Bidding system based on enhanced hand evaluation

[tnevolin]

Some time ago I've started discussion about enhanced hand evaluation here.

http://www.bridgebase.com/forums/topic/74642-new-hand-evaluation-method/

Feel free to comment on in. Thanks to all contributors I've updated method slightly increasing its precision and simplifying it even more. This is the updated version.

https://drive.google.com/open?id=0BxM2JfK2YtucRDlySlFJa09RNFU

 

I also wrote an article on this method internals and the way I calculated it here.

https://drive.google.com/open?id=0BxM2JfK2YtucQWxsZmlUR2ZUblU

 

In this article I highlighted benefits this method can bring to the bidding system both in the way of better precision and information concealment. Based on these ideas, I created a sketch bidding system that tries to maximize these benefits potential. Here is it on the google drive.

Description: https://drive.google.com/open?id=0BxM2JfK2YtucOXBHZzdYQW5KWWs

Guidelines: https://drive.google.com/open?id=0BxM2JfK2YtucZ0R6VmRKQnFXdjQ

Bidding chart: https://drive.google.com/open?id=0BxM2JfK2YtucTFFmekQ1ZEFLYVE

 

Even though it is still a skeleton system it is fully playable and covers all contract ranges from partials to slams. One can immediately start using it as it is. However, since it is quite simple and straightforward in nature, any number of popular conventions can be added and used with it. It is pretty flexible.

 

I'd appreciate any comments, suggestions, and criticism that help me improve it even better. Thank you.

[helene_t]

The idea of having 1♦ deny a single-suited hand (i.e. single-suited diamonds open 2♦) is quite popular on this forum in the form of "IMPrecision". I think it is a good idea.

 

Minor issue: The natural 3NT opening is probably not the best use of that opening.

 

Major issue: Negative double should not be limited. And, consequently, freebids generally promise 5+ cards.

 

IMO the 16-19 range for 1NT is too wide.

[jogs]

 

Bidding chart: https://drive.google.com/open?id=0BxM2JfK2YtucTFFmekQ1ZEFLYVE

 

 

This just looks like your version of precision. In the seventies many club pairs in the USA tried precision. In the beginning the system destroyed the field. Then players learned to interfere with the 1♣ auctions. Only the strongest partnership were able to handle interference. Most US club pairs went back to 2/1.

The major huddle most pairs could not overcome was firm agreements on the meaning of bid against interference. It was hard to land on their feet on part score boards. There was no need to make major changes on the initial action.

 

As helene_t says 1NT ranges should be only 3 points(14-16 HCP). Higher NT ranges should be 2 points.

[tnevolin]

The idea of having 1♦ deny a single-suited hand (i.e. single-suited diamonds open 2♦) is quite popular on this forum in the form of "IMPrecision". I think it is a good idea.

 

Minor issue: The natural 3NT opening is probably not the best use of that opening.

 

Major issue: Negative double should not be limited. And, consequently, freebids generally promise 5+ cards.

 

IMO the 16-19 range for 1NT is too wide.

Thank you for comments.

I agree with 3NT comment. I can change it to whatever, since it is very rare case anyway.

Negative double make sense. I'll change it.

[tnevolin]

This just looks like your version of precision. In the seventies many club pairs in the USA tried precision. In the beginning the system destroyed the field. Then players learned to interfere with the 1♣ auctions. Only the strongest partnership were able to handle interference. Most US club pairs went back to 2/1.

The major huddle most pairs could not overcome was firm agreements on the meaning of bid against interference. It was hard to land on their feet on part score boards. There was no need to make major changes on the initial action.

 

As helene_t says 1NT ranges should be only 3 points(14-16 HCP). Higher NT ranges should be 2 points.

I didn't plan to either mimic existing systems or create something unique too. Just picked up whatever good concepts are there to serve the purpose. The main ideas are to use strong opening and avoid detailed distribution description. The fact that 1c looks simlar to precision and 1 major response looks similar to SAYC is a coincidence.

I agree that NT intervals may be more narrow since I don't use strength update for them. Let me think how to rearrange them.

[1eyedjack]

I agree that NT intervals may be more narrow since I don't use strength update for them. Let me think how to rearrange them.

It is certainly classical thought that a 1NT opener/rebid range should be 3 points (for the pedantic mathematicians who define "range" differently that means a difference of 2 between the highest and lowest possibility, such as 12-14, 15-17 etc). I have long subscribed to that view also, and perhaps still do even now, but my conviction is wavering, and certainly I now think that a 4 point range, if inferior, is not inferior by so great a margin as I would have said a few years ago.

 

The argument for the 3 point range tends to concentrate on the population of available uncontested sequences, and a desire to avoid languishing in 2N or 3M too often.

 

But as repeatedly observed, it is a 4-handed game, and most sequences that open 1m end up contested. There is I think a premium to be had in being the first person to get to limit their hand as balanced and within a reasonably narrow range, especially if in the process you force the opponents to come in at the 2 level if so inclined. That argues for sacrificing some of the risks associated with a 4 point range.

[tnevolin]

It is certainly classical thought that a 1NT opener/rebid range should be 3 points (for the pedantic mathematicians who define "range" differently that means a difference of 2 between the highest and lowest possibility, such as 12-14, 15-17 etc). I have long subscribed to that view also, and perhaps still do even now, but my conviction is wavering, and certainly I now think that a 4 point range, if inferior, is not inferior by so great a margin as I would have said a few years ago.

 

The argument for the 3 point range tends to concentrate on the population of available uncontested sequences, and a desire to avoid languishing in 2N or 3M too often.

 

But as repeatedly observed, it is a 4-handed game, and most sequences that open 1m end up contested. There is I think a premium to be had in being the first person to get to limit their hand as balanced and within a reasonably narrow range, especially if in the process you force the opponents to come in at the 2 level if so inclined. That argues for sacrificing some of the risks associated with a 4 point range.

Not inclined to prefer either way, I am still curious about different treatments for NT bid and suit bid ranges. I'd much appreciate if anyone can explain me the idea behind it. Here is what I see in bidding systems so far. Let's take SAYC as an example. Balanced hand ranges are: 12-14, 15-17, 18-19, 20-21, 22-24, 25+. Imbalanced hand ranges are: 12-15, 16-18, 19-21, 22+. First thing to notice is that balanced hands include 5 card minors and, sometimes, even 5 card majors. So there is a high chance you end up in suit contract after opening balanced hand. From the other side you have a high chance to end up in NT after suit opening as well. Narrower balanced hand ranges do not serve NT contract decision exclusively. Another thing to notice is that average balanced range width is only 1 point less than suit one: 2-3 vs. 3-4. Not that huge difference. Lastly, not all bidding sequences lead to single range. Sometimes you don't have time to narrow it down before final decision which result in 4,5,6 wide ranges for both balanced and imbalanced hands. All in all distinctive hand (balanced/imbalanced) ranges benefit corresponding contact type (NT/suit) directly about ~1/4 of the time. Other cases are either mixed or inconclusive.

Summarizing above, I'll rephrase two main questions here:

1. Why it is so important to range balanced hands more narrow than imbalanced ones? Why not to treat them equally?
   
2. Even if there is a big idea behind #1, the SAYC seems to implement it quite poorly for NT contracts. Does it worth the hassle at all?
   

[Zelandakh]

There are a couple of reasons for the difference between balanced and imbalanced ranges but in the end it comes down to captaincy and the scoring system. When making a limit bid, we effectively transfer the primary decision-making to partner. Practice has shown that treating balanced hands in this way tends to work well. The hand is strongly defined and partner can either set the contract or ask us to clarify some minor details and then setting the contract. Unbalanced hands are less well defined in a single bid and it is therefore not good to transfer captaincy immediately. The continuations then reflect this difference.

 

The other thing to notice is that we have an extra level of bidding space for a suit game than a NT one. If the bidding begins 1NT - 2NT, we have to decide immediately "game or no game". By contrast, after 1♠ - 2♠, we have various ways of inviting partner, thus allowing us to limit our hand much more precisely. Similarly, partner can also limit their hand, by raising to 3♠ rather than 2♠ for example. And this is in addition to the fact that we are already starting higher after a 1NT opening. The end effect is that a one of a suit opening can accommodate a far larger range than 1NT can. Of course the wider your range the more problematic it can be but that is a whole discussion of itself for system designers.

[tnevolin]

Thanks, Zel. Quite a good explanation.

[jogs]

There are a couple of reasons for the difference between balanced and imbalanced ranges but in the end it comes down to captaincy and the scoring system. When making a limit bid, we effectively transfer the primary decision-making to partner. Practice has shown that treating balanced hands in this way tends to work well. The hand is strongly defined and partner can either set the contract or ask us to clarify some minor details and then setting the contract.

This is the prevailing wisdom of most theorists. But I don't believe it. For game purposes this thinking is acceptable. When the contract is 1 to 3 NT the standard deviation of the expected results is about 1 trick per board. But for slams there should be a lower std dev.

For game contracts controls is a secondary parameter. For slams it is the primary parameter. 15 to 16 HCP is a narrow range in HCP, but not in controls.

♠ Axxx ♥ Axx ♦ Axx ♣ Axx 16 HCP

♠ QJxx ♥ QJx ♦ QJx ♣ KQJ 15 HCP

The 16 point hand has 8 controls. The 15 point hand has 1 control. In terms of controls this is not a narrow range. The ace hand is much better suited for slams than the aceless hand.

[Zelandakh]

♠ QJxx ♥ QJx ♦ QJx ♣ KQJ 15 HCP

This is not 15hcp in any measure other than walrus count. Given that this is a thread series about improved hand evaluation, it is laughable that you should post this hand to try and make a cheap point. Instead you take away from your (valid) argument and make yourself look like a clueless fool.

[jogs]

So walruses are the final authority of point count!? Let's get serious. When will the bridge community admit that there is no fixed value for any of the honors. The object of the auction to to determine the number of tricks we can make in our best strain. 4-3-2-1 are just rough, very rough approximations of the value of those honors. The actual value is loosely a bell curve. Most of us are using 4 as the mean value of an ace. Many experts claim 4.5 is more correct. During the initial action it is sufficient to know 4 is probably low. During the auction we should continually revise the estimate of our tricks.

[tnevolin]

So walruses are the final authority of point count!? Let's get serious. When will the bridge community admit that there is no fixed value for any of the honors. The object of the auction to to determine the number of tricks we can make in our best strain. 4-3-2-1 are just rough, very rough approximations of the value of those honors. The actual value is loosely a bell curve. Most of us are using 4 as the mean value of an ace. Many experts claim 4.5 is more correct. During the initial action it is sufficient to know 4 is probably low. During the auction we should continually revise the estimate of our tricks.

Jogs, you have a valid point. I saw you posted once in initial thread "New hand evaluation method". It is exactly what you are taling about. I'd recommend you read my article, the hand evaluation document itself and scan comments in this thread. You may find many ideas you like there. This thread is more about building a viable bidding system on top of said evaluation method. I agree with Zelandakh on narrower balanced hand ranges. Even though it may not matter from hand evaluation point, it matters from bidding sequence point.

[Stefan_O]

If you play SAYC-ish (ie 2-over-1 not GF), I also think 1NT=14-16 is technically the best approach.

 

Main reason is you can then bid eg. 1H-2D-2NT=12-13 (rather than 12-14, which is too wide when you have no room to invite).

(This situation is the same if you play a strong 1C-system, where 2-over-1M is not GF.)

 

In a natural system, an elegant way is to place some NT-ranges in 2D-multi.

You can then use this opening-structure with balanced hands:

	1X followed by lowest NT-bid = 12-13
1NT = 14-16
1X followed by jump NT-bid = 17-18
2NT = 19-20
2D-2H-2NT = 21-22.
2C-2D-2NT = 23-24.
2D-2H-3NT = 25-26.
2C-2D-3NT = 27-28.

In a strong 1C (without 2D-multi), the 17+ NT-ranges need to be handled differently, though.

 

____________________________________________________________________

ps. the way I prefer weak openings with 2D-multi is this:

2D = Weak major 8-10 or 21-22-NT.
2H/2S = Weak major 5-7.

This way, you can also safely open weak-2's more often than std SAYC.

[Stefan_O]

I'd appreciate any comments, suggestions, and criticism that help me improve it even better. Thank you.

 

Hi Tim,

 

In your Chart, when you say "No 5 major, no 6 minor, no singlets"

 

What means "singlet" here?

Not familiar with that term...

[Stefan_O]

I'd appreciate any comments, suggestions, and criticism that help me improve it even better. Thank you.

 

I actually get a lot of questions re your notation...

 

In Chart.docx, pg2, Response, it says:

+1 suit 6-11 4+ (major preference)

+1 major (free bid) 6-11 5+

What exactly does "+1" mean?

Does this table apply to normal 1-over-1?

And what is the difference between tables "Response" and "Advance"?

 

Clarifications needed :)

[Zelandakh]

You can then use this opening-structure with balanced hands:

There are many possible structures with various advantages and disadvantages. Some posters will recognise this one:

 

5M(332)

==

12-14: open 1M; rebid lower of 1NT/2M

15-17: open 1M; rebid 2NT

18-19: open 1M; rebid 3NT

 

Bal without 5M

==

12-14: open 1m; rebid minNT

15-17: open 1NT

18-19: open 1NT; rebid jumpNT

 

Another popular idea involves nebulous minor openings:-

 

11-13: open 1♣; rebid 1NT (or first step if playing transfer responses)

14-16: open 1NT

17-19: open 1♦; rebid minNT

 

I have left out the 20+ point hands because that is often independent of other choices, for example:-

 

20-d22: open 2NT

g22-24: open 2♣; rebid 2NT

25+: open 2♣; rebid 3NT

 

or:-

20-21: open 2NT

22-23: open 2♣; rebid 2NT

24-25: open 2♣; rebid 2♥, then 2NT

 

Using the 2♦ opening to aid with the NT ladder is obviously also possible, not only with a weak/strong Multi but also with Benji Twos or the Mexican 2♦. My own scheme based off of a strong 1♣ opening is:-

 

11+-14: open 1NT

15-17: open 1♣; rebid 1NT

18-20: open 1♣; rebid 1♥, then 1NT

21-22: open 1♣; rebid 2NT

23-24: open 1♣; rebid 1♥, then 2NT

25-26: open 1♣; rebid 3♣ with 4-5♥, 3♦ with 4-5♠, 3♥ with no major

27-28: open 1♣; rebid 1♥, then 3♣ with 4-5♥, 3♦ with 4-5♠, 3♥ with no major

29-30: open 1♣; rebid 3NT

31-32: open 1♣; rebid 1♥, then 3NT

 

And there are hundreds of other alternatives. I would not like to claim any knowledge of what is the absolute best method possible but I will say that I am fairly sure a scheme based on a weak/strong Multi is not it except perhaps for specific systems. It would certainly not be first choice whenever I had a reasonable alternative available.

[Zelandakh]

What means "singlet" here?

Not familiar with that term...

Clearly it means singleton. It is basically one way of defining a balanced hand.

 

What exactly does "+1" mean?

It looks to mean non-jump to me, with +2 meaning jump and +3 double jump.

 

And what is the difference between tables "Response" and "Advance"?

The usual difference is that response refers to the partner of Opener and advance to the partner of Overcaller. I assume it is the same here but I agree this ought to be cleaned up with defensive bidding moved to a separate section.

[Stefan_O]

I'd appreciate any comments, suggestions, and criticism that help me improve it even better. Thank you.

 

I really like your work on improving hand-evaluation :)

 

I fail, however, to see how that is related to the bidding-system you present here?

Maybe I am missing some context...?

 

Wouldn't it be more efficient to just start out from a familiar and proven system,

like SAYC or 2/1 (or some existing proven strong-1C system, if such is preferred)

and make the bidding-structure more systematic by (re)defining bids and ranges in

"Evolin points" scale?

 

Would be interesting to hear you motivation behind this new bidding-system.

[Stefan_O]

Clearly it means singleton. It is basically one way of defining a balanced hand.

 

I was considering that option, but then I find no suitable opening bid at all when you have, for example, 5-card minor + 4-3-1 (12-15), which does not make sense...

Maybe better to let Tim clarify himself... :)

[jogs]

Where do tricks come from?

 

When you are a child you ask your parents, "where do babies come from?" If you wish to design a better valuation system, "you should ask yourself, where do tricks come from?" Which parameters have the greatest effects on tricks.

Every current valuation system is making assumptions which aren't true. Why are each of the suits treated as equals? Examine the hand histories. Final contracts are in the majors more than twice as often as in the minors. Why is the CA worth the same as the SA? If the CA is worth 4, the SA should be worth 8. More tricks are made by trumps than in the other suits.

Why do aces, kings, queens and jacks have fixed values?

S A5432 H A432 D A32 C A

Those four aces can't each be responsible from the same number of tricks. The honors are variables. Its value is dependent on other parameters.

S K5432 H K432 D K32 C K

Those four kings can't each be responsible from the same number of tricks. The value of the club king is a mystery.

evolin, if you truly wish to design a better, a more accurate valuation system, you must seek the answers to these questions. Solve those and many other questions will come later.

Of course all current valuation systems really only evaluate my own hand. The initial summing of the points in my hand should be treated as the prologue. It is the initial value of my hand, used for the early bids. This valuation has a high variance of possible outcomes. Still it is perfectly adequate for determining the opening bid. Normally only valuations on the boundaries need to be exact. Should I open or should I pass? With all hands in the middle of the range the exact value does not matter for the opening bid.

The next phase should be attempting to evaluate how well partner's hand fit with my hand. On this next phase we should attempt to determine how partner's 13 cards fit with my 13 cards. Now use this information to attempt to measure our partnership tricks. The original valuation should be used on a interim basis. Once the trump suit is identified, switch to estimating partnership tricks.

Our Tricks

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

Our expected tricks is equal to our combined trumps plus a HCP displacement. e is the error of the estimates. This equation is only the first two parameters. It is for the general case. Notice that our tricks is dependent of the strain.

This equation can be expanded to include the effects of all 2 times 13 cards of the partnership. Each of the 4 suits will have effects on the estimates. The configuration of the honors in each suit will have effects on the estimates. During the auction one should attempt to visualize the effects of these other parameters and continually revise the count of our tricks. This is for the specific case.

[Stefan_O]

evolin, if you truly wish to design a better, a more accurate valuation system, you must seek the answers to these questions.

 

Nyah... not necessarily -- there's more than one way to do it... :)

Using statistical methods, you can certainly discover patterns and relationships without answering any "whys" at all.

Apparently, that is more the approach Tim has taken...

 

The crucial question here: Does it give you an advantage over other, existing methods?

This we will not know until it's been tested in practical play, of course.

 

That said --- if we find one workable method to improve evaluation, obviously, it does not exclude that other approaches even more workable might be discovered in the future... :)

[jogs]

Nyah... not necessarily -- there's more than one way to do it... :)

Using statistical methods, you can certainly discover patterns and relationships without answering any "whys" at all.

Apparently, that is more the approach Tim has taken...

 

The crucial question here: Does it give you an advantage over other, existing methods?

This we will not know until it's been tested in practical play, of course.

 

That said --- if we find one workable method to improve evaluation, obviously, it does not exclude that other approaches even more workable might be discovered in the future... :)

Not so easy. But I'm not the one trying to design a new system. Am only interested in improving bidding judgment.

[tnevolin]

Hi Tim,

 

In your Chart, when you say "No 5 major, no 6 minor, no singlets"

 

What means "singlet" here?

Not familiar with that term...

Sorry, typo. Should read "singleton".

[tnevolin]

I actually get a lot of questions re your notation...

 

In Chart.docx, pg2, Response, it says:

+1 suit 6-11 4+ (major preference)

+1 major (free bid) 6-11 5+

What exactly does "+1" mean?

Does this table apply to normal 1-over-1?

And what is the difference between tables "Response" and "Advance"?

 

Clarifications needed :)

Yep, I've already added terms description at the top. Sorry, I didn't do this earlier. That's why I appreciate all comments.

+X means bidding at X available level for this denomination. I.e. +1 is cheapest bid in this denomination, +2 - jump, +3 - double jump, etc.