The "Bridge Methods" Blog

13/09/2009

Welcome to Bridge Methods

Filed under: DIP,Systems — BM @ 23:33

Welcome.

This blog contains my thinking as I refine DIP, the strong club system I’ve been working on for a while.  It also deals with methods of targetted interest, with anything from little to total relevance to any system I’m currently considering.  Finally, it deals with random items of bridge interest that may pop-up from time to time.

Well, the above was true.  However recent events have led me to spend some time working on a new method S1N (named after the “Vienna like” very strong and artificial 1NT opening which underpins the rest of the structure).  However, S1N inherits from much of the work done on DIP, in particular the Jammer openings, so there is much crossover.

In due course, the underlying site, http://www.bridgemethods.com, will be created and developed to retain and flesh out the more persistent aspects of what is discussed in the blog.  I hope you find something of interest below.

Regards, BM.

29/08/2010

Advantages accruing to the 2H opening in S1N

Filed under: S1N,Systems — BM @ 16:35

The decision to slightly adjust the S1N openings, and in particular, to play 2 as roughly 11-15 hcp with 4 and a longer minor has had some interesting, and mostly positive, surprising repercussions.

Firstly, there is the obvious removal of the rebid problem with 5m4 hands that many natural methods have after a 1 response.  In S1N, the two relevant sequences would other be 1 1 (showing spades) and 1 1.

Secondly, now that 1 1 is now no longer needed in its natural form (though opener can reverse into 2 when stronger than a 2 opening) has allowed a problem with 2/1 to be addressed.  I like to play 2/1 as absolutely game forcing, but this means that the sequence 1 1NT needs to be quite wide ranging (and arguably, unplayable).  Taking advantage of the property described above, we are now going to play the following “Kaplan Inversion” inspired responses to 1

  • 1:  ART, <5M, F1
  • 1:  5+
  • 1NT:  5+
  • 2:  NAT, FG
  • etc

The above gives the obvious advantages of knowing immediately that responder has 5 in the major, but also, after the new 1 1 1NT “catchall” sequence (no game interested opposite <9+ hcp) it gives responder second chance with invitational values to do something sensible.  Also, the spade fits still get discovered via1 1 1.

Thirdly, it also helps in inverted minor raise sequences.  In S1N, the 1 opening includes 12-17 BAL hands without a 5M.  Once again, 1 2 2 is not needed (as much) in its natural sense, so we can play the following [note, we are playing shape showing continuations after an inverted minor raise - I can't remember the fancy name for this] “Kokish” inspired continuations …

  • 2:  4+
  • 2: 4+ (reversing strength) or14-17 BAL, requests a 2 R, then
    • 2NT:  14-15 BAL
    • 3NT:  16-17 BAL
    • others – as per a strong 5+/4+ hand
  • 2:  4+
  • 2NT:  12-13 (pre-declining a simple invite, if you like).
  • 3:  MIN, NF
  • 3// splinters
  • etc

All in all, I’m quite liking how this somewhat accidentally fits in!

Regards, BM

22/08/2010

S1N v2

Filed under: S1N,Systems — BM @ 11:59

It’s been a busy time at home – August has brought a new baby into the family.  Nevertheless, I hope to get some sleep sometime, preferably before the middle of September when I return to work.  Also, the middle of September is effectively the start of my bridge calendar for the year, so I’d better complete any refinements of S1N in time for it!

One thing my partner in “S1N” has taken issue with is the use of 2 naturally – I think some pressure needs to be taken off 1 as it contains all the weak and strong notrump openings.  He is not convinced of the veracity of this use and instead prefers to use it as part of the pre-emptive barrage.  I have considered this and decided that I can’t keep saying no to this opinion indefinitely as it might cause resentment.  I think it’s best to trial it his way, which will lead to one of three outcomes

  1. I will be more or less proved right, but it will only hurt us a little bit in the meantime (this is what I think is the likely scenario)
  2. I will be proved right, but it will hurt us a lot (this is the bad scenario)
  3. I will be proved wrong (in which case, only my bridge intellect will be bruised)

As part of the “negotiation”, I have secured one compromise and a joint positive outcome.  The compromise is that 2 will be played as a minimum opening with 4+ (usually exactly 4) and a longer minor.  This achieves two positive effects

  1. It takes a modicum of pressure of the 1 opening, and
  2. It removes the rebid problem from this hand type after 1m 1 or equivalent sequences (in S1N, 1 1 shows spades, but it still exhibits the same problem)

If this bid does not prove its worth, then I suppose its back to some form of Weak Twos in the majors (with some mild adjustment to the 2 to 3 openings, which contain two way pre-empts, probably making them similar 2NT to 3 openings).

As to the positive outcome (I think) the 2 Jammer now gets moved to 2.  This gives more safety, in particular the ability to pass 2 (which we often do to the 2 Jammer), but also to have 2 2 as PoC, 2 3 as invitational and 2 4 as usually right-sided.

This leaves the S1N v2 1st seat openings now as below, with the following percentages using some sensible assumptions and ignoring pre-empts for the present.

  • 1:  12-20 hcp, 4+ or 12-17 BAL
  • 1:  12-20 hcp, 4+ or 18-20 BAL
  • 1:  11-20 hcp, 5+
  • 1:  11-20 hcp, 5+
  • 1NT:  ART 20+ hcp or equivalent, 21+ if BAL
  • 2:  7-11 hcp long heart short pointed Jammer (4-5, 3-5 and 3-5/0-2 or 3-5/0-2)
  • 2:  7-11 hcp long spade short round Jammer (4-5, 3-5 and 3-5/0-2 or 3-5/0-2)
  • 2:  11-15 hcp, 4, 5+m

The percentage each opening appears to occur (ten million hand sample) are

  • Pass:  54.03 %
  • 1:  13.94 %
  • 1:  4.75 %
  • 1:  5.23 %
  • 1:  5.42 %
  • 1NT:  0.86 %
  • 2:  6.91 %
  • 2:  7.08 %
  • 2:  1.79 %

Let’s see how it goes!

Regards, BM

01/08/2010

Jammer 2NT continuations

Filed under: S1N,Systems — BM @ 16:59

July has passed me by- busy at work and busy preparing for a newborn early this month.  It might be a good time to get a post in!

Jammer continues to serve well on the infrequent chances I get to play & practise.  Before I go into the 2NT continuations, it is worth noting also that 2 3, 2 3 and 2 3 are now played as invitational.  There is no invite available in clubs after a 2 opening due to space considerations (thought as yet, the 4m responses to 2/ are undefined).  The responses to the strong 2NT relay are shown below, after a 2 opening first, then their equivalents after a 2 opening.

First, 3 always shows length in the non-anchor minor, 3+ length in the non-anchor major.  This is because more relay space is needed in the latter case to resolve relative major suit lengths.

  • 2() 2NT 3:  spades, diamonds and clubs (hearts, clubs and diamonds)
  • 2() 2NT 3+:  spades, diamonds and hearts (hearts, clubs and spades)

Resolution then goes as follows.  When the non-anchor major is held

  • 2() 2NT 4:  5=3=5=0 (3=5=0=5), i.e. long anchor major and minor
  • 2() 2NT 4:  5=5=3=0 (5=5=0=3), i.e. long both majors
  • 2() 2NT 4:  some 5440 short in the non-anchor minor, then 4 asks 5 card suit
    • 4 4=5=4=0 (4=5=0=4)
    • 4 5=4=4=0 (5=4=0=4)
    • 4NT 4=4=5=0 (4=4=0=5)
  • 2() 2NT 3NT:  4=5=3=1 (5=4=1=3), i.e. longer in the non-anchor major
  • 2() 2NT 3:   4/4 M’s, then 4 asks
    • 4 4=4=4=1 (4=4=1=4)
    • 4 4=4=3=2 (4=4=2=3)
  • 2() 2NT 3:  BAL with the anchor major longer, then 3 asks
    • 3NT 4=3=4=2 (3=4=2=4)
    • 4 5=3=3=2 (3=5=2=3)
  • 2() 2NT 3:  UNBAL with the anchor major longer, then 3 asks (3/4 then 3/5 then 4/5 for memory)
    • 3 4=3=5=1 (3=4=1=5)
    • 3NT 5=3=4=1 (3=5=1=4)
    • 4 5=4=3=1 (4=5=1=3)

When the non-anchor minor is held, and after a 3 continuation to the 3 response

  • 2() 2NT 3 3 4:  5=0=5=3 (0=5=3=5), i.e. long anchor major and minor (as above)
  • 2() 2NT 3 3 4:  5=0=3=5 (0=5=5=3), i.e. long anchor major and non-anchor minor (by elimination)
  • 2() 2NT 3 3 4:  some 5440 short in the non-anchor major, then 4 asks 5 card suit
    • 4 4=0=4=5 (0=5=4=4)
    • 4 5=0=4=4 (0=4=4=5)
    • 4NT 4=0=5=4 (0=4=5=4)
    • mnemonic above – “NT shows diamonds” or “long clubs before diamonds”
  • 2() 2NT 3 3 3NT:  4 in anchor major and diamonds not clubs (mnemonic – “NT shows diamonds”), then 4 asks
    • 4 4=2=4=3 (2=4=4=3)
    • 4 4=1=5=3 (1=4=5=3)
  • 2() 2NT 3 3 3:  5 in anchor major, then 4asks
    • 4 5=2=3=3 (2=5=3=3)
    • 4 5=1=3=4 (1=5=3=4)
    • 4 5=1=4=3 (1=5=4=3)
    • mnemonic above – “long clubs before diamonds”
  • 2() 2NT 3 3 3:  4 in anchor major and clubs (and maybe diamonds) then 3 asks
    • 3NT 4=1=4=4 (1=4=4=4)
    • 4 4=2=3=4 (2=4=3=4)
    • 4 4=1=3=5 (1=4=3=5)

There are a few obvious design decisions that have been made above, most notably, that all void types initially show 4 and above.  This is not foolproof, but I believe (and testing has thus far borne out) that if partner was willing to force opposite a potential 4432, then there is almost always safety above 3NT with an extreme shape.  This means there is now less pressure on shape resolution below 3NT.  Also, it may appear strange that one sometimes resolves fairly common shapes above 3NT.  If you look closely, however, the location of the shortage and relative major suit length was known at 3NT or below.  In many if not most cases, this final resolution will not be needed, and if it is, those bits that resolve above 3NT will promise extra length.

One final point of interest is slam investigation.  Current thinking is that a “reverse end-signal” should be used.  That is, if 4 is not explicitly required for a shape showing purpose, then it should command a puppet to 4, after which

  • 4 – RKCB in clubs
  • 4NT – RKCB in diamonds
  • 5 – RKCB in hearts
  • 5 – RCKB in spades

We could do some more work on the responses (a la Shortwood or similar) but at present, we haven’t.

Regards, BM

24/06/2010

S1N and Jammer again

Filed under: S1N,Systems — BM @ 22:12

After one further practice run (a “win”) on BridgeBase, we wheeled out S1N and the Jammers again at a decent (arguably strong) Swiss Pairs event recently.  We had two system-ironing-out glitches, otherwise we may well have won comfortably (as it was, we finished second).

The Jammers are proving particularly effective.  Normally, I am quite opinionated as to the defence to various exotic weaponry as I almost always find it clear cut.  Playing this style of Jammer and watching others try and deal with it, I confess that I’m not sure I know what the optimum defence is.  My gut instinct would normally be to play double as take-out of the known major (of spades in the 2 opening and of hearts in the 2 opening).  However, this type of approach is typically consistent with a style which strains to enter the auction.  For most opponents doing this against the Jammer openings (which are often 4432′s) it typically hasn’t worked well.  Equally, complete conservatism on their part has rarely worked either – we have made most (maybe all?) of the part scores we’ve been allowed to play.

So, what to do?  I suspect (and this is something I would normally abhor elsewhere, where the opponents have shown or implied shape) that playing double as showing a 12-14, 13-15, 12-15 or similar Weak NT, with a view to allowing partner to pass with xxxx or Hxx even, in the opponents suit, might be the way to go.  This would ratchet up the IMP count, no doubt, but I think if this defensive method and moderately good judgement are applied diligently, it might rebalance the auction in favour of the opponents of the Jammer openings.  It may well be that the Jammer openings are a bit like the house in BlackJack:  theoretically, the odds are marginally in favour of the punter, but the punters on average play so badly that the house does very well in the game.

Still, it is still early days in this experiment and maybe we have some Jammer based hidings coming our way yet.  Until then, I am working on the constructive continuations after the Jammer openings, which also have promise.

Regards, BM

29/05/2010

S1N and Jammer debut

Filed under: S1N,Systems — BM @ 21:27

I played in a congress last weekend for the first time in an eternity, partly to help out some friends of a new partner and partly to wheel out S1N for the first time.

S1N incorporates the same version of Jammer that was mentioned in an October 2009 blog entry of mine, namely 2 as a long spade short round Jammer and 2 as a long heart short pointed Jammer.  On balance, the Jammers proved a success.  They were high frequency as expected.  There were two ridiculous results, one extremely positive where the opponents had a bidding misunderstanding and one one extremely negative, where partner, still less used to them than me, took a flight of fancy in a ridiculous save.  In my analysis they cancelled each other out.

This leaves the rest, where there was a stream of flat to small positive outcomes.  As an example, twice we picked off their spade suit with a 2D opening and prevented effective competition, allowing us to play in a comfortable partscore.  It’s still early days, but I think these bids will be long term net winners.

As to broader S1N, it went fine.  There were a lot of 1 openings, which is to be expected (given all the 12-17 BAL hands without a 5M are contained therein).  Our methods in response, agreed on the train to the venue, were adequate, but I think we’ll need to do quite a bit of work opposite this opening to make the most of the method.

Regards, BM

10/05/2010

Axioms of S1N

Filed under: S1N,Systems — BM @ 20:41

Moving on from my “Substitution, Relays, or something else” blog entry of 25/04/2010, I have been thinking about whether it makes sense to produce a list of axioms.  Clearly, I already tend to think along this sort of line anyway, but what particularly inspired me was an article in the latest Bridge World about an allegedly superior method for opener to respond to a game forcing raise, a la Jacoby 2NT.

Though I wasn’t much a fan of the method presented, it did have at least one sensible basic premise:  a priori, opener should not show shortage as he is likely to be declarer.  However, you will also know that I am a fan of the Balanced Hand Principle and when one uses it in this context, opener will end up showing shortages (and/or extra length) sometimes.  How to resolve the dilemma.

Hence, I began considering the merits of producing bidding axioms.  It doesn’t take much thinking to come to the conclusion that it’s not correct to consider potential axioms as being applicable to all systems.  Any axiom involving jump shifts, for example, is likely to be quite different in a strong club context versus a more straightforward natural approach.  I did however conclude that a system can and probably should have a set of axioms listed somewhere.  If nothing else, it should both help enforce coherence of design and hence also reduce memory strain.

I am therefore approaching the design of S1N with this idea in mind, taking an almost abstract view of it during the early design stages.

Back to the Balanced Hand Principle dilemma of earlier, it remains to reconcile which is the more important.  I suspect the hiding of opener’s shortage does have merit in context – so an effort to modify the basic Balanced Hand Principle structure whilst keeping as many of the benefits as possible is probably in order.

Regards, BM

25/04/2010

Substitution, Relays, or something else?

Filed under: Methods,S1N — BM @ 22:31

In fleshing out a basic set of rules for S1N, one of the main things to decide is the systemic approach as to how to show shortage (and length where relevant, though it’s not a relay system per se, so this is slightly less relevant).  I come from a relay background, so that is the preferred approach for me, but many people use substitution, or a mix of substitution and relays:  often, because they don’t consider it from an overall system and memory perspective.

To illustrate what I mean, let’s consider the hypothetical auction 12 2NT, where 2NT asks about responder’s shape.  Let’s also assume in this hypothetical auction that 3 would should a balanced hand and that 3through 3 need somehow to show shortage.  You’ve got at least four options:  I’ll put them in my order of preference

  1. Relay style:  Shortage down the line (and, where relevant,  length up the line, though some, perversely in my opinion, don’t regard this as an essential complement)
  2. Transfer substitution style
  3. Relay style:  Shortage up the line
  4. Simple substitution style

Let’s look at (4) first.  In simple substitution, bids describe something about the suit bid.  So here, 3 would show a diamond shortage and 3 would show a heart shortage.  3, the trump suit, acts a substitute for the unavailable 3 bid and shows a club shortage.  It’s pretty clear to me this is poor as over 3 and 3, the opposition have a clear chance to double (where it is for a lead, or for to suggest a sacrifice, of both, doesn’t matter:  it’s easy for them).

Looking at (3) next, shortage up the line, at first glance it doesn’t appear too bad:  3 shows club shortage, 3shows diamond shortage and 3 shows heart shortage.  However, imagine a hypothetical 1 2 2auction.  Now all the shortages line up with the bid suits:  not good.  You are then faced with the choice of accepting this, or creating an exception, which gives its own set of memory problems.

Transfer substitution, no (2), improves on simple substitution.  You can play it transfer up or transfer down:  here I’ll assume transfer up.  So, 3 shows a heart shortage and 3 shows a club shortage.  3 acts for the unavailable 3 bid and so shows a diamond shortage.  There’s nothing fundamentally wrong with this approach other than being slightly more difficult to remember that some other methods (also, occasionally you get the problem where there are two substitutes required, necessitating a rule to determine which to apply to which missing bid).

My preference is of course (1), relay style, shortage down the line.  In this case, that gives us 3 showing a heart shortage, 3 showing a diamond shortage and 3 showing a club shortage.  So, no coincidence of suit bid and shortage.  Even in the 1 2 2 hypothetical auction, this remains true.  Some fairly basic mathematics will tell you that in this type of scheme, you will have at most one bid coinciding with the suit shown, and often none, as in the two examples above.  Hence, an easy to apply rule with little or no need to memorise exceptions.

Interestingly, I learned relays length down, shortage up.  I think this is an understandable but unfortunate extension of wanting to shows the majors first.  I don’t intend to go back!

Regards, BM

19/04/2010

S1N

Filed under: S1N,Systems — BM @ 21:13

It may indeed be a sin, but I am in the throes of a new potential partnership and looking at playing something along the following lines

1:  NAT 12-20 hcp or 12-17 hcp BAL

1:  NAT 12-20 hcp or 18-20 BAL

1: 5+, 11-20 hcp

1:  5+, 11-20 hcp

1NT: ART, 21+ hcp if BAL, 20+ hcp if UNBAL

2:  6+ or 5+/4M, 11-15 hcp

2+:  various weak hand types, not strictly relevant to considerations

I am interested in any general comments or ideas, but particularly, in two things

  1. Would you play a 3 tiered response structure to 1NT (i.e. Double NEG, NEG and POS) or 2 tiered response structure (i.e. Non-POS and POS)?
  2. Do you think something like the 2C opening as described is wise (the alternative being wrap it up in the 1C opening)?

By the way, “S1N”?  It’s for “Strong 1NT”.

Regards, BM

05/04/2010

Composite Expected Parity – Reprise

Filed under: DIP,Systems — BM @ 22:28

In my late March blog entry, I mused about the possibilities of Composite Expected Parity.  Since then, I dusted off my Deal programming skills and ran a ten million hand simulation.  For simplicity, I used QPs (i.e. Q=3 based points) rather than ZZ points (similar, but singleton K=1 and singleton Q=0).  I hope you will agree, however, that the results are likely to be directly comparable to those which would be obtained for ZZ points.  Finally, I discarded hands of <3 or >14 QP’s:  the bottom limit because it is the weakest hand where such a question might be asked in DIP and similar methods (i.e. a DIP semi-POS) and the upper limit, somewhat more arbitrary, to ensure the equivalent of adding AKQ twice in controls to a base of 2 QP’s.

The raw figures, converted to percentages, were as follows.  With respect to the paritys, K is mentioned first, so EO means Even King, Odd Ace parity.  I’ve highlighted what “Expected Parity” predicts in green.

QP’s EE EO OE OO
3 0.61% 3.99% 6.58% 0.00%
4 2.50% 6.59% 3.61% 0.00%
5 3.63% 3.65% 0.79% 6.62%
6 4.21% 0.77% 0.66% 9.66%
7 3.97% 3.67% 0.78% 4.68%
8 1.82% 4.64% 3.94% 0.87%
9 0.38% 2.56% 4.71% 0.82%
10 1.78% 1.09% 1.98% 0.86%
11 1.95% 0.39% 0.33% 1.10%
12 0.75% 0.11% 0.34% 0.92%
13 0.15% 0.35% 0.32% 0.31%
14 0.04% 0.32% 0.15% 0.04%

Before I progress further, a definition.  The efficiency of any parity method I have looked at is measured in steps to the first denial after showing the paritys.  In DIP, this means the denial of a K in the longest/lowest suit.  So to illustrate this with an example, imagine a sequence where 3 shows exact shape (say 4531) 3 asks controls, 3NT shows base (say 6 QP’s) and 4 is the first parity ask.  To illustrate this (note:  OO is expected parity for 6 QP’s), the auction might continue any of the following ways (remember, the A parity expectation reverses if the K parity one is not as expected)

  • OO:  3 steps:  4 4 (Odd K, Odd A, deny K)
  • EE:  4 steps:  44 4 4NT (Even K, Even A, deny K)
  • EO:  5 steps:  4 4 4 4 4NT 5 (Even K, Odd A, deny K)
  • OE:  4 steps:  4 4 4 4NT (odd A, even K, deny K)

Of course, it is still possible I have messed up the subsequent calculations (I am fairly confident in the percentages already listed above) but hopefully not.  On the basis of my calculations, I have determined the following

  • 3.48 steps:  Theoretical optimal parity:  If you remembered all the percentages above wrote, this is what you could achieve.  Note, 3.48 steps represents just under halfway between 4 and 4NT in the example I used above.
  • 3.51 steps:  Composite expected parity, as outlined in the previous blog entry.
  • 3.61 steps:  Expected parity, K first (my preference)
  • 3.64 steps:  Expected parity, A first
  • 3.96 steps:  AOKO, i.e. arbitrarily showing odd parity first in all cases, A first
  • 4.04 steps:  AEKE, i.e. arbitrarily showing even parity first in all cases, A first:  this is the default method used by most methods that I have seen used in practice

So, have I concluded that Composite Expected Parity has the promise I thought it might have?  Ostensibly yes, but in practice no.  Why?  Because the only reason it (and similarly in proportion, Optimal Parity) is superior to my previous and now current again preference of “simple” K based Expected Parity is that it can immediately discount the OO possibility for 3 and 4 QP’s (which cannot happen).  With an “exception” to Expected Parity, where if the first ask shows odd you go straight into specific K showing, Expected Parity achieves the same performance as Composite Expected Parity, without the downsides.  For example, re-hashing the example above for this expected tweak with a 3 QP hand (OE expected parity)

  • OE:  4 4 (Odd K, implicitly Even A as cannot be AK, deny K)
  • EO:  4 44 4NT (Even K, Odd A, deny K)
  • EE:  4 44 4 4NT 5 (Even K, Even A, deny K)
  • OO:  does not exist

Why is this, or if you like, why did my line of thought on Composite Expected Parity prove fruitless?  The reason is that I made a logical mistake that I have seen others make in different circumstances, and unfortunately, did not learn from.  It was only when I began manipulating the data that I realised the same error.  Another place I have seen this faux pas is when I was presented with an allegedly superior method for single suited resolution that standard symmetric relay.  To show the efficiency of this new method, the person presenting showed me where the shapes resolved.  What they had missed was the loss of zooming efficiency in their new method – they would have been better off considering where Base Controls would have been shown for each shape in their new method.  This is why in my example above I consider the end point as being when the first specific K is denied rather than when the paritys are shown - analagous to Base Controls in the example above.

So, what have I learned?

  • Expected Parity, something that devised myself (though as always, standing on the shoulders of others) is very efficient, and with the simple tweak for QP = 3 or 4, more than half a step more efficient than the AEKE that most practitioners work with.  DIP will stick with this.
  • Expected Parity is at its weakest for “1+(n*3)QP’s”, notably 10 and 13 QP’s.
  • Expected Parity works less well, the higher the QP’s.  When I started writing out the honour dispersions that might occur, the reason for this became apparent:  the more honours there are around the less places there are for new honours to go.  So, where expected parity favours even combinations of honours (e.g. AKQ more likely than AA for 6 QP’s) the higher QP levels favour A’s over K’s over Q’s (e.g. AAAKQQ more likely than AAKKQQQ for 13 QP’s).
  • The practical upshot of the above two points is that Expected Parity is at its worst for 10 and 13 QP’s.  However, the simplicity of the rule and the overall efficiency of the method argues for non-tampering, even knowing this.
  • If you really hate the idea of showing K’s first in Expected Parity, you could show A’s first with only a marginal loss of efficiency.

So, to sum up, if you’re into this sort of thing I suggest you give K-based expected parity a go!

Regards, BM

21/03/2010

Composite Expected Parity

Filed under: DIP,Systems — BM @ 22:17

In my 23/12/2009 post “Frustration showing specific controls – DIP part one”, I talked about, amongst other things, Expected Parity.  I intend to return that sub-topic today.

As I was emptying this dishwasher this morning, I was wondering if it were both possible and sensible for RR to have two different way of showing specific controls, dependent on the actual honour dispersion in his hand.  For example, DIP prefers a variant of the Scandanavian JVCB method, but it, like everything else I have come across, has some horror hands for it.  So, I was thinking, maybe it was plausible to play something like

  • Step One:  unsuitable for JVCB, subsequent responses in some other sort of method
  • Step Two:  K-parity
  • Step Three:  A-parity
  • etc

I’m not sure about the merits of this, but it may be worth a further look at another time.  What it did do, however, was get me thinking how to make up for the lost step that any such approach would occur when defaulting to the primary control showing method (i.e. modified JVCB as in Step Two+ above).  It was then I came up with Composite Expected Parity.

It might be reasonable to make the assumption, after hearing your Q=3 control count response, that if partner was sufficiently interested to ask about your K parity, he will also be interested (or at least, it would usually be safe to do so, if he were) in your A parity.  So, why not show them both at once?

The method revolves around having four steps rather than two in response to the first parity ask. I haven’t decided which order they should be yet, and won’t until I’ve run a simulation, but the following is my best guess how it will play out.  Assume the auction has gone … 3 control ask, 4 B+1 QP’s, 4 R

  • 4:  Expected K-parity, Expected A-parity
  • 4:  Non-expected K-parity, Non-expected A-parity
  • 4NT:  Expected K-parity, Non-expected A-parity
  • 5:  Non-expected K-parity, Expected A-parity
  • 5+:  to be determined whether zooming is appropriate

Note that unlike simple Expected Parity, as you are deciding your expectation up front, you don’t modify the second parity on the fly.  Instead, the probability modification will be taken into account in the order in which those four steps above ultimately end up to be (post simulation).

Whether this technique is viable will depend on a couple of factors, at least.

  1. Whether the assumption on safety broadly holds true, and
  2. Should the four steps be in the same, or mostly the same order, for all QP counts?

If (2) above doesn’t hold true, the memory strain may be too great.  However, the method holds great promise – sometimes saving two steps and often saving one.

I have another new idea as well, about multiple uses for 4 and 4 in control showing sequences, but that will have to wait for another time.

Regards, BM

Older Posts »

Powered by WordPress