[foobar]

@kwiktrix had posted about an alternative to Symmetric relay on BW:

https://bridgewinner...292355#c1292355
https://bridgewinner...s-interference/

From reading the posts, the premise is definitely interesting, but it's a little difficult to tease out the details of the relay scheme given that it's interwoven with the semi-positive and 1♠ DN. Also, it appears that the 1♠ DN response may also contain some balanced GF hands "1♠: (1) DN without a 6+m or 7+M 1suiter; (2 - Advanced) Balanced 9+-11HCP, denying 4controls)". A minor quibble about "By re-arranging the response schedule, this frees up the 1♦ response for game forcing use. While this concept has been used in a non-relay context, it has never been applied to any flavor of Symmetric Relay." is that Moscito did use 1♦ for a GF response and employed symmetric relays after that.

It will be great to have a discussion on the Charron relay scheme assuming the following:

• Can the relay still function assuming a 16+ ♣ with 1♦ as 0-7? Perhaps, one alternative is to use 1♦ as some GF or DN (with 1♣ - 1♦ - 1♥ - 1♠ showing the DN).
  
• Can such a relay scheme handle *some* reverse relays over a balanced GF response (presumably 1♠)?
  
• Can the same relay scheme be repurposed over an unbalanced opening with 5+CM?

[awm]

Suppose we compare IMPrecision structure vs. Charon over 1♠, since we both use 2♣ as a relay.

For IMPrecision:

5422 hands resolve at 3♦
5431 hands resolve at 3♥
6421 hands resolve at 3♠
6430 hands resolve at 3nt
5521 hands resolve at 3♠
5530 hands resolve at 3nt
6331 hands resolve at 3♥, 3♠, and 3nt (depending on shortness)
6322 hands resolve at 3♥, 3♠, and 3nt (depending on shortness)
5440 hands resolve at 3♥
7321 hands resolve at 3♥, 3♠, and 3nt (depending on shortness)

For Charon, it looks like:

5422 hands resolve at 3♥ (lose one step)
5431 hands resolve at 3♦ (win one step)
6421 hands resolve at 3♥ (win one step)
6430 hands resolve at 3♠ (win one step)
5521 hands resolve at 3nt (lose one step)
5530 hands resolve at 4♣ (lose one step)
6331 hands resolve at 3nt (lose one step on average)
6322 hands resovle at 3♠ (equal on average)
5440 hands resolve at 3♠ (lose one step)
7321 hands resolve at 4♦ (lose three steps on average)

Charon is also excluding 5332 (which IMPrecision covers). I've left out some of the weirder shapes as they are unlikely to matter much.

I'm not sure what is going on in Charon with the sequence 1♠-2♣-2♦-2♥-2♠ or the sequence 1♠-2♣-2♠-2NT-3♣-3♦-3♥; maybe you're able to save a step on some auctions because these are available? It does mess with the memory of a certain bid showing a certain pattern though.

[foobar]

Here's my attempt to reconstruct the Charron relay scheme over 1♣ - 1♥ (♠ key suit). Note that we are dealing with "key suits" so it can be length or shortness.

The key differences from Symmetric appear in bold.

1♣ - 1♥ - 1♠ - 1N:
...2♣: ♠ + ♣
........2♥: Reversed, then 2N = high short; 3♦+ = Same as Symmetric, except for this order: (5431, 6421, 7321, 6430, 6520)
..............2N: High short, then per 3[diamond below]
........2♠: Switch for "type 3 hands" (5440, 5530, 6511, 7411)
..............3♣: Reverser
..............3♦: Not used
..............3♥+: 5440, 5530, 6511, 7411
........3♣: Special switch for "type 2"; then 3♥+ = 4432, 5422, 5531
........3D: Same as Symmetric

1♣ - 1♥ - 1♠:
...2♣: ♠ + ♦, then as above

1♣ - 1♥:
...2♦: Type 4/5 switch (Single key suit; includes all single suited hands, except 7321; also 4333, 4441)
.......2♠: High unique suit
.......2N: Mid unique suit
.......3♣: Type 5 switch (4441, 4333, 3N=7222)
.......3D: 5332
.......3H: 6322
.......3S: 6331
.......3N: 7330


Some observations:

• The short-legged hands (reversed or non-reversed) are resolved mostly same as Symmetric with 7321 displacing 6430 by one step. The 5422 is an outlier and is resolved at +2
  
• The single suited hands appear to resolve at +1
  
• There are no separate modules for balanced, long legged, short legged and three suited hands
  
• Unlike Symmetric, suits are probably not known for sure until shape is resolved
  
• The space efficiency appears to be mixed, with some common shapes resolving higher
  
• The relay over the 1M opening is different from the above

[kwiktrix]

 foobar, on 2024-September-21, 15:23, said:

Some observations:

• The short-legged hands (reversed or non-reversed) are resolved mostly same as Symmetric with 7321 displacing 6430 by one step. The 5422 is an outlier and is resolved at +2
  
• The single suited hands appear to resolve at +1
  
• There are no separate modules for balanced, long legged, short legged and three suited hands
  
• Unlike Symmetric, suits are probably not known for sure until shape is resolved
  
• The space efficiency appears to be mixed, with some common shapes resolving higher
  
• The relay over the 1M opening is different from the above

Atul - thank you for introducing Charron Relay to BBO Forums.

As you have said, it uses keysuits (a suit with non-duplicated suit length) that could potentially be short. But apart from 1suiters (for which there is an available relay break sequence in Charron - see below), a Relay System has one objective: pattern out the responder’s exact shape. How suits are used to reach this objective is irrelevant, provided that the exact shape is derived in the end. By using the keysuit concept, it is mathematically more efficient, and uses fewer steps. This might not be obvious because many auctions end at or near the same level as KK or Symmetric. But then Charron frees up every direct response to 1♣ above 1♠ to show semi positive or very shapely junk hands.

Since my BridgeWinners post, I’ve rearranged the groupings slightly - your idea of Group numbering is sensible so I'll comply. Note that Group 1,2 and 3 follow the same "track" initially by showing a primary and secondary keysuit. Both Group 1 and Group 2 then show reverses if applicable, then Group 2 "switches" tracks (they have "even" remaining suit lengths) with 3♣. Group 3 switches tracks immediately after defining the primary and secondary keysuits with 2♠.

Groups 4 and 5 only define a primary keysuit initially (going through 1N shows the minor suit "equivalent"), then immediately switches tracks with 2♦. Then bids of 2♠, 2N or "direct to pattern" identify where the secondary keysuit is located in a 1suiter. A bid of 3♣ identifies a Group 5 hand with only a primary keysuit.

Group 1 (4 unique suits) are the same as short-legged and includes (in order of frequency): 5431, 6421, 7321, 6430, 6520, (7420 as well if 4♦ is included)

Group 2 (Balanced) has been revised. The 5332 pattern replaces 5521 and includes (in order of frequency): 4432, 5332, 5422. All of these pattern out at or below 3N)

Group 3 (2/3suiters) has also been revised with the 5521 pattern replacing the 7411. Includes (in order of frequency): 5521, 5440, 5530, 6511

Group 4 (1suiters) has also been revised. 5332 has moved to balanced and replaced by 7411. Now includes (in order of frequency): 6322, 6331, 7330, 7411, 8221

Group 5 (1KS patterns) remains unchanged. Includes (in order of frequency): 4333, 4441, 7222

One point that seems to be missed - Charron is completely consistent in its relay methodology, unlike KK and other symmetrics. For example, note how different the short legged and balanced relay methods are in KK. Or 7321 which is inordinately complex.

To your comments:

• Four suit hands (short legged) are virtually indentical with KK, with the advantage that the 1♣ opener is declaring more often. KK mishandles 7321. 5422 can be resolved at 3♥ - it's simply at 3N now as I optimize rightsidedness.
  
• Incorrect - for example, witness this auction from KK (Answers to 1suiter Quiz #4). ♠KJT864 ♥975 ♦K8 ♣Q7. 1♣-1♥ (showing spades): 1♠-3♣ (showing spade 1suiter, equal short (2 low doubletons or 7222): 3♦-3♠ (6322). Contrast this with Charron: 1♣-1♥ (showing NOT hearts): 1♠-2♦ (spade 1suiter, or 1KS pattern with unique spade suit): 2♥-2♠ (spade 1suiter with HIGH secondary keysuit): 2N-3♦ (6322). Resolved at 3♦, not 3♠. More importantly, since Charron identifies the likelihood of a 1suiter at 2♦ (unlike KK which is at the 3level), Charron allows for a relay-break to identify suit strength.
  
  Here’s how it works:
  

  ...2♦
  2♥ = Symmetric shape continuations 
  2♠ = <Relay Break> Do you have length? I have A, K or both
       2N = other control + more ——> 3♣
            3♦ = QJ(T)
            3♥ = Q(T)
            3♠ = JT
            3N = J
       3♣ = none or no length ——> 3♦
            3♥ = 1444
            3♠ = 4333
            3N = length, no honors
       3♦ = neither + QJ(T)
       3♥ = neither + Q(T)
       3♠ = neither + JT
       3N = neither + J 
  2N = <Relay Break> Do you have length? I have neither A or K, but Q(JT)
       3♣ = none or no length ——> 3♦
            3♥ = 1444
            3♠ = 4333
            3N = length, no controls
       3♦ = AK(JT)
       3♥ = A(JT)  
       3♠ = KJ(T)
       3N = K(T)
  

  
  
• I'm not sure the point here? Essentially Charron only needs 2 Modules - 1 using 2keysuits initially, and the other using 1keysuit which goes through 2♦. All Modules follow the same logic.
  
• True
  
• Untrue. The space efficiency is mathematically superior. Some shapes resolve higher to allow more options for semi-positive auctions.
  
• True - but it follows the same keysuit logic. The primary keysuit has already been defined.

[foobar]

kwiktrix, on 2024-September-25, 13:42, said:

By using the keysuit concept, it is mathematically more efficient, and uses fewer steps. This might not be obvious because many auctions end at or near the same level as KK or Symmetric. But then Charron frees up every direct response to 1♣ above 1♠ to show semi positive or very shapely junk hands.

Thanks for the detailed response. Can you elaborate a little more on the mathematical efficiency part? Specifically, is this a function of stipulating a 1♠ DN response, or something else? In other words, can we improve on Symmetric with the proviso that all responses except 1♣ - 1♦ are GF?

kwiktrix, on 2024-September-25, 13:42, said:

Group 2 (Balanced) has been revised. The 5332 pattern replaces 5521 and includes (in order of frequency): 4432, 5332, 5422. All of these pattern out at or below 3N)

I don't have the complete scheme mapped out, but this seems like a great revision. IMO, clubbing all balanced hands in a single response makes a lot of sense in any relay scheme, and 4333 is excluded for obvious reasons given the design constraints.

kwiktrix, on 2024-September-25, 13:42, said:

One point that seems to be missed - Charron is completely consistent in its relay methodology, unlike KK and other symmetrics. For example, note how different the short legged and balanced relay methods are in KK. Or 7321 which is inordinately complex.

IMO, the differences between SL and LL are fairly logical since they follow the pattern of showing shortness, then resolving the specific shape. I will unconditionally concede the point about balanced relays because they are indeed all over the place. For the 7321 hands, can't recall how KK-relay bids it, but a typical Symmetric relay with ♠ might go: 1♣ - 1♥ (♠) - 1♠ ® - 2♠ (single suited; high short) - 2N ® - 3♥ / 3♠. Note that single suited schemes that also include 5♠332 typically compress 73(21). The assertion is that the (21) is seldom relevant, and I don't have data to debate it one way or the other.

kwiktrix, on 2024-September-25, 13:42, said:

[*]Four suit hands (short legged) are virtually indentical with KK, with the advantage that the 1♣ opener is declaring more often. KK mishandles 7321. 5422 can be resolved at 3♥ - it's simply at 3N now as I optimize rightsidedness.

How would something like TOSR compare in terms of right siding?

kwiktrix, on 2024-September-25, 13:42, said:

[*]Incorrect - for example, witness this auction from KK (Answers to 1suiter Quiz #4). ♠KJT864 ♥975 ♦K8 ♣Q7. 1♣-1♥ (showing spades): 1♠-3♣ (showing spade 1suiter, equal short (2 low doubletons or 7222): 3♦-3♠ (6322). Contrast this with Charron: 1♣-1♥ (showing NOT hearts): 1♠-2♦ (spade 1suiter, or 1KS pattern with unique spade suit): 2♥-2♠ (spade 1suiter with HIGH secondary keysuit): 2N-3♦ (6322). Resolved at 3♦, not 3♠. More importantly, since Charron identifies the likelihood of a 1suiter at 2♦ (unlike KK which is at the 3level), Charron allows for a relay-break to identify suit strength.

This looks really interesting, and I will have to read the complete document to fully understand how it works (hint, hint ).

kwiktrix, on 2024-September-25, 13:42, said:

[*] I'm not sure the point here? Essentially Charron only needs 2 Modules - 1 using 2keysuits initially, and the other using 1keysuit which goes through 2♦. All Modules follow the same logic.

It was intended to be a compliment .

kwiktrix, on 2024-September-25, 13:42, said:

[*] Untrue. The space efficiency is mathematically superior. Some shapes resolve higher to allow more options for semi-positive auctions.

Please see the question related to DN and SPs above. I am genuinely curious about how things will look in a system that mandates no SPs.

kwiktrix, on 2024-September-25, 13:42, said:

[*] True - but it follows the same keysuit logic. The primary keysuit has already been defined.

Fair enough, should be comparable in terms of mnemonic ease. FWIW, Tarzan Club (played by WCs Mueler-De Wijs) repurposes almost exactly the same relay scheme over 1M opening by using 1♠ / 1N as the GFR. They do open 5-5+M hands with 1♥ so that everything can fit in.

[kwiktrix]

foobar, on 2024-September-25, 21:37, said:

Thanks for the detailed response. Can you elaborate a little more on the mathematical efficiency part? Specifically, is this a function of stipulating a 1♠ DN response, or something else? In other words, can we improve on Symmetric with the proviso that all responses except 1♣ - 1♦ are GF?

Right now, 1♣ - 1♥ is unconditionally GF. 1♣ - 1♦ also contains all BAL semi-positives (SP). SP defined as "NOT Junk (JK)" and "NOT GF", and JK is defined as 0-6 UNLESS there is a 75% chance or more of an entry (KK, A, KQ suited). It should be noted that JK comes in BAL/SemiBAL and more shapely patterns. The 1♠ response is strictly for the former. To rework Charron such that it follows KK (loses access to 1♦) loses the benefits of heart rightsiding.

Let me give it some thought - diamonds would no longer be specifically declared as in Charron (more rightsiding downside).

Quote

IMO, the differences between SL and LL are fairly logical since they follow the pattern of showing shortness, then resolving the specific shape. I will unconditionally concede the point about balanced relays because they are indeed all over the place. For the 7321 hands, can't recall how KK-relay bids it, but a typical Symmetric relay with ♠ might go: 1♣ - 1♥ (♠) - 1♠ ® - 2♠ (single suited; high short) - 2N ® - 3♥ / 3♠. Note that single suited schemes that also include 5♠332 typically compress 73(21). The assertion is that the (21) is seldom relevant, and I don't have data to debate it one way or the other.

7321 is more common than 6430...

Quote

How would something like TOSR compare in terms of right siding?

As I understand it, TOSR follows this sequencing: 1♦=0-8, 1♥=♠, 1♠=BAL, 1N=♥, 2♣=♦, 2♦=♣, 2♥+=♣+♦. Charron follows this sequencing: 1♦=♥, 1♥=NOT♥, then 1N=♣, 2♣=♦, zooming to both majors (after 1♦) or both minors (after 1♥).

Quote

This looks really interesting, and I will have to read the complete document to fully understand how it works (hint, hint ).

One benefit that I did not mention to the <Relay Break> after 2♦ is that the exact hand shape for 1keysuit hands resolves in the same manner as if it were relayed out for pattern. So what the 2♠ and 2N <Relay Breaks> do is clarify whether responder is 1suited or 1keysuited - and if 1suited, the exact quality of the suit for notrump purposes.

Quote

Fair enough, should be comparable in terms of mnemonic ease. FWIW, Tarzan Club (played by WCs Mueler-De Wijs) repurposes almost exactly the same relay scheme over 1M opening by using 1♠ / 1N as the GFR. They do open 5-5+M hands with 1♥ so that everything can fit in.

If I used 1N over 1M, I could do better. Here's a quick concept...

1H = 10-14 UNB 5+h (h longest or equally long, except 5h6any
     1S = ART F1, not GF
          1N =  ART 4s
          2m = 4+m
     1N = ART GF
          2C = diamond 2nd keysuit  ——> 2D
               2H = AVAILABLE as a "switch" - perhaps for some GI hands
              [2S = high short ——> 2N]
               2N = even ——> 3C
                    3D = 2542
                    3H = 1651
                    3S = 1561
                    3N = 1741
               3C = extreme ——> 3D
                    3H = 5620
                    3S = 6520
                    3N = 7321
                    4C = 7420
               3D = 5431
               3H = 6421
               3S = 5521
               3N = 5440
               4C = 5530
          2D = club 2nd keysuit ——> 2H
              [2S = high short ——> 2N]
               2N = even short ——> 3C
                    3D = 2524
                    3H = 1615
                    3S = 1516
                    3N = 1714
               3C = extreme ——> 3D
                    3H = 5620
                    3S = 6520
                    3N = 7321
                    4C = 7420
               3D = 5431
               3H = 6421
               3S = 5521
               3N = 5440
               4C = 5530
          2H = single suited (6+) ——> 2S
              [2N = high 2nd KS ——> 3C]
              [3C = mid 2nd KS ——> 3D]
               3D = 7222 (no short)
               3H = 6322
               3S = 6331
               3N = 7330
               4C = 8221
          spade 2nd keysuit...
         [2S = high short ——> 2N]
          2N = even ——> 3C
               3D = 4522
               3H = 5611
               3S = 6511
               3N = 7411
          3C = extreme ——> 3D
               3H = 5620
               3S = 6520
               3N = 7321
               4C = 7420
          3D = 5431
          3H = 6421
          3S = 5521
          3N = 5440
          4C = 5530