[DavidKok]

I've been rereading the notes of the Auby Ebenius Club. Previously I avoided the relay as it is complicated, nonstandard and highly non-symmetric. Today I picked it back up again, and it's really not as scary as I thought it was. I failed to find older discussion on these methods so I figured I'd just write up my own. The purpose of this post is to generate discussion on the approach and help me understand how this method works.

To find the sources of these methods check out https://relaysystems...-relay-systems/ and https://relaysystems...th-and-honours/, in particular the 'AEC Uncontested' and 'AEC ZZ-Bidding' pdf's.

Before beginning, two notes.

• This post is currently not intended as a full explanation of these methods. In particular, I'm mostly trying to make sense of the (exact) shape showing sequences and different ways to bid the hands. That means I'm ignoring most of the slam gadgets - e.g. the odd/even aces, the exact honour scans etc.
  
• While seemingly complete I found the source material very difficult to read, also in large part because the structure is so different from the mostly-symmetric relays I've seen. This makes it much more likely that I've gotten some explanations confused or made a mistake somewhere, please point this out if you spot this.

Overview
The Auby Ebenius club is a Precision-like system with a strong club (16+ unbal or 17+ bal), limited strength five card majors (approx 10-15) and a nebulous diamond (0+ diamonds, 11-15) and a variable notrump (10-12 when favourable, 14-16 otherwise). It uses both 2m openings as natural and constructive (11-15, denies a 4cM, shows 6(+)m or 5m4om) to take strain off of 1♦. Personally I am a big fan of this structure, but in this post I only want to focus on the relay structure. If you want to discuss the system as a whole there was older discussion on it here.
Compared to other relay structures a few aspects stand out:

• They use 'ZZ points', also known as relay points (A=3, K=2, Q=1, but a singleton king or queen is downgraded by a point) throughout as a replacement for HCP for relay purposes and even as an additional opening requirement.
  
• Their relay is not at all symmetric, improving efficiency but making it much more difficult to understand.
  
• Symmetric relay splits hands in five types (balanced, single-suited, short two-suiter, long two-suiter, three-suited). Instead ZZ relay splits it in three grand types:
  • No-shortage (at least 2 cards in each suit). Includes shapes like 5422 ot 7222.
    
  • Shortage (a singleton or void somewhere) without 10(+) cards in any two suits.
    
  • 10(+) cards in two suits.
  These three are then further refined later using a set of modules that apply within these categories:
  • The 4* scheme (read: the four card scheme) for hands of any 4333 or 4432 shape.
    
  • The 5+ scheme for hands having shown 5+ cards in a known suit without any other information.
    
  • The 10* scheme (read: the ten card scheme) for hands having shown 10 cards in two suits.
    
  • The 1-suiter and 2-suiter schemes (relevant only after the 2m openings).
    
  • The 1-0-2 scheme (relevant only after the 1♦ opening).
• ZZ relay does not show 11(+) cards in two suits, so the "10(+)" category above really is just a '10' category. As a result hands with a 6-5 distribution or 7-4 distribution have to distort their shape and pick which features to show. It also does not show 8(+) cards in a suit.
  
• Unlike other relays, ZZ relay has a strength split for all hand types, i.e. they show min vs max well before showing full shape resolution. Their split is 0-4 ZZ/5-7 ZZ/8+ ZZ for responses to a strong club, which corresponds roughly to the traditional HCP ranges 0-7/8-12/13+.

Something that confused me quite a bit is the connection between the schemes and the hand types. Clearly the 4* scheme is for no-shortage hands, but e.g. the 5+ scheme needs to be modified to include the fact that we know that responder does or does not have shortage, as we only invoke it after splitting by major hand type. The pdf's behind the links above explain this in more (confusing) detail. These schemes are supposed to be extremely general, and in each application of a scheme we skip over the steps that we already know and compress the responses. This makes the relay more difficult to learn but also more efficient, and also allows the scheme to be used over both 1M-2♣ and 1♣-<a major suit showing response>. Personally I felt that I could benefit from having everything spelled out at least once, so I started writing.

Incorrect but helpful first try
As with most relay and a lot of non-relay methods we often overload certain responses and make them multi-purpose to fit particular hand types, and ZZ relay is no exception. However, this complicated matters quite a lot for me when trying to understand it, so instead I focused on the most common hand types and repeating structures. My idea is that once this is understood we can add the exceptions back in without losing sight of the whole. As a result I have a simplified but incorrect overview that summarises how I think about these bids in ZZ relay:

• 1♦: Weak (0-4 ZZ, i.e. ~0-7 HCP) any distribution.
  
• 1♥: No-shortage, (almost) any distribution, wide ranging positive (5+ ZZ, i.e. ~8+ HCP).
  
• 1♠: Shortage, (almost) any distribution without 5(+) spades, limited positive (5-7 ZZ, i.e. ~8-12 HCP).
  
• 1NT: 5(+)♠, shortage, wide ranging positive (5+ ZZ, i.e. ~8+ HCP).
  
• 2♣: 5(+)♦, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP).
  
• 2♦: 5(+)♥, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP).
  
• 2♥: Reserved for plugging a specific system gap later.
  
• 2♠+: 5(+)♣, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP), zoom to the 5+ scheme.

5+ scheme applied to super-positive with clubs
So let's look at that 5+ scheme with clubs as the long suit and knowledge that there is shortage somewhere. It looks as follows:

• 2♠: 10* scheme, so 10 cards in two suits and one suit is clubs. Since we show the highest suit first with 5-5 distribution this is always a 6-4.
  
• 2NT: High short, i.e. 0-1♠.
  
• 3♣: Middle short, i.e. 0-1♥.
  
• 3♦: Low short, low side suit, so exactly 3=4=1=5.
  
• 3♥: Low short, single-suited, so '3=3=1=7 minus a card'.
  
• 3♠: Low short, high side suit, so exactly 4=3=1=5.
  
• 3NT: The same as 3♠, but zooming to further strength clarification?

What stands out to me is that the middle short is +1 compared to the other shapes, and also the resolution of the single-suiter is kind of unsatisfactory. I don't know how they handle 5440-type hands, they are supposed to be part of this scheme as well. Probably my inference that 3♦ shows 3=4=1=5 is wrong and it is supposed to read "3=4=1=5 or 4=4=0=5 with 3♥ asking"?

10* scheme applied to super-positive with clubs
Next up let's try the 10* scheme in this example.
1♣-2♠; 2NT-?

• 3♣: ♦.
  
• 3♦: ♥.
  
• 3♥: ♠, high short, so exactly 4=1=2=6 or 4=0=3=6.
  
• 3♠: ♠, low short, so exactly 4=2=1=6 (patterns listed with a numerical rule).
  
• 3NT: ♠, low short, so exactly 4=3=0=6 (patterns listed with a numerical rule). Zoom to strength clarification.

In situations where it was unclear whether the shape was 5-5 or 6-4 we'd have to clarify that as the next step before showing shortage.

Continuations over some other responses
With that example out of the way let's go back to the initial responses:

• 1♥: No-shortage, (almost) any distribution, wide ranging positive (5+ ZZ, i.e. ~8+ HCP).
  
• 1♠: Shortage, (almost) any distribution without 5(+) spades, limited positive (5-7 ZZ, i.e. ~8-12 HCP). 1NT relays and responder bids identical to first round responses of 2♣ or higher, but now shows 5-7 ZZ points instead.
  
• 1NT: 5(+)♠, shortage, wide ranging positive (5+ ZZ, i.e. ~8+ HCP).
  
• 2♣: 5(+)♦, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP). 2♦ relays and responder bids identical to first round responses of 2♠ or higher, but now shows diamonds instead. The rebid of 2♥ is instead used to plug a system gap later.
  
• 2♦: 5(+)♥, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP). 2♥ relays and responder bids identical to first round responses of 2♠ or higher, but now shows hearts instead. Note that this means that on the 10* scheme we now have to clarify 6♥4m versus 5♥5m on a 3m rebid.
  
• 2♥: -
  
• 2♠+: 5(+)♣, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP), zoom to the 5+ scheme.

That covers most of the unbalanced hands with a 5(+)-card suit, regardless of strength. One problem is the unbalanced hands with 5(+) spades. We have put these in 1NT, but now we don't have the strength split via 1♠ (and we really wouldn't want to have unbalanced hands with long spades in the 1♠ response anyway). Instead ZZ relay uses a trick they call the "Min 2 step" principle to split the ranges after the 1NT response showing wide-ranging spades. Here are the relay followups:
1♣-1NT; 2♣-?

• 2♦: Reserved for plugging a specific system gap later.
  
• 2♥: 5-7 ZZ, not high short.
  
• 2♠: 5-7 ZZ, high short.
  
• 2NT: 8+ ZZ, high short
  
• 3♣: 8+ ZZ, middle short
  
• 3♦: 8+ ZZ, low short, low side suit so exactly 5=3=4=1.
  
• 3♥: 8+ ZZ, low short, high side suit so exactly 5=4=3=1.
  
• 3♠: 8+ ZZ, 5=4=4=0(??)
  
• 3NT: 8+ ZZ, '7=1=3=3 minus a card', zoom for strength?

As you can see I'm a bit confused by this near the end, I don't quite understand the option that I suspect is 5=4=4=0, nor how we resolve the 6-card spade suit versus 7-card spade suit in time. To make this work the 10* scheme has to not be a part of our 1NT response - so we put it in the direct 2♥ response instead and even come out at -1 (although we unfortunately do have to bid 1♠-then-2♥ to show 5-7 ZZ 10* with primary spades)! Then the scheme above is equal in relay height compared to the unbalance clubs hands and even goes -1 on all 5-7 ZZ hands, the most common set of the bunch. This "Min 2 step" principle is interesting to note: often relay schemes have a structure of the form

• Step 1: A toggle (for example, here: minimum).
  
• Step 2: Some exception hand types.
  
• Step 3+: A list of shape descriptions without the toggle (here: maximum).

In situations where there is no clear choice for the exception you can instead move this around a bit by picking a 'first hand type on the list', and then inserting:

• Step 1: The toggle, not the first hand type on the list.
  
• Step 2: The first hand type on the list with the toggle.
  
• Step 3+: A list of shape descriptions without the toggle.

No-shortage hands
We're closing in completing the structure now, so let's look at 1♣-1♥. Since this response denies shortage this is an excellent opportunity to use reverse relay, where opener decides to show their shape and (minimum) strength instead to let the (mostly) balanced hand evaluate our combined assets. Alternatively, opener can continue relaying for full shape information. The rebids over 1♣-1♥ are:

• 1♠: Relay
  
• 1NT+: Identical to immediate responses of 1NT or higher to 1♣, promising shortage (otherwise why reverse relay). They are limited to 9-13 ZZ, stronger hands should just relay instead.

Resuming with the relay 1♣-1♥; 1♠-? instead responder rebids:

• 1NT: A super-positive (8+ ZZ) with a 5(+) suit or some specific exceptions.
  
• 2♣: 5-7 ZZ bal without a 5(+) suit, i.e. any 4333 or 4432. After this use the 4* scheme.
  
• 2♦: 8+ ZZ bal without a 5(+) suit, i.e. any 4333 or 4432. After this use the 4* scheme (and are a step higher than facing 5-7 ZZ).
  
• 2♥: 5-7 ZZ 5m(332) or 5m4om.
  
• 2♠: 5-7 ZZ 6m(322) or 7m222.
  
• 2NT: 5-7 ZZ 5♠332.
  
• 3♣: 5-7 ZZ 5♣4M22.
  
• 3♦: 5-7 ZZ 2=4=5=2.
  
• 3♥ 5-7 ZZ 4=2=5=2.

This is a lot of memory work, and there are a lot of annoying exceptions. It is fairly straightforward to resolve exact shape on the next round with the schemes or numerical rules for patterning out. However, I've kept the system holes hidden so far - these are the no-shortage hands with a 5(+)cM - i.e. the 5♠332, 5♠422, 6♠322, 7♠222 and then the whole list again with hearts. Also the 4441's have to go somewhere. These mostly need to be memorised and use up the sequences 1♣-1♥; 1♠-1NT; 2♣-2♦, 1♣-1NT; 2♣-2♦, 1♣-2♣; 2♦-2♥ and one shape, 4=4=1=4, even gets added to 1♣-2♥ (our 8+ ZZ 10* primary spades bid). In particular this also means that there are some hands without shortage in the direct 1NT response (and therefore not all hands without shortage respond with 1♥), some unbalanced hands with primary spades in the 1♠ response, and responder ends up bidding their longest suit as an early response every now and then.

The 4* scheme
The 4* scheme is very simple, to the point that I'm worried it might not be optimal. Using the example of the 1♣-1♥; 1♠-2♣; 2♦ auction (showing 5-7 ZZ and 4333 or 4432) the scheme is:

• 2♥: A doubleton major (2♠ asks, then high short first, low short zooms to numeric shape-showing of the 4432).
  
• 2♠: Any 4333, then 2NT asks and we do numeric shape-showing.
  
• 2NT: A doubleton diamond, then numeric shape-showing.
  
• 3♣: 3=4=4=2.
  
• 3♦: 4=3=4=2.
  
• 3♥: 4=4=3=2.

I think this always patterns out at 3♠ at the highest, so the 8+ ZZ version patterns out at 3NT sometimes.

Conclusion
I think this is an interesting relay structure. It shares similarities with both Mecklite (splitting 0-7, 8-11, 12+) and Transfer-oriented symmetric relay (trying to use transfers into our super-positives as well as delayed transfers with our regular positives). The full notes define several more clever relay breaks and tricks on certain auctions, and also cover the 1♣-1♦ followup structure (it uses 1♥ as Birthright - hearts or strong any, with room to fall back into relay). It is also a lot of work, not symmetric, and (in my opinion) not explained clearly in the documentation. There are also some hands that end up patterning out quite unfortunately, e.g. if I deal myself a 5=5=1=2 with 6 ZZ points I think the correct auction is 1♣-1♠ (5-7 ZZ); 1NT-2♥ (10* primary spades); 2♠-3♥ (5♠5♥ secondary hearts - remember that 4=4=1=4? It pushes up the hearts. Since this is the highest secondary suit we zoom to show 5-5 (first step) versus 6-4 (second step)); 3♠-3NT (high short); 4♣-4♦ to get to ESS, though I did also show my 5-7 ZZ along the way. The modules can be reused over the limited openings, though sometimes with slight modifications. On average I think this shows strength and shape lower than the symmetric schemes I've seen, but you also have to work much harder to get there. I've stuck with the 'simple' parts of the relay so far and I'd love to receive feedback on both the structure and my (lack of) understanding of it.

Oh and before I forget, the actual meaning of the first round responses:

• 1♦: Weak (0-4 ZZ, i.e. ~0-7 HCP) any distribution.
  
• 1♥: No-shortage wide ranging positive (5+ ZZ, i.e. ~8+ HCP), contains all no-shortage hands other than option (2) in 1NT.
  
• 1♠: Shortage, any distribution without 5(+) spades, limited positive (5-7 ZZ, i.e. ~8-12 HCP).
  
• 1NT: 1) 5(+)♠ not 10*, shortage, wide ranging positive (5+ ZZ, i.e. ~8+ HCP) or 2) 5(+)♥, wide ranging positive 5+ ZZ, no shortage, not 5♥332 5-7 ZZ
  
• 2♣: 5(+)♦ or 4♦441, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP).
  
• 2♦: 5(+)♥, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP).
  
• 2♥: 10* scheme 5(+)♠ or 4=4=1=4, super-positive 8+ ZZ.
  
• 2♠+: 5(+)♣, shortage, super-positive (8+ ZZ, i.e. ~13+ HCP), zoom to the 5+ scheme.

[DavidKok]

I wrote a long reply earlier today but I was careless in not saving a backup before trying to post. A communication error with the server swallowed it whole, and it was gone. Oops. Let's try again.

This post hasn't seen any discussion, but I'm still interested in the ZZ relay. To find more information about how it works I looked up Nyström-Upmark games on vugraph. In fact, I looked at all their vugraph games dating back to 2019 (there are no older games by them on there). Unfortunately relay games are rare, and I found a total of exactly 37 relay auctions by them in these 4+ years, of which 26 were over 1♣, 2 over 1♦, 2 over 1♥, 3 over 1♠, 2 over 2♣ and 2 over 2♦. It is quite possible that I missed some, in particular I vaguely remember dismissing another relay auction over 1♦.

Combining the above with my earlier notes I think it's fair to say that there's several aspects I don't get, in particular relating to the 10* scheme. Of these 37 relay auctions there were four deals with 10+ cards in two suits and the 10* scheme was only used on two of those deals (and one was a 1♦ opening where opener showed 4♥6(+)m cheaply). I think they may have removed the scheme from their structure over 1♣ and 1M entirely, moving in the rest of the responses. Though I don't know how, why, or what to do with the hands that would have qualified for it. This idea is further corroborated by an example in the pdf: "Ex: 1♠-2♣;2NT (8+zz and any 5521/6421)", compared to board 14 where opener rebid 2NT with a 5=1=3=4. However, there are also other (older) auctions that do have the 10* scheme, leaving me confused.

Other than this most other auctions agree with my limited understanding of the relay, which is a good sign. There are two more main areas of confusion for me: the single-suited resolution with shortage (I labeled this "7331 minus a card" but this may well be inaccurate) and the 'special shapes' answer in the 5+ scheme (does this show exactly 5440, or are other distributions included too?). Below is a link to every deal where I am not sure that I understand the auction. If people have an idea or suggestion please let me know.

• Board 12: 1♣ is strong, 1♥ is GF positive no shortage, 2♦ is a relay break, 9-13 ZZ, 5(+)♥, some shortage, 2♥ relay and now I am not sure. The notes say they don't use the min 2 step principle on a 1NT relay break here, which is also the only way I can make sense of the followups. If they do not have a min 2 step princple here I expect the followups to be:
  • 2♠ = 10* scheme
    
  • 2NT = high short (spades)
    
  • 3♣ = middle short (diamonds)
    
  • 3♦+ = low short
  which agrees with the bid (3♣ showing short diamonds) but implies that the 10* scheme was still there in 2023.
  
• Board 24 straight up is the 10* scheme in reverse relay, opener's 2♥ rebid shows a maximum (11-13 ZZ?) with 5(+)♠ and 10+ cards in two suits, 2NT shows clubs and 3♥ confirms the 6-4 (in fact, I think that zooms to high short, so exactly 6=1=2=4 or perhaps also 6=0=3=4).
  
• Board 8 is more interesting than it may seem, because I think 3NT was a zoom response to relay break. ZZ relay zooms with primary clubs, then further with low shortage, then further with a high suit, so exactly 4=3=1=5. In fact, this is exactly the 5+ scheme from my previous post. However opener is not allowed to go past 3NT facing a hand of unknown strength, so I think 3♠ shows 9-10 ZZ and 3NT shows 11-13 ZZ.
  
• Board 22 has me thoroughly stumped. I thought this would be part of the 10* unbal primary spades structure. Instead 1NT-then-2♥ initiates the min 2 step unbal 5(+) spades step 3+ structure, i.e. 5-7 ZZ with 5(+) spades and no heart shortage. However, I expected 2NT on the next round to then show a club secondary suit, which is clearly not the case. Perhaps the 10* scheme is included here after all somehow, or they don't use the min 2 step rule facing passed hands?
  
• Board 13: again I'm stuck on the sequence intended to show unbal with long spades, 2NT to show short clubs would only make sense if they don't use the min 2 step rule and do not have the 10* scheme here, in which case it shows exactly a 5=3=4=1 (or possibly a 5=4=4=0).
  
• Board 2 similar to before, a zoom on the low short high length maximum hand. This clarifies the response structure over 1M-2♣ a bit, assuming there is a 10* scheme in place:
  • 2♦ = Minimum (5-7 ZZ), shortage.
    
  • 2♥ = Minimum (5-7 ZZ), no short.
    
  • 2♠ = Maximum (8-10 ZZ), no short.
    
  • 2NT = Maximum, 10* scheme.
    
  • 3♣ = Maximum, high short.
    
  • 3♦ = Maximum, middle short.
    
  • 3♥ = Maximum, low short, low suit (so 3=5=4=1 or perhaps 4=5=4=0).
    
  • 3♠ = Maximum, low short, high suit (so 4=5=3=1).
    
  • 3NT = Maximum, low short, single-suited (so 7=3=3=1 minus a card).
• Board 14 unfortunately directly contradicts the explanation above, as this 2NT is not a 10* scheme hand. Back to the drawing board?
  
• Board 24 makes my confusion complete, as it now uses the 3♠ immediate zoom rebid to show a single-suited hand rather than the other major two-suiter with a low shortage.

I'm still trying to get a better understanding of these methods. The approach over 1M-2♣ is particularly interesting, as it seems to be more efficient than the more popular alternatives like Ambra or Grammy that see play. If I have the time and energy I might attempt to compute a frequency score of where the ZZ relay patterns out, but I notice that they almost always have shape resolution by 3♥ in addition to having given strength information, i.e. save the better part of a full question compared to other relay schemes I've seen.

[foobar]

 DavidKok, on 2024-January-14, 09:21, said:

I'm still trying to get a better understanding of these methods. The approach over 1M-2♣ is particularly interesting, as it seems to be more efficient than the more popular alternatives like Ambra or Grammy that see play. If I have the time and energy I might attempt to compute a frequency score of where the ZZ relay patterns out, but I notice that they almost always have shape resolution by 3♥ in addition to having given strength information, i.e. save the better part of a full question compared to other relay schemes I've seen.

That seems really impressive given that SCUM achieves +1 over 1M - 2♣, and it doesn't even have to cater to the 5332 hands. My guess is that some uncommon patterns must fare worse?

[DavidKok]

Yes, I think some 5-5 and 6-4 patterns in particular end up quite high. I'm also a bit worried about the 7(321) shapes. This might still be a gain on average, and I don't really understand how they handle these hand types (from the Vugraph records: they haven't had a relay auction with a hand type like this on vugraph in at least four years and possibly longer - presumably the auction is frequently competitive when they hold this) as they are part of the confusing 10* scheme.

Another interesting example is board 3 here, where Al Hollander said he thought that 4♠ was the final shape-showing call (again, on a 6-4 type hand) though it is unclear to me how this is supposed to work. I expected a 2♠ rebid by responder, initiating the 10* scheme with 5-7 ZZ and primary clubs. If I try to make it fit I think I can't quite make my idea match the zoom from 3♦ to 4♠ - counting the shapes using the AEC "Frequency considerations" level system and numeric scheme I get the following bids:

• 3♥ = side diamond suit
  
• 3♠ = single-suited, i.e. 3=1=3=6 (instead of '3=1=3=7 minus a card')
  
• 3NT = side spade suit, zooming to exact shape so 4=1=3=5
  
• 4♣ = 4=1=2=6
  
• 4♦ = 4=0=3=6
  
• 4♥ = 3=1=2=7
  
• 4♠ = 3=0=3=7

Perhaps there's a 4=0=4=5 in there (but aren't you supposed to bid 3♥ first with that?) but even then I'm a step off from 4♠ showing ther 4=0=3=6.

[nullve]

 DavidKok, on 2024-January-14, 14:56, said:

I expected a 2♠ rebid by responder, initiating the 10* scheme with 5-7 ZZ and primary clubs.

Is it clear from the notes that they include (6430) in the 10* scheme here?

AEC notes said:

The 10* scheme is used when SE shows 10* in two suits w/ the long suits initially shown as
distributed 64/55. The scheme is used both when a void is possible and when it is not.

[DavidKok]

I'm not sure what you mean. Which hand types do you think are in the 10* scheme with primary clubs?

[nullve]

 DavidKok, on 2024-January-15, 12:18, said:

I'm not sure what you mean. Which hand types do you think are in the 10* scheme with primary clubs?

Just the (6421) type, perhaps:

Quote

The 5+ scheme is used when SE has shown a specific 5+ suit, but nothing else about distribution.

1 step No SHO, then:

1 step (5332)
2 steps (6322)/(7222) (1-suiter w/o SHO)
3+ steps (5422)

2 steps (6421), (5521) 10* scheme
3 steps High SHO, but denies distributions eligible for step 2, as does all higher bids
4 steps Middle SHO
5 steps- Low SHO

[DavidKok]

Thank you! That would explain several of the examples that currently have me confused. I'll go through my list of relay auctions again with this base assumption and see if I can make sense of the bidding.

[nullve]

Perhaps the VuGrapsh operator mistyped and the auction went

Upmark Nyström

1♣-1♠(1)
1N-3♣(2)
3♦-4♦(3)
5♣-P

(1) 5-7 zz, shortness (= singleton/void) somwehere, not primary S unless (5521)/(6421)
(2) primary C (and 5+ C), H shortness, not (6421)
(3) 4036, 5 zz

instead, because it seems to me that the relevant structure is

1♣-1♠
1N-?:

(...)
2♠+ = primary C (and 5+ C)
In more detail:
2♠ = 6C(421)
2N = 5+C4-D4-H1-S, not 6C(421)
3♣ = 5+C4-D1-H4-S, not 6C(421)
...3♦
......3♥ = 4D3-S*
......3♠ = 1-suited
......3N = 4135
......4♣ = 4045*
......4♦ = 4036, 5 zz
......4♥ = 4036, 6 zz
......4♠+ = 4036, 7 zz
3♦+ = 5+C1-D4-H4-S, not 6C(421)

* Teller shows side a suit in a major first with (440) remaining

[DavidKok]

Thank you, that structure makes sense. Even if this is not what Nyström-Upmark currently play (though it may well be) I would personally be more comfortable with this structure for future analysis. It is somewhat easy to follow and abides with their general principles, and resolves most of my 10* scheme confusion. I haven't had time yet to go through the deals again with the new interpretation of 10* (presumably including (6421) and (5521) only) but that approach looks internally consistent to me and likely to pattern out quite low.

[sieong]

On that board, based on the notes, my interpretation is that the following rules are in effect:

• Zoom limit is 5N (shortness is known, relay at 3 level)
  
• NUMB is in effect, and so is FC.
  
• Possible shapes at this point are 4=1=3=5 (FC 0), 4=0=4=5 (FC 1), 4=0=3=6 (FC 2)
  
  • 4=1=2=6 is ruled out because of 10* scheme
    
  • No 5-5 since they show higher 5-card suit first
    
  • 4=0=4=5 is included because they show higher 4-card suit first for (5440) shapes.

Based on these, I think the responses meant

• 3N: 4=1=3=5, 5-6zz
  
• 4♣: 4=1=3=5, 7zz (FC 0 compared to next shape of 4=0=4=5)
  
• 4♦: 4=0=4=5, 5-6zz
  
• 4♥: 4=0=4=5, 7zz (FC 1 compared to next shape of 4=0=3=6)
  
• 4♠: 4=0=3=6, 5zz (zooming at this point, range is smaller than 5zz, so min step is an exact count)

Fwiw, the auctions from 2019 look strange, and made me wonder if they were experimenting at the time. There may be more hands from older competitions that one can look at from the vugraph project. I believe Nystrom-Bertheau and Cullin-Upmark were both playing AEC.

[DavidKok]

Thank you, that explains that board! I'll look into older games by the two pairs you mentioned soon.

I'm still also thoroughly confused by the relay over 1M-2♣. For convenience I've linked all their games I could find that relayed this way, along with a brief summary (I've included the Mulberry 4♦ signoff).

• 1♥: Board 21, 1♥-2♣; 2♥-2♠; 3♣-4♦; 4♥, was a 7 ZZ 3=5=3=2.
  
• 1♥: Board 2, 1♥-2♣; 3♠-4♣; 4♦ and then there was a misunderstanding, was an 8 ZZ 4=5=3=1.
  
• 1♠: Board 14, 1♠-2♣; 2♦-2♥; 3♥-4♦; 4♥-5♦, was a 6 ZZ 5=4=3=1.
  
• 1♠: Board 14, 1♠-2♣; 2NT-3♣; 3♦-3♥; 3NT-6♠, was a 9 ZZ 5=1=3=4.
  
• 1♠: Board 24, 1♠-2♣; 3♠-4♦; 4♥-4♠, was a 9 ZZ 6=3=3=1.

I expected the 1M-2♣ relay with min 2 step principle to look like:

• 2♦ = Minimum (5-7 ZZ), shortage.
  
• 2♥ = Minimum (5-7 ZZ), no short.
  
• 2♠ = Maximum (8-10 ZZ), no short.
  
• 2NT = Maximum, 10* scheme.
  
• 3♣ = Maximum, high short.
  
• 3♦ = Maximum, middle short.
  
• 3♥ = Maximum, low short, low suit.
  
• 3♠ = Maximum, low short, high suit.
  
• 3NT = Maximum, low short, single-suited.

The fourth and fifth deal on my list don't conform to my scheme, the 2NT rebid in deal four is not a 10* scheme and the 3♠ bid is not the other major with club shortage. I'm not sure what's going on here. I remark that deal number 3 does conform to my scheme, as the steps in the minimum unbalanced branch are one below those in the maximum unbalanced branch thanks to the min 2 step principle.

[DavidKok]

 sieong, on 2024-January-18, 00:03, said:

Fwiw, the auctions from 2019 look strange, and made me wonder if they were experimenting at the time. There may be more hands from older competitions that one can look at from the vugraph project. I believe Nystrom-Bertheau and Cullin-Upmark were both playing AEC.

I failed to search properly on vugraph last time(s), but with a bit of effort more boards by Nyström-Upmark were readily available going back as far as 2011 (and perhaps even older). I've gone through the set back to 2017 but no older years yet, and I think I have a better grasp of the structure now. Thank you everybody for your insights! The 10* scheme indeed does not apply to hands with a void, so it is specifically (6421) or (5521) hand types. In addition in many sequences some (5521)'s can be ruled out as the higher ranking suit is always the anchor suit of those, so e.g. a 10* hand with primary clubs is always 6♣(421), and a 10* hand with primary hearts and secondary spades is always 4♠6♥(21).

Based on this the scheme is almost exactly as I posted before, with the understanding of the 10* scheme above and one modification. The last few bids are supposed to read:

• 3♠ = Maximum, low short, high suit, not a special distribution so exactly (5431).
  
• 3NT = Maximum, low short, high suit, special distribution (5530 or 6430).
  
• 4♣ = Maximum, single suited, low short, zoom to exact shape so (6331)
  
• 4♦ = Maximum, single suited, low short, zoom to 'exact' shape so (7321) - ZZ relay does not clarify the 3-card suit and 2-card suit.
  
• 4♥ = Maximum, single suited, low short, zoom to exact shape so (7330), zoom to showing exact ZZ so 8-9ZZ
  
• 4♠ = (7330) low short, 10 ZZ
  
• 4NT = (7330) low short, 11 ZZ, zoom to scanning for aces and kings so even number of aces.

I'll omit the rest of the exact honour scan (it's even/odd aces, then even/odd kings, then scanning suits in a complicated manner).
This seems to agree with most of the boards I've been able to find with this start, thought there are a few notable exceptions. My best guess is they started modifying the scheme at some point. This is most obvious when comparing this board 24 (from 2019) with this board 14 (from 2018), where a different rebid is chosen with the same shape and ZZ (in fact, perhaps this board from 2018 was the motivation for the change?). At first glance it seems uncomfortable to me that 6331 patterns out past 3NT sometimes (with a minimum it'll be a step lower, but with a middle short it'll be a step higher again).
Also this board 27 from 2017 has me somewhat confused. 2♦ minimum with shortage, 2NT high short not 10*, but what are the rebids over 3♣? 3♦ shows a club suit, 3♥ a 5=1=4=3, 3♠ a special distribution (so 5=0=5=3 or 6=0=4=3), but what is 3NT? I'd expect that to be the start of the single-suited scheme, i.e. 6=1=3=3. Maybe they are immediately zooming on the 5=0=5=3 vs 6=0=4=3 hands? Somewhat related is the 2020 auction on board 22: 1NT is unbal with primary spades but not 10*, or bal with 5(+) hearts. 2♥ is unbal primary spades, min 2 step (so in particular: a minimum, i.e. 4-5 ZZ by a passed hand I think, unbal primary spades and not high shortage), 2NT is middle shortage (3♣+ would be low shortage), then over 3♣ I think it would have been:

• 3♦ = Low suit
  
• 3♥ = High suit, generic 5431 so exactly 5=4=1=3
  
• 3♠ = High suit, special distribution so 5=5=0=3 or 6=4=0=3
  
• 3NT = Exactly 6=3=1=3, start of the zoom to exact shape showing of single suited spades hands with diamond shortage.

As can be seen on the board this is not the case. I'm not sure why, or what the continuations mean. Perhaps they split the 6=4=0=3 out immediately and the single-suiters start at 4♣, perhaps there's a 3♠/3NT swap (ZZ relay does this in a number of situations), or perhaps I'm way off. Possibly I'm already misunderstanding the previous round of bidding, as I also can't explain board 13 (I'd expect 2NT to show a maximum passed unbal hand with primary spades and heart shortage).

Also I don't understand the scheme over the no-shortness bids yet - I thought it would be structured the same way as the 5(+)♠ no-shortness scheme over 1♣-1♥; 1♠-1NT; 2♣-2♦; 2♥ but without the strength split - that is:

• Step 1 = (5332)
  
• Step 2 = 5M4m22
  
• Step 3 = 5M4oM22
  
• Step 4+ = Single-suited (so (6322) or (7222)), zooming to exact shape and then on to exact ZZ(?)

However, this seems to conflict with board 21 (where step 2 contains a 5332) and board 2 (where step 3 is single-suited, but there was no zoom to exact shape).

[sieong]

I am not able to explain the no shortness hands - I suspect that this has been changed at some point. Here are my guesses for the other sequences.

Fwiw, Cullin-Michielsen in the Swedish trial should be playing a variation of AEC as well. I think Mats Nisland posted in Bridgewinners that they are competing this weekend.

 DavidKok, on 2024-January-19, 08:35, said:

Based on this the scheme is almost exactly as I posted before, with the understanding of the 10* scheme above and one modification. The last few bids are supposed to read:

• 3♠ = Maximum, low short, high suit, not a special distribution so exactly (5431).
  
• 3NT = Maximum, low short, high suit, special distribution (5530 or 6430).
  
• 4♣ = Maximum, single suited, low short, zoom to exact shape so (6331)
  
• 4♦ = Maximum, single suited, low short, zoom to 'exact' shape so (7321) - ZZ relay does not clarify the 3-card suit and 2-card suit.
  
• 4♥ = Maximum, single suited, low short, zoom to exact shape so (7330), zoom to showing exact ZZ so 8-9ZZ
  
• 4♠ = (7330) low short, 10 ZZ
  
• 4NT = (7330) low short, 11 ZZ, zoom to scanning for aces and kings so even number of aces.

I believe this may be dependent on whether responder zooms with the shortness or if the shortness is already known and asker makes a relay, and subject to both section 1.3 (exceptions from exceptions) and section 1.4 (zoom limit).

For example, after 1♥-2♣, I think exceptions 1.3.1 and 1.4.3 apply when the responses zoom above 3NT, meaning

• 3♠ = 8-11zz, 4=5=3=1
  
• 3NT = 8-11zz, 4=6=3=0 or 5=5=3=0
  
• 4♣ = 8-11zz, 3=6=3=1 (NUMB applies, but no FC since 4♥ is the limit - if the opening was 1♠, this would have been 8-9zz, 3=6=3=1).
  
• 4♦ = 8-11zz, 2*=7=2*=1
  
• 4♥ = 8-11zz, 3=7=3=0 (in particular, no zoom from here)

Note also that over a 5+ minor (e.g., after 1♣ opening), the responses are different (pg 9).

 DavidKok, on 2024-January-19, 08:35, said:

Also this board 27 from 2017 has me somewhat confused. 2♦ minimum with shortage, 2NT high short not 10*, but what are the rebids over 3♣? 3♦ shows a club suit, 3♥ a 5=1=4=3, 3♠ a special distribution (so 5=0=5=3 or 6=0=4=3), but what is 3NT? I'd expect that to be the start of the single-suited scheme, i.e. 6=1=3=3. Maybe they are immediately zooming on the 5=0=5=3 vs 6=0=4=3 hands?

With shortness known (2N) and a 3-level relay, I think zoom limit is now 5N. Together with NUMB and FC, I think the meaning of the bids were:

• 3♦ = 4+♣
  
• 3♥ = 5=1=4=3
  
• 3♠ = 5=0=5=3 or 6=0=4=3
  
• 3NT = 5-6zz, 6=1=3=3 (FC0)
  
• 4♣ = 7zz, 6=1=3=3 (FC0)
  
• 4♦ = 5-6zz, 7=1=2*=2* (FC1)
  
• 4♥ = 7zz, 7=1=2*=2* (FC1)
  
• 4♠ = zooming with 7=0=3=3

 DavidKok, on 2024-January-19, 08:35, said:

Somewhat related is the 2020 auction on board 22: 1NT is unbal with primary spades but not 10*, or bal with 5(+) hearts. 2♥ is unbal primary spades, min 2 step (so in particular: a minimum, i.e. 4-5 ZZ by a passed hand I think, unbal primary spades and not high shortage), 2NT is middle shortage (3♣+ would be low shortage), then over 3♣ I think it would have been:

• 3♦ = Low suit
  
• 3♥ = High suit, generic 5431 so exactly 5=4=1=3
  
• 3♠ = High suit, special distribution so 5=5=0=3 or 6=4=0=3
  
• 3NT = Exactly 6=3=1=3, start of the zoom to exact shape showing of single suited spades hands with diamond shortage.

As can be seen on the board this is not the case. I'm not sure why, or what the continuations mean. Perhaps they split the 6=4=0=3 out immediately and the single-suiters start at 4♣, perhaps there's a 3♠/3NT swap (ZZ relay does this in a number of situations), or perhaps I'm way off. Possibly I'm already misunderstanding the previous round of bidding, as I also can't explain board 13 (I'd expect 2NT to show a maximum passed unbal hand with primary spades and heart shortage).

I am not sure about this one either. I suspect that the passed hand sequences are different. It is also possible that maybe FC is in use here (so both 3♥ and 3♠ are 5431, and 3N are the hands with void). The rules of FC are not totally clear in the notes.

 DavidKok, on 2024-January-19, 08:35, said:

Also I don't understand the scheme over the no-shortness bids yet - I thought it would be structured the same way as the 5(+)♠ no-shortness scheme over 1♣-1♥; 1♠-1NT; 2♣-2♦; 2♥ but without the strength split - that is:

• Step 1 = (5332)
  
• Step 2 = 5M4m22
  
• Step 3 = 5M4oM22
  
• Step 4+ = Single-suited (so (6322) or (7222)), zooming to exact shape and then on to exact ZZ(?)

However, this seems to conflict with board 21 (where step 2 contains a 5332) and board 2 (where step 3 is single-suited, but there was no zoom to exact shape).

Agree that Board 21 is not explainable using the notes. Board 2 could be explained by the 5* scheme section - the shapes are shown differently compared to the specific sequences after 1♣.

[DavidKok]

Thank you, the structure is starting to make more sense to me. At the same time the FC rules and exceptions on exceptions are giving me a lot of difficulty, and I'd love to construct a simplified (albeit less efficient) version similar to the 'skeleton' relay I introduced in the first post, as a benchmark. I've put looking for more example games on hold for the moment, though I intend to return to it later.

My next step would be to write my own overview of this simplified relay, perhaps in pdf form. This thread has been immensely helpful in clearing up which hand types go where, information that is not explained as clearly as I'd like in the pdf's linked earlier. I think a clear summary along with some example sequences could be very helpful. After that I might try to program the shape showing part of my simplified relay structure into dealer to get some statistical information, e.g. how frequently do we pattern out at a low level, which hands systemically end up high, how often do we bypass the optimal spot using my naive version &c. Would there be any interest in such a pdf and/or this information?

[foobar]

 DavidKok, on 2024-January-21, 04:24, said:

Would there be any interest in such a pdf and/or this information?

Yes please .

[straube]

1C-16+, 17+ balanced
1D-bal, 3-suited or 5+D/5+C
.....2H-5S/4H, less than inv
.....2S-6S, less than inv
.....1H
..........1S-xyz
..........1N-xyz
..........2C-5/4+ minors, either way
...............2H-to play
...............2S-4SF
..........2D-good raise
..........2H-bad raise
.....1S
..........1N-11-13 bal or min 1444 or min 1(5m43) or min 04(54)
..........2C-13-15 1444 or max 1((5m43) or max 04(54)
..........2D-5D/5C
...............2H-4SF
...............2S-inv
...............2N-inv, f
....................3C-weak
...............3C-correction
1M-5+
1N-14-16
2C-6C, possible 4M or 4D
.....2D-asking
2D-6D, possible 4M or 4D
.....2H-asking
2M-weak
2N-19+-21 bal

[foobar]

 straube, on 2024-January-23, 18:23, said:

1C-16+, 17+ balanced
1D-bal, 3-suited or 5+D/5+C
.....2H-5S/4H, less than inv
.....2S-6S, less than inv
.....1H
..........1S-xyz
..........1N-xyz
..........2C-5/4+ minors, either way
...............2H-to play
...............2S-4SF
..........2D-good raise
..........2H-bad raise
.....1S
..........1N-11-13 bal or min 1444 or min 1(5m43) or min 04(54)
..........2C-13-15 1444 or max 1((5m43) or max 04(54)
..........2D-5D/5C
...............2H-4SF
...............2S-inv
...............2N-inv, f
....................3C-weak
...............3C-correction
1M-5+
1N-14-16
2C-6C, possible 4M or 4D
.....2D-asking
2D-6D, possible 4M or 4D
.....2H-asking
2M-weak
2N-19+-21 bal

Believe this is the IMPrecision opening structure, right (maybe except for 2N)? This one (my preference) differs in 1♦(can't have 55m, 54m) and 2m (can't have 4CM).

[straube]

 foobar, on 2024-January-24, 10:24, said:

Believe this is the IMPrecision opening structure, right (maybe except for 2N)? This one (my preference) differs in 1♦(can't have 55m, 54m) and 2m (can't have 4CM).

Yes it is (except for 2N). I included responses and rebids from Imprecision because I think they lend themselves well to a simple Precision structure.

I think opening 4216 with 1D is very bad. Opening 1255 2N with a constructive hand is also bad because there's no room to sift out strength or tolerance for responder's major. 1D-1S, 2D as 5/5 leaves a lot more room.

[foobar]

 straube, on 2024-January-24, 12:34, said:

Yes it is (except for 2N). I included responses and rebids from Imprecision because I think they lend themselves well to a simple Precision structure.

I think opening 4216 with 1D is very bad. Opening 1255 2N with a constructive hand is also bad because there's no room to sift out strength or tolerance for responder's major. 1D-1S, 2D as 5/5 leaves a lot more room.

This may be a religious argument, but I think excluding the more common 54+m hands from 1D trumps the 6m4M hands. Also, the 55+m hands can be folded into 2m, freeing 2N if desired.