Point counting method

[A2003]

Which method is best for counting high card points for hand evaluation?

Anyone doing these methods in the tournaments? Is there any other methods?

 

Zar point count:

Calculating the Zar Points has 2 parts – calculating the High-card Points (HP) and the Distribution

Points (DP).

For the high-card points we use the 6-4-2-1 scheme which adds the sum of your controls (A=2, K=1) to

your standard Milton HCP, in the 4-3-2-1 scheme (A=4, K=3, Q=2, J=1).

The High-card Zar points (HC) you are already very familiar with (Milton HCP + Controls or 6-

4-2-1)

- The difference between the lengths of the Longest and the Shortest suits (we call it S2)

- The sum of the lengths of the Longest 2 suits (we call it L2);

That’s all: HC + S2 + L2.

 

Milton Work Point Count

Ace4

King3

Queen2

Jack1

Two-10s 1

 

Robertson Point Count

Ace 7

King 5

Queen 3

Jack 2

10= 1

 

Reith Point Count

Ace 6

King 5

Queen 3

Jack 2

10= 1

 

Culbertson Point Count:

Include Milton work point count and below

Opener's hand count:

5 card trump suit if raised 1

6 card trump suit if raised 2

7+ card trump suit if raised 3+

4+ card side suit 1+

Responder's (dummy) hand count: deduct points for 4-3-3-3 hand, 3 trumps (versus 4+):

6+ card trump suit 2+

Void 1

Two singletons 1

 

Sheinwold Point Count:

Include Milton work point count and below

Initial Hand Valuation

Void 3

Singleton 2

Doubleton 1

Aceless Hand -1

3-4 Aces"worth more"

Queen or Jack

Only worth full value when accompanied by Ace or King

Singleton King or Doubleton Queen or Jack

"not worth full value"

 

Declarer Hand Valuation

5th card in trump suit 1

6th card in trump suit 3 (2 additional over 5th)

7th card in trump suit5

(4 additional over 5th)

Dummy Hand Valuation

4+ trump and Doubleton 1

4+ trump and Singleton 3

4+ trump and Void 5

 

Goren Point Count:

Include Milton work point count and below

Void 3

Singleton 2

Doubleton 1

Responder's (dummy) hand count: deduct points for 4-3-3-3 hand, 3 trumps (versus 4+), non-guarded honors; add point for face card in trump (allowing up to 4 HCP in trump)

Void 5

Singleton 3

Doubleton 1

Karpin Point Count:

Include Milton work point count and below

Opener's hand count:

5 card suit 1

6 card suit 2

7+ card suit (etc) 3+

Responder's (dummy) hand count,

4+ trump, Void 3

4+ trump, Singleton 2

4+ trump, Doubleton 1

3 trumps, Void 2

3 trumps, Singleton 1

3 trumps, Doubleton 0

 

Roth Point Count:

Include Milton work point count and below

Opener's hand count: Add

Void 3

Singleton 2

Doubleton 1

6+card suit 1+

 

Responder's (dummy) hand count: add 1+ point when opener supports responder's 5+ card suit

0-2 trump Void 0

0-2 trump Singleton 0

0-2 trump Doubleton 0

 

3 trump Void 3

3 trump Singleton 2

3 trump Doubleton 1

 

4+ trump Void 4

4+ trump Singleton 3

4+ trump Doubleton 2

 

Kantar Point Count:

Include Milton work point count and below

Declarer Distribution Points

7 - 8 card Trump fit, Doubleton 0

8 card Trump fit, Doubleton 1 for each Doubleton

9+ card Trump fit, Doubleton 2 extra

 

7 - 8 card Trump fit, Singleton Honor in Side Suit -1

8 card Trump fit, Singleton Honor in Side Suit -1

9+ card Trump fit, Singleton Honor in Side Suit -1

 

Bergen Point Count:

Include Milton work point count and below

Void Value with Trump Fit

Declarer 3

Dummy with 5 trump 5

Dummy with 4 trump 4

Dummy with 3 trump 3

Singleton Value with Trump Fit

"Normal" 2

Dummy with 4+ trump 3

Extra Length Points in Notrump

5 card suit 1

6 card suit 3

7 card suit 5

"Great suit" 1 Extra

"Poor suit" Subtract 1

Extra Length Points in Side Suit with Fit

5 card suit 1, unless balanced (including 5-3-3-2)

6 card suit 3, but 2 if 6-3-3-2

7 card suit 5, but 4 if 7-2-2-2

 

Losing Trick Count:

Simply stated, once partnership has identified a 8+ card suit fit (unless the opener has a very strong-long major, e.g., a "self-sustaining suit), each suit has between 0 to 3 losers; Aces and Kings are not losers (except a singleton King), Queens may or may not be losers depending on the suit support. Adjustments to LTC include: A J 10 [x...] = 1.5 losers, Q x x = 2.5 losers, while J 10 x = 3 losers and Q x= 2 losers, but are a "plus".

[Siegmund]

There are many others. None of them is perfect, many of them are used at least some of the time.

 

If you are interested in what has been done to search for the "best" point count, have a look at Thomas Andrews's hand evaluation articles.

 

LTC is very different from the others you have listed, in that it works very well when you have an appropriate fit, but is of very limited use for opening bids and non-raise responses, except as a measure of what potential your hand may have if partner turns up with a fit.

 

From your list, I tend to start with Karpin, and from the next round onward see what LTC says. If the two methods disagree by one trick I believe the LTC, if two tricks, usually split the difference.

[mgoetze]

Milton Work Point Count

Ace4

King3

Queen2

Jack1

I think this one is pretty good.

[blackshoe]

interesting you should ask, as I just finished reading Better Balanced Bidding: The Banzai Method, by David Jackson and Ron Klinger (Masterpoint Press, 2010). Jackson is a New Zealand expert, and his nickname is "Banzai". His method is based on a paper published in 1987 by Richard Cowan, an Australian, in Applied Statistics, a journal of the Royal Statistical Society. Basically, the paper showed that when both hands are balanced, a 5-4-3-2-1 count gives a more accurate estimate of the trick taking potential (ttp) of the two hands than either the "standard" Milton Work (4-3-2-1-0) count or the "extended" (4-3-2-1-1/2) version. Using hands from actual championship play, the authors show how "Banzai" works, and how it works better, in many cases, than either version of "Milton". It also shows that Banzai isn't perfect — sometimes it gets it wrong, and sometimes Milton gets you to a making game that Banzai doesn't, even though Banzai is "right" about the ttp of the two hands. Luck in the lay of the cards, mostly. The authors also talk about how to use Banzai when one hand is unbalanced.

 

Very interesting book, all things considered. I found his examples of "16 point" (Milton) openers that Banzai shows aren't strong enough for a strong NT very interesting. Example: four bare aces is 16 Milton points, and 20 Banzai points. 20 Banzai points is right in the middle of the weak NT range (18 - 21 Banzai points). I doubt there will be a mass movement to the new method, though. After all "I've been using the Work Count since 1903 (sic), and I don't see any reason to change now!" :P

 

I want to try it out.

[Statto]

The Banzai count is flawed as Thomas Andrews has already pointed out. It is based on the trick taking potential of suits in isolation (Richard Cowen, 1987), with no consideration of the whole hand. For example, KQJ10-xxx-KQJ10-xx opposite xx-KQJ10-xxx-KQJ10 equates to making 12 tricks in NT in the analysis it is based on, when clearly it isn't.

 

In answer to the OP:

1) Are you trying to impersonate 32519 :P

2) I use Milton as a base guide and upgrade/downgrade as necessary, unless it is a highly distributional hand then I start looking at tricks or perhaps losers. A 6-4 shape is always worth some upgrade for a suit contract, it's on the cusp of being highly distributional.

[blackshoe]

The Banzai count is flawed as Thomas Andrews has already pointed out. It is based on the trick taking potential of suits in isolation (Richard Cowen, 1987), with no consideration of the whole hand. For example, KQJ10-xxx-KQJ10-xx opposite xx-KQJ10-xxx-KQJ10 equates to making 12 tricks in NT in the analysis it is based on, when clearly it isn't.

Have you read the book? According to the Banzai count as presented there, there is no way these two hands equate to 6NT. There's 40 Banzai points between them, and 6NT requires at least 49. These hands should bid to 3NT according to the Banzai count.

 

As to it being flawed, of course it's flawed. No point count method is perfect. The question is, does it provide a more or a less accurate view of the trick taking potential of two balanced hands than does Milton? The book makes a good case, IMO, that it is more accurate than Milton for most balanced hands.

 

I grant you that Cowen's paper didn't examine things in the context of whole hands, but only individual suits. But I daresay Jackson has done his homework in this area. If you've read the book and you still disagree, well, so be it.

[Statto]

Have you read the book?

As some reviewers on Amazon have said, I don't need to. However, I have discussed it with people who have.

 

According to the Banzai count as presented there, there is no way these two hands equate to 6NT. There's 40 Banzai points between them, and 6NT requires at least 49. These hands should bid to 3NT according to the Banzai count.

Of course this doesn't equate to bidding 6NT. That's not the point I was making, which was that, in the statistical analysis it is based on, these holdings are valued as being worth 6NT.

 

The question is, does it provide a more or a less accurate view of the trick taking potential of two balanced hands than does Milton?

Milton I still think is better for NT, better still is counting ½ point for tens in a 42 point deck. However, it reinforces the fact that Quacks and Tens are good value in balanced hands for NT contracts, something that some players may lose sight of.

[Zelandakh]

Culbertson came out with several evaluation methods so it depends on which book you are reading as to which is "official". For example, one of his ideas was to take "trump length minus shortest suit" as distributional points when raising partner. In practise he preferred never to use a (modern) point count method at all but rather Honour Tricks combined with simple judgement.

 

One method I am mildly surprised you did not include is modified Milton where an ace counts as 4.5 and an unsupported queen 1.5. I think every system subtracts a bit for doubleton honours where they do not have full value, even if the simplified base numbers do not. One reasonable way of dividing evaluation methods is to take those that use the 3-2-1 model (including Zars and MLTC for example) as one group, with Milton and co (4-3-2) as another and then fill in the more unusual schemes according to whether they are top or bottom heavy in relation to these. Note also that you can combine facets of different schemes. If you like Zar 6-4-2-1 for honours but hate the way it adds distribution you could take the distributional model from, say, Goren. Just scale it up accordingly to something like 6.5-4-1.3 instead of 5-3-1, or scale down the honour count to 4.5-3-1.5-0.75.

 

What works "best" is probably at least partly to do wit the temperament of the players concenrned. If you like to upgrade aggressively then Milton allows this. If you use Zars then the upgrades are already built into the system and you have to be more wary of the misfit hands where you suddenly need to apply the negative adjustment. There are good arguments to say that starting with a primarily NT-based evaluation and then switching to a method that allows aggressive upgrading when a fit is found is a useful method. Of course system plays a part there too. At the end of the day, all of the evaluation methods are simply a tool to help judgement and with the appropriate adjustments should (usually) end up with the same result.

[jjbrr]

4321 not close

[CSGibson]

I prefer counting on fingers & toes. If I run out of fingers & toes, I open 2♣.

[han]

Have you read the book? According to the Banzai count as presented there, there is no way these two hands equate to 6NT. There's 40 Banzai points between them, and 6NT requires at least 49. These hands should bid to 3NT according to the Banzai count.

 

Oh come on, Statto explained it very well, why not read what he wrote!

[billw55]

I reject any method I can't figure in my head in 10 seconds or less. In fact, I generally reject anything too mathematical, as this tends to become a crutch. Ultimately there is no substitute for judgement.

 

4-3-2-1 plus my own discretion is good enough for me.

[blackshoe]

Oh come on, Statto explained it very well, why not read what he wrote!

I did read what he wrote. I stand by my response. The authors of the book I mentioned pointed out that the paper on which it was based did not discuss whole hands, in fact did not provide any hand examples at all. The paper was solely about determining the relative values of the honor cards in a suit. This is not a reason to reject the method in favor of the Work count. After all, at least there's a mathematical basis for the numbers. In the Work count, the basis, afaik, is "seems like a good relative ranking, why not?" Besides the Work count's relative ranking is known to be wrong.

 

Look, if you want to argue against it because you can't add up as many as twenty values in your head, instead of sixteen, that's just fine. If you want to argue that judgment is more important that any count method, well, I agree. All I'm saying is that this Banzai method should not be rejected just because it's unfamiliar. The book gives examples of where it outshines the Work count (plus judgment, presumably, because the examples are from expert play), and also some where it doesn't. In that regard, at least, the authors are not failing to show that sometimes it doesn't work as well as Work, or luck, or whatever. The question in my mind is how will it perform at the table in lower level competition? I submit that we can't know that until a substantial number of people have tried it. That won't happen if people aren't willing to give it a shot.

[han]

The main reason to reject the Banzai count is that it is REALLY BAD!!!

 

Since 1987 there has been a lot of statistical analysis, by people who do have a good understanding of statistics. The optimal honor count for notrump contracts has been investigated very thoroughly by Thomas Andrews, go to his site and read his articles! They are not much harder to read than the Klinger book, and you don't have to pay for them, there is no reason not to read his articles if you are interested in the subject.

 

And of course the 1-2-3-4 count is not optimal. Nobody, not even jjbrr, thinks that the method is optimal. But, as turns out, it is considerably better than the Banzai count. With these results well known, how can one honestly defend a method as poor as the Banzai count? And moreover, dismiss valid arguments that demonstrate that the premises on which the Banzai count is build are very flawed?

 

To repeat:

 

- The 1-2-3-4 count is certainly not optimal, but for balanced hands it isn't so far from being optimal.

 

- I don't reject the Banzai count because it is unfamiliar to me (it isn't).

 

- I don't reject the Banzai count because it is complicated (it isn't).

 

- After reading articles on the subject and doing many simulations myself, my impression is that the Banzai count is considerably worse.

 

- There is nothing magical about "having a mathematical foundation", especially when this foundation is flawed. One should use a method because it is good, not for some emotional reason that is based on nothing.

 

I have no clue why a respected author and bridge player such as Ron Klinger would write a book about the Banzai count. I don't know him. My most friendly guess is that he does not have a strong background in statistics and is unaware of the research of others.

[cherdano]

I don't need statistical studies to know that any method suggesting a hand with 4 aces is a weak NT is terribly flawed.

[blackshoe]

You tell me to go to a web site, and don't provide a link? Thanks. Yeah, I can google it, and I will, but... :blink:

 

Maybe Banzai is bad. Milton is certainly bad, and as I said seems to me to be used "for some emotional reason that is based on nothing".

 

Cherdano (and others): please continue to keep your thinking squarely inside the box. Innovation is anathema! <_<

[Tomi2]

The main reason to reject the Banzai count is that it is REALLY BAD!!!

 

Since 1987 there has been a lot of statistical analysis, by people who do have a good understanding of statistics. The optimal honor count for notrump contracts has been investigated very thoroughly by Thomas Andrews, go to his site and read his articles! They are not much harder to read than the Klinger book, and you don't have to pay for them, there is no reason not to read his articles if you are interested in the subject.

 

And of course the 1-2-3-4 count is not optimal.

 

 

did I miss something or did nobody mentioned Kaplan/Rubens?

 

everytime I go down in a slam or a game that was not percentage I see I had at least 0.5 KR points less then I expected (in fact I overbid)

KR is like a miracle to me, its always right

[blackshoe]

Okay, I've read Thomas Andrews' stuff, and that led me to some old articles on rec. games.bridge about "Tysen Points" (6-4-2-1), and some other stuff. It seems there's been more done in this area, by various people, than I knew. Somebody ought to write a book. :D

 

At the moment, do folks agree that hand evaluation is still more art than science?

[chasetb]

Kaplan-Rubens is horrible for NT contracts, and for suit contracts it is wrong more than you might think. For suit contracts, go with Ace = 4.5 / King = 3.1 / Queen = 1.7 / Jack = 0.7 . For simplicity, you can round King down to 3, and give Queen 1.75 and Jack 0.75, so you work with fourths rather than tenths.

 

For NT contracts, I have been using a sort of hybrid method. Work's 4-3-2-1 is used for honors, but then I count up the number of honors, subtract that number from 7 (using a 15-17 HCP NT, that's close to average), and multiply that number by 0.5. Then finally, look at distribution and placement of honors, as well as the opponents and go from there. Against better players, I downgrade as much as I upgrade, but against worse players, I hardly ever downgrade.

 

Ultimately, I think there is plenty of science, but science can't replace judgment, experience, and luck.

[cherdano]

Maybe Banzai is bad. Milton is certainly bad, and as I said seems to me to be used "for some emotional reason that is based on nothing".

 

Cherdano (and others): please continue to keep your thinking squarely inside the box. Innovation is anathema! <_<

 

Banzai is worse than Milton. Milton undervalues ace for almost all purposes, and Banzai values them even less.

 

And thanks for your recommendation, but I when I can tell some outside-the-box thinking is crap, I will stay inside, thank you.

[han]

You tell me to go to a web site, and don't provide a link?

 

I did not mean to be difficult but Siegmund had already posted a link. Tysen has also done some very interesting analysis, unfortunately he doesn't post here anymore.

[blackshoe]

I looked back at Siegmund's post before I posted that. I didn't see a link. It's there now. Perhaps I'm going blind. :unsure:

 

I did read Tysen's original three articles from r.g.b., and saved the third one, which has the practical point count method he derived. He said he intended a fourth article, but I couldn't find it. Perhaps he never got around to writing it.

 

Earlier you mentioned the search for the optimal point count method. According to Tysen, Binky Points (from Thomas Andrew's original work) are that method, but Binky Points are too complicated for this tired old brain to use at the table, so (again according to Tysen), that makes TSP the optimal method (6421 plus a few adjustments, including 1 point for a suit with two honors, which is the only place where tens get any love). Do you agree? If not, is there an optimal method?

[helene_t]

When I fitted the logistic regression model

(3NT makes) ~ #aces + #kings + #queens + #jacks

based on the GIB DD library I was chocked to see how accurate it MW is.

 

If all the information we have is the combined number of aces, kings, queens and jacks in the two hands, there is very little scope for improvement.

 

But of course one can come up with better methods if one adds more information. Shape, location of honours, collaborating honours, spots, rightsiding.

 

I use MW until a major suit fit has been found. And then I switch to gut feelings or some kind of modified LTC.

[hrothgar]

When I fitted the logistic regression model

(3NT makes) ~ #aces + #kings + #queens + #jacks

based on the GIB DD library I was chocked to see how accurate it MW is.

 

If all the information we have is the combined number of aces, kings, queens and jacks in the two hands, there is very little scope for improvement.

 

But of course one can come up with better methods if one adds more information. Shape, location of honours, collaborating honours, spots, rightsiding.

 

I use MW until a major suit fit has been found. And then I switch to gut feelings or some kind of modified LTC.

 

Given the size of the data set, you could probably introduce additional terms without much risk of overfitting.

 

For example, treat the Ace of Spade, Ace of Hearts, Ace of Diamonds, ... Jack of Clubs as separate predictors.

Add cross terms for Ace of Spades + King of Spades and the like...

Perform the same logistic regression model...

 

The new GeneralizedLinearModel.stepwise method in the 12b release of Statistics Tbx will automatically check which combinations are statistically significant.

 

(Regretfully, I don't have access to the products any more or I'd run this myself)

[the hog]

As some reviewers on Amazon have said, I don't need to. However, I have discussed it with people who have.

 

 

Of course this doesn't equate to bidding 6NT. That's not the point I was making, which was that, in the statistical analysis it is based on, these holdings are valued as being worth 6NT.

 

 

Milton I still think is better for NT, better still is counting ½ point for tens in a 42 point deck. However, it reinforces the fact that Quacks and Tens are good value in balanced hands for NT contracts, something that some players may lose sight of.

 

 

Regardless of whether it is good or bad, I am always bemused when people criticise an argument without even having read the book. It is hard to take such a post seriously.