[DinDIP]

(Background: Delta Club is a Symmetric variant based on the Polish ideas of separating hands without a shortage [0-1 cards] from those with a shortage, and showing the shortage rather than length first with the shapely hands. It is based on work Ian Casselton and I started a decade ago with much of the shape-showing module developed by Ian. See here . An outline of Delta Club and some interim conclusions about its effectiveness is here.)

For the past 11 weeks the usual game in which partner and I have been playing Delta Club has been suspended while real bridge tournaments have been played (many on RealBridge, as it happens). Because of ill health I haven’t been able to play so I’ve spent the time thinking about how the ideas and shortness-showing tools of Delta Club can be used when playing “normal” Symmetric methods, i.e. where long suits are shown first.

I think there are three areas in particular where Symmetric can be improved:

• Showing shortness opposite BAL hands
  
• Reverse relays; and
  
• Exclusion asks.

Showing shortness opposite BAL hands
If an unbalanced relayer does not have a way to show shortness then finding the right contract can be problematic. The original Symmetric responses to a strong club included 1N with BAL hands. That leaves a lot of room for opener to break the relay and find the best fit and level. However, it comes with the price of making responder declarer in notrumps, the most likely strain when responder is balanced. After decades of splitting the balanced hands between 2♣ (at least one four-card major) and 2♦ (no major, includes 5m332) I now think it’s better to split by strength, with 2♣ showing 11+ and 2♦ 8-10. One reason is that opener rarely relays for slam when responder is 8-10.

Because 2♦ consumes more bidding space there are fewer possible chain-breaks available to opener. This limits opener’s opportunities to describe his hand but there are effective options. One is to bid the suit in which opener is short, with 2N showing a hand with no slam interest that wants to declare notrumps while leaving room for responder to explore for major-suit fits via Puppet Stayman.

A better option, my testing suggests, is to transfer to the shortage, allowing responder to “accept” with no four-card major but interest in a 5-3 (or 4-3) major-suit fit or some contract other than 3N. (More complex alternatives are easy to devise but have heavier memory loads.)

There are a few general rules that are helpful:

• Showing shortness does not promise a four-card major or a three-suiter. With Kxx x AKJxx AQxx or Kxx x AQx AQJxxx opener should show his H shortage.
  
• A bid of 3N by responder shows a stopper (often more) but opener’s 3N is designed to allow the partnership to play there if responder wanted to show a four-card major but play in 3N if there is no 4-4 (or 4-5) fit.
  
• A bid of opener’s short suit at the four level by either player shows maximum values with no wastage.

While most of the time these shortness-showing breaks will focus the partnership on a choice of games, opener can also use them to try for slam. One easy-to-show hand with slam interest is when opener has a 6+card major and a shortage. For example, with AKQTxx AQx x Kxx opener wants to be in slam if responder has a hand like xx Kxx xxxx AQxx or xxx Kxx Axxxx Qx but not if responder has wasted values in diamonds.

Reverse relays
Delta Club allows opener to describe all unbalanced hands when responder has a GF hand with no shortage and many unbalanced hands when a GF responder is also unbalanced. Given that opener will only reverse relay with a pure hand, one that will easily be able to show its values in denial cue bidding, when the shortage is opposite a 4+card suit it’s usually better to have described than to have asked. Delta Club is random in achieving this goal because opener (1) does not know which is (are) responder’s 4+card suit(s); and (2) which suit opener can show shortness in is fixed in advance -- for example after a 1N response to 1♣ opener can only shows shortness in spades as we show shortness in the order SHDC.

When, as in “standard” Symmetric, responder shows length rather than shortness it is clear that the most useful shortage to show is in responder’s suit. And Delta provides an efficient, Symmetric structure for doing so:
2♦ 5+ lowest suit or three-suiter or 5+5+ with lowest and middle suits
2♠ three-suiter (3!D+) or 5+5+ with lowest and middle suits (3!C
2♥ 5+ middle suit
2♠ 5+5+ with highest suit (3♣ = lowest suit, 3♦ = middle suit)
2N 5+ highest suit, 4 lowest suit
3♣ one-suiter in highest suit (3♥ = 6331 etc)
3♦ 5-4-3-1
3♥ 6-4-2-1
etc
(Obviously, the order in which suits are shown can be adjusted.)

This makes it possible to reverse relay over a 1♥, 1♠ or 1N response, with each partnership needing to decide whether to start shape-showing with step+1 or always at 2♦. The former gains steps; the latter ensures consistency and a lower memory load, and provides some additional sequences for opener to break relay. My testing suggests that it is profitable to use this step (or two) to show another family of hands that are often better telling than asking: hands with 4+card support for the suit responder has shown and a singleton or (especially) a void.

Reverse relaying works best when opener has few or no cards that will be hard to show in denial cue bidding. For example, when opener has AQxxx x AKQx Qxx and partner responds 1!S (showing 4+ hearts) to the 1!C opening opener should start to tell. Responder will know that KQxx in hearts are wasted for slam but will be a secure stopper for notrumps while Axxx is a good holding for a suit contract and a fair one for notrumps, and xxxx is even better for a suit contract but terrible for notrumps. If opener’s hand were KQJTxx x AQxx Ax then it’s better to retain captaincy by asking.

Exclusion asks
Once teller has made a second shape-showing bid we use step+1 relay breaks to show shortness in the second suit. This has two objectives:
1. to ensure that teller has stoppers in the suit if asker offers 3N as a contract and to be able to look for a slam if teller has no wasted values; and
2. to get teller to ignore the K and Q of asker’s short suit when responding to an inquiry about strength. (The first reference I have to this is Paul Marston’s notes from the mid 2000s.)

In a practice deal opener held AKQ85 J3 4 AQJT6. Responder showed hearts then diamonds so opener bid step+1 to show a D shortage. When responder’s shape was shown to be 1=5=4=3 opener wanted to play 3N unless responder had extra values or nothing wasted in diamonds. With a hand like x AKxxx KJxx xxx responder would have passed. With the actual hand (9 AKT74 J862 K87) responder showed 7 working 321 points

[straube]

Hi David.

I just read your description of Delta. Not what you're looking for, but I wanted to tell you that my own experience with SCREAM was horrible. SCREAM's club structure is similar (I think) to Delta (1D=most GF, 1H=semipositive, 1S-DN, 1N+ =shapely GF). We were losing way too many fits opposite the SP and DNs. Somehow we didn't even gain on the GF hands.

I thought Delta's non-club openings were very interesting. Would you mind emailing me the notes?

[foobar]

This is somewhat tangential to the post, but SCUM uses 1♥ to show all balanced hands (including 5332). It allows opener to reverse relay using the exact same responses, and greatly reduces the memory load.

For example:
1♣ - 1♥ (balanced) - 1N (reverse relay with ♠)
1♣ - 1N (positive response with ♠)
1♣ - 1♥ (balanced) - 2♣ (opener is balanced as well)

This does come at a cost (symmetric+1), but there's something to be same about being able to use the exact same relay structure for positive responses, reverse relays, 1M - 2C (GFR), and also possibly over 2m openings.

Also a +1 for straube's observation about the Delta 1C response scheme.YMMV, but IME it was awful.

[Kungsgeten]

I think that the Swedish relay system developed by Daniel Auby and Johan Ebenius focuses on separating hands with and without shortness early. I am not familiar with the details though. The method is still played by a couple of Swedish pros.