very accurate in terms of what you can make. But the par spot is frequently identical regardless of whether you have this ace. This is what LOTT models.

Here are the competitive problems with the bidding backbone:

(1) Strong is a loser in competition. Partner has some idea of the combined potential of the hands (what we can make), especially if 1 is not a pure hcp bid, but has very little idea of how high to compete or when to double. Of course, this effects all precision-like systems and there are compensating gains.

(2) The multi-way 1 will be very difficult to deal with in competition, since you could have a minor suit super-fit (almost always right to bid five over four) or opener could have a chunky balanced hand with enough controls/hcp to boost it into the 31-35 ZAR range (almost never right to compete, almost always right to double). If you hear 1-3M or 1-4M you are very very fixed.

(3) Four card majors which could be balanced or two-way canape (any of 5cM+4cm, 4cM+5cm, or 4cM balanced okay). These tend to do very badly in competition -- I know people who play them and honestly they don't do well. The issue is that it's very hard to work out how much of partner's strength is offensive or defensive, or how big your major suit fit might be in a competitive auction. It can also be difficult to get to the right partscore even unobstructed, another issue which the simulations don't seem to much deal with.

(4) Preempts. Is it really losing bridge to open with less than 26 ZAR? If neither you nor partner can make a game, you still need to bid to the par spot to stop opponents from running all over you. It's quite possible that you can make 2 with less than 52 ZAR, and that's all you usually need to sacrifice over opponents' making 4. This is ignoring the pressure value of the opening preempt as well.
Adam W. Meyerson
a.k.a. Appeal Without Merit
0

#180 User is offline   Zar 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 153
  • Joined: 2004-April-03

Posted 2005-August-31, 21:11

>
So are you saying that you've never actually tested misfit points to see if they work?
<

Tested? I never do this kind of stuff - why bother? Just put it out there and go to the next big thing. If you are wrong, people will let you know – you’ll patch it and keep going. Testing is for people with low self-esteem.

You have to have confidence in yourself.

>
Any scheme for computing the making spot will fail when asked to compute the par spot, and vice versa, simply because these spots are frequently not the same.
<

I actually posted an answer to Mike’s question along those lines just above. However, after that I kept thinking about it and came to the conclusion that they ACTUALLY may be related. The example I gave with pulling 3 Spade cards from West’s hand and exchanging them for 3 Heart cards from East’s hand made me think about all the consequences of that. Your 26 cards stay the same but the suit-breaks changes dramatically. The point though is that this would affect BOTH the par and your top-contract (due to the unfriendly breaks you introduce).

I’ll run some tests with the Dealmaster and the Deeepfiness (the database is not gonna cut it here). I’ll let you know – there are might be deeper relationships here than the ones that meet the eye ...

>
To relate this to ZAR points, an ace is worth something like 6 ZAR (4 for hcp + 2 controls). If you take a pair of hands and change them by replacing an ace with a small card, you will get a pair of hands with 6 ZAR fewer. This would seem to be true regardless of Fit/Misfit points (which are based upon the distribution of the hands and not how big the cards are).
<

That is certainly correct.

>
Thus ZAR would predict that you should be about one level lower (actually slightly more) on the hand pair without the ace. Very accurate in terms of what you can make.
<

True again.

>
But the par spot is frequently identical regardless of whether you have this ace. This is what LOTT models.
<

Again – I don’t want to talk off-the-top of my head, but I’ll do some runs and let you know. Definitely the subject is worth studying.

>
Here are the competitive problems with the bidding backbone:

(1) Strong ♣ is a loser in competition. Partner has some idea of the combined potential of the hands (what we can make), especially if 1♣ is not a pure hcp bid, but has very little idea of how high to compete or when to double. Of course, this effects all precision-like systems and there are compensating gains.
<

Actually the Backbone is just SEAMINGLY like the Strong 1C systems. I guess you touch on the issue by saying that “1♣ is not a pure hcp bid”, but I’d like to put just a couple of words around that. When you open 1C before me, I know that you have 16+ HCP and that chances are that IF we have to compete (referring to vulnerability here), it PROBABALY would be in the sac area.

When I open 1C before you, you have no clue what’s going on really. I may have 28 HCP in Aces and Kings and cut your head off if you dare to shove it under the sword. BUT I may also have as little as 10 HCP for that Strong 1C opening (please see page 20 for reference). So what do you do – sacrifice or construction? Or it doesn’t matter :-) (referring to our previous conversation).

>
(2) The multi-way 1♦ will be very difficult to deal with in competition, since you could have a minor suit super-fit (almost always right to bid five over four) or opener could have a chunky balanced hand with enough controls/hcp to boost it into the 31-35 ZAR range (almost never right to compete, almost always right to double).
<

You can NEVER have a balanced hand when you open 1D.

It’s important to realize that.

>
If you hear 1♦-3M or 1♦-4M you are very very fixed.
<

Aren’t you? :-)

I am actually much less fixed than you due to the point mentioned above – when I open 1D you know that I just cannot have a balanced hand. And when (and if) I open my mouth again, you can count easily where we belong.

>
(3) Four card majors which could be balanced or two-way canape (any of 5cM+4cm, 4cM+5cm, or 4cM balanced okay).
<

The length-restriction is well stated – it is impossible to open 1M and have ANY 6-card suit, be it major or minor. It targets again the negative inferencing.

>
These tend to do very badly in competition -- I know people who play them and honestly they don't do well. The issue is that it's very hard to work out how much of partner's strength is offensive or defensive, or how big your major suit fit might be in a competitive auction.
<

Oh, we are back to the 5-card major again. This is discussed with numbers in the book. I’ll point you to the proper place – around page 30 (just to save space, I am not ducking the question).

>
It can also be difficult to get to the right partscore even unobstructed, another issue which the simulations don't seem to much deal with.
<

Can you please elaborate here?

I may be missing something...

>
(4) Preempts.
<

What da heck is that? :-)

>
Is it really losing bridge to open with less than 26 ZAR?
<

The answer is YES if you have a balanced hand. If you have 13 HCP and less than 26 Zar Points, let your opponents suffer instead of going down. Check any records of any tournament, you might be surprised.

Now, if you have UNBALANCED hand ... I’ll keep my mouth shut :-)

As mentioned in a previous reply tough, you are FREE to use preempts and overload the corresponding (to that suit) bids.

>
If neither you nor partner can make a game, you still need to bid to the par spot to stop opponents from running all over you. It's quite possible that you can make 2♠ with less than 52 ZAR,
<

Very true.

>
and that's all you usually need to sacrifice over opponents' making 4♥. This is ignoring the pressure value of the opening preempt as well.
<

I have nowhere to go here.

I did agree before and will agree now. But I still believe that it’s a matter of style and priorities more than being a vital part of the construction. Overload one level down and adjust according to your preference. I may do it myself :-)

ZAR
0

  • 19 Pages +
  • « First
  • 7
  • 8
  • 9
  • 10
  • 11
  • Last »
  • You cannot start a new topic
  • You cannot reply to this topic

3 User(s) are reading this topic
0 members, 3 guests, 0 anonymous users