[straube]

Here's an example of how Moscito could use reverse relays that spin off. The suits can be arranged to taste and I haven't really looked at what's optimal.

The balanced hands are 5332s, 4333s and 4432s and don't include any 4441s. If I have the math right, responder doesn't need 1C-2N+ as 5440s although there is a little spillover with a few of them (so one might use 3D up or whatever is needed). There is little change for responder. He is usually +1 as is usual for Moscito, but that's the consequence of using immediate 1N+ responses for semipositives. First, how responder might show his hand after 1C-1D, 1H....

1C-1D,
.....1H-
..........1S-bal or H/C or H
...............1N-relays (usually balanced)
....................2C-H or H/C
.........................2D-relays
..............................2H-H at +0
..............................2S-H/C reverser at +1
....................2D+ balanced shapes
..........1N-spades
...............2C-relays
....................2D-S/C at +1
....................2H-S at +0
....................2S-S/D reverser at +1
..........2C-majors or 3-suited short minor
...............2H-3-suited short minor at +1
..........2D-C/D at +1
..........2H-C at +0
..........2S-3-suited short major at +1
..........2N-D bal at +0

Now opener also gets to show his hand and this is where the savings are...

1C-1D, 1H-1S
..........2C-S/C at +0
..........2D-S at -1
..........2H-S/D reverser at +0

1C-1D
..........1S-hearts
...............2C-H/C at +0
...............2D-H at -1
...............2H-H/D reverser at +0
..........1N-majors
...............2D-3-suited, short minor at +0
...............2H-H/S reverser at +0
..........2C-C/D at +0
..........2D-C at -1
..........2H-3-suited, short major at +0
..........2S-D bal at -1

Opener is helped by two things...
.....1) freeing up 1C-1D, 1N and collapsing all balanced hands into 1C-1D, 1H
.....2) being able to temporize with 1C-1D, 1H and then reverse relaying over a 1S rebid.

[hrothgar]

 straube, on 2011-April-22, 22:01, said:

Now opener also gets to show his hand and this is where the savings are...

Silly question: Why precisely, would opener want to reverse the relay after

1♣ - 1♦
1♥ - 1♠

It's not enough to be able to reverse a relay.
You also need to consider WHY you'd want to reverse a relay.

During an earlier post, I suggested that there are a number of reasons why people want to reverse relays.

One reason is limiting range.
Another reason is trying to identify a misfit as quickly as possible should you can safely stop at 3N

Once the relay responder has started to show shape, I use relay breaks to show a misfit

1. If relay responder has show a two suited hand, the relay break shows the other two suits
2. If relay responder has only show one suit, the relay break shows shortage (singleton or void) in that suit

I really think that you are putting the cart before the horse by jumping straight in a designing relay structures without giving any serious consideration to what the ***** you're trying to show and why you want to do so.

[straube]

 hrothgar, on 2011-April-23, 07:35, said:

Silly question: Why precisely, would opener want to reverse the relay after

1♣ - 1♦
1♥ - 1♠

It's not enough to be able to reverse a relay.
You also need to consider WHY you'd want to reverse a relay.

The design goal is to increase the chances of having the balanced hand be captain when partner is unbalanced. As of now, your Moscito version doesn't do this when opener rebids 1N.

hrothgar said:

I am a firm believer balanced hands should ask rather than show.
I like that the auction

1♣ - 1♦
1N

is as rare as possible.

Imo, you are straining not to use the valuable 1C-1D, 1N sequence. The fact that the hand you offered for a maximum NT rebid...

♠ KT94
♥ QJ8
♦ AQJ6
♣ QJ3

has zero chance for upgrade makes me think how very aware you must be of the balanced hand principle.

By putting responder's balanced shapes (which by frequency are the majority) into a 1S rebid one....

1) preserves the NT declaration for whenever bal opposite bal which is very common
2) retains for opener the ability to be captain for more unbalanced opposite unbalanced hands. 1C-1D, 1H and now responder can throw off a few more unbalanced shapes before opener commits to doing so.
3) lets opener show all of his unbalanced hands more economically in total. He gets two bites at the apple. 1C-1D, 1S and higher and 1C-1D, 1H-1S, 2C and higher. Right now your 1C-1D, 1N is excluded from showing a shapely hand (excepting the 4441s). On average he goes from +1 to +0 and is able to show the 5440s; I think that's huge.

I understand that limiting hands and abbreviated auctions are also design goals. I suppose, too, that you could have opener reverse relay with only the limited hands that are unbalanced if you want and just retain captaincy with the bigger hands as you do now. You'd still be +0 instead of +1.

I feel like it's often so often difficult to tell whether the hands are fitting until complete shape has been shown. I would rather retain captaincy for the balanced hand.

I don't think I have more to add. Thanks for the interesting discussion.

[akhare]

 straube, on 2011-April-23, 08:24, said:

The design goal is to increase the chances of having the balanced hand be captain when partner is unbalanced. As of now, your Moscito version doesn't do this when opener rebids 1N.

It seems that you have made up your mind that that the Balanced Hand Principle is inviolable overarching design goal.

Given that, it's difficult to see how anyone can make headway regarding other potential use of the 1N reverse relay (like with min balanced hands)...

[Edit]
That probably came across as too harsh.

Basically, I was trying to see if it was possible to objectively evaluate the tradeoffs of low information auctions like 1♣ - 1♦ - 1N - 3N or 1♣ - 1♦ - 1N - <transfer> - 4M vs. other design goals (balanced hand principle for example).

On a secondary note, it may be possible to use relay breaks if the reverse relay is used to show a minimum hand.

[Free]

 akhare, on 2011-April-22, 16:17, said:

On the flip side, isn't it true that balanced hands resolve lower? Most balanced hands should be able to complete their QP / PCB / DCB scan by 4N or so.

Contrast this with a 5530 long legged shape resolution that resolves at say 3N. The QP ask / DCB / PCB scan likely won't end before the 5-level..

In my experience balanced hands resolve higher than the most common unbalanced hands. The best example is 5431 which is at 3♦. Balanced hands usually resolve at 3♥/♠/NT. I must admit that in many cases you can skip the last step and start scans for 12- or 14-card-distributions (which is what I do very often to save space). Then you start scanning lower, obviously, but the scans still take you sky high very quickly since 4 suits need to be scanned. It's a huge difference, and I really hate to relay out a balanced hand completely because you end up too high too often imo. I disagree that balanced hands are able to complete their QP / PCB / DCB scan by 4NT, it's much much higher. 12/14-card scans are usually acceptable though. Perhaps my scanning methods aren't efficient enough.

The scans take most space when a step isn't skipped. For example, if you skip a step during the first scan to show 1 or 2 top honours, then it's most efficient if you have 1-2 tophonours in each suit. Having a 5530 this change is much higher than having a balanced hand.

[akhare]

Actually, this discussion has helped me realize that it may be possible to combine both design goals (BHP) and potential relay breaks with two balanced hands opposite each other.

In our scheme, the sequence 1♣ - 1♦ (GF) - 1♥ ® - 1♠, 1♠ shows (♥+♣) OR balanced.

The sequence proceeds:

1♣ - 1♦ (GF) - 1♥ ® - 1♠ - 1N:

.....2♣: ♥+♣
.....2♦+ <Balanced module, including 5332 hands>

In retrospect, it's blindingly obvious to flip the two, i.e.:

.....2♣: Bal hands with 4/5 card major
.....2♦: Bal hands without 4/5 card major
.....2♥+: ♥+♣, reversed etc.

Now, over 2♣ / 2♦, if one is especially ambitious about BPH, there's an opportunity to use 2♠+ to spin off single suited reverse relay that couldn't be shown before .

Alternatively, 2♠+ can be used as a relay break showing a minimum balanced hand, ostensibly expressing doubt about playing in 3N.

Thoughts?

[straube]

 akhare, on 2011-April-24, 13:51, said:

Actually, this discussion has helped me realize that it may be possible to combine both design goals (BHP) and potential relay breaks with two balanced hands opposite each other.

In our scheme, the sequence 1♣ - 1♦ (GF) - 1♥ ® - 1♠, 1♠ shows (♥+♣) OR balanced.

The sequence proceeds:

1♣ - 1♦ (GF) - 1♥ ® - 1♠ - 1N:

.....2♣: ♥+♣
.....2♦+ <Balanced module, including 5332 hands>

In retrospect, it's blindingly obvious to flip the two, i.e.:

.....2♣: Bal hands with 4/5 card major
.....2♦: Bal hands without 4/5 card major
.....2♥+: ♥+♣, reversed etc.

Now, over 2♣ / 2♦, if one is especially ambitious about BPH, there's an opportunity to use 2♠+ to spin off single suited reverse relay that couldn't be shown before .

Alternatively, 2♠+ can be used as a relay break showing a minimum balanced hand, ostensibly expressing doubt about playing in 3N.

Thoughts?

I think this is a very good idea. It will lead to more abbreviated auctions for minimum opposite minimum.

[straube]

Richard, I tallied results for 100 balanced hands to see how many of 15 pts or higher would have exactly 8 QPs. I excluded 14 pt hands because I assumed that you would likely open those 1N when they only have 8 QPs. Only 2 out of the 100 were 15+ with exactly 8. In contrast, 1C-1D, 1N as majors or 3-suited short minor constitutes about 8% of total patterns.

What do you think of akhare's idea? Seem like it would address some of the same concerns that your 1C-1D, 1N presently does.

[straube]

With akhare's idea, I'm thinking that after...

1C-1D, 1H-1S, 1N-2C

....opener will most likely want to break relay with minimums with a four-card major

2H to show four hearts and possibly four spades
.....3N to show responder has a minimum with four spades
.....4H to play

2S to show four spades and deny four hearts
.....3N minimum with four hearts
.....4S to play

These sorts of auctions should be very common. The don't even lose much when opener has 4/4 in the majors because responder will necessarily be playing spade contracts when this has been revealed.

Other uses are less clear because opener will be declaring NT and will want to hide his pattern. For example, we could use 2C-3H to show 5-3-2-3 minimum but that defeats the primary purpose of not giving away unnecessary information.

[hrothgar]

 straube, on 2011-April-24, 14:47, said:

Richard, I tallied results for 100 balanced hands to see how many of 15 pts or higher would have exactly 8 QPs. I excluded 14 pt hands because I assumed that you would likely open those 1N when they only have 8 QPs. Only 2 out of the 100 were 15+ with exactly 8. In contrast, 1C-1D, 1N as majors or 3-suited short minor constitutes about 8% of total patterns.

What do you think of akhare's idea? Seem like it would address some of the same concerns that your 1C-1D, 1N presently does.

Question: why you ran your simulation, did you constrain it such that responder has a Game forcing hand?

Comment:

I'm fairly happy with my current relay scheme.
It works well and its easy to remember.

If I were to introduce a chance, I would probably

1. Have all balanced hands relay with 1♥
2. Use 1♣ - 1♦ - 1N to show some annoying hand type (solid 6+ card suit or some such)
3. Use 1♣ - 1♦ - 1♥ - 1N - 2♦ to SHOW shape using our normal response structure after NT openings

[akhare]

 straube, on 2011-April-24, 16:12, said:

Other uses are less clear because opener will be declaring NT and will want to hide his pattern. For example, we could use 2C-3H to show 5-3-2-3 minimum but that defeats the primary purpose of not giving away unnecessary information.

Other extensions that remain in the relay scheme might be:

2N: Min 5♥332
3♣: Min 4333 with a minor
3♦ - 3♠: Min 5♠332

Alternatively for less revealing auction:

2N: Min 5♥332
3♣ Min 5♠332
3♦: Min 4♣333
3♥: Min 4♦333

[straube]

 hrothgar, on 2011-April-24, 16:27, said:

Question: why you ran your simulation, did you constrain it such that responder has a Game forcing hand?

I constrained responder to have 10+ hcps

[straube]

Here's how our bidding went last night. 4H showed 14 QPs. After that, there wasn't much point to asking for specific cards because I couldn't have the right cards for a grand.

straube ♠AJ965♥AK42♦AJ10♣A

akhare ♠KQ8♥973♦KQ86♣Q75

1♣ P 1♦ P
1NT P 2♣ P
3♦ P 3♥ P (3D shows 5431 specific)
4♥ P 6♠ P
P P
6♠ E NS: 0 EW: 0

[Free]

 hrothgar, on 2011-April-24, 16:27, said:

I'm fairly happy with my current relay scheme.
It works well and its easy to remember.

True, it's elegant, flexible and pretty easy on the memory. On the other hand it's probably possible to improve it, that's why I posted my ideas. Imo the importance of balanced hand asking is much higher than generally accepted. It's also not included in the current scheme, the philosophy at the moment is to ask with extra's and show with 9-11 QP although we like to break the rule with balanced hands.

Note that I'd open your example hand with 1NT 11+-14HCP 6+QP rather than 1♣ 15+HCP 9+QP. Yes, it does have 15HCP, but it doesn't have the necessary 9 QP. I definitely want to avoid problems when partner starts asking.

 hrothgar, on 2011-April-24, 16:27, said:

If I were to introduce a chance, I would probably

1. Have all balanced hands relay with 1♥
2. Use 1♣ - 1♦ - 1N to show some annoying hand type (solid 6+ card suit or some such)
3. Use 1♣ - 1♦ - 1♥ - 1N - 2♦ to SHOW shape using our normal response structure after NT openings

This (1. and 2.) is like the next step after my suggestion. Instead of using 1NT for all balanced hands, lower it to 1♥ and use it as a relay. It suits the "balanced hand asks" principle much better and you can still relay out all hands at the same height (and even use the same scheme).
Also, opener saves space when he decides to show his hand, because he can't be balanced anymore. So all steps from 2♣ and higher can be lowered, or you can use 1NT to show any 5440.

I like this and I think it's much more efficient than the current scheme. Moreover it still allows opener the flexibility to ask with unbalanced shapes anyway. The biggest drawback (imo) of the current relayscheme is that it wrongsides NT contracts when responder is balanced as well. If you're prepared to lose the perfect symmetry, then an easy solution can be found. Otherwise it will be hard, unless you use the basic symmetric relayscheme +1 step (the one before semipositives were in play). If I remember correctly, it would be something like:
1♠ = 4+H unbal (doesn't include 1=4=4=4)
1NT = 4+S unbal (doesn't include 4=1=4=4)
2♣ = bal
2♦ = 6+D / 3-suited with both minors
2♥ = 6+C
2♠+ = D+C
Opener can use the same scheme, and replace 2♣ with any 5440.

[straube]

 Free, on 2011-April-27, 04:06, said:

True, it's elegant, flexible and pretty easy on the memory. On the other hand it's probably possible to improve it, that's why I posted my ideas. Imo the importance of balanced hand asking is much higher than generally accepted. It's also not included in the current scheme, the philosophy at the moment is to ask with extra's and show with 9-11 QP although we like to break the rule with balanced hands.

Note that I'd open your example hand with 1NT 11+-14HCP 6+QP rather than 1♣ 15+HCP 9+QP. Yes, it does have 15HCP, but it doesn't have the necessary 9 QP. I definitely want to avoid problems when partner starts asking.


This (1. and 2.) is like the next step after my suggestion. Instead of using 1NT for all balanced hands, lower it to 1♥ and use it as a relay. It suits the "balanced hand asks" principle much better and you can still relay out all hands at the same height (and even use the same scheme).
Also, opener saves space when he decides to show his hand, because he can't be balanced anymore. So all steps from 2♣ and higher can be lowered, or you can use 1NT to show any 5440.

I like this and I think it's much more efficient than the current scheme. Moreover it still allows opener the flexibility to ask with unbalanced shapes anyway. The biggest drawback (imo) of the current relayscheme is that it wrongsides NT contracts when responder is balanced as well. If you're prepared to lose the perfect symmetry, then an easy solution can be found. Otherwise it will be hard, unless you use the basic symmetric relayscheme +1 step (the one before semipositives were in play). If I remember correctly, it would be something like:
1♠ = 4+H unbal (doesn't include 1=4=4=4)
1NT = 4+S unbal (doesn't include 4=1=4=4)
2♣ = bal
2♦ = 6+D / 3-suited with both minors
2♥ = 6+C
2♠+ = D+C
Opener can use the same scheme, and replace 2♣ with any 5440.

I think you're close, but the next step is letting 1C-1D, 1H-1S show (mostly) balanced shapes. This means that opener can divide his shapely hands between 1C-1D, 1S+ and 1C-1D, 1H-1S, 2C+
That's how opener gets to be +0 and it's how opener gets to declare when balanced opposite balanced.

[Free]

 straube, on 2011-April-27, 07:27, said:

I think you're close, but the next step is letting 1C-1D, 1H-1S show (mostly) balanced shapes. This means that opener can divide his shapely hands between 1C-1D, 1S+ and 1C-1D, 1H-1S, 2C+
That's how opener gets to be +0 and it's how opener gets to declare when balanced opposite balanced.

Good point about using 1♠ as balanced. Lowering responder's balanced hands will also result in better slam bidding in the balanced-balanced case, because that's where we need space the most. However, dividing opener's shapely hands is dangerous because you're never sure responder will bid 1♠. I'd rather use it to show 5422's and 6322's for example, where opener decided to relay but now changes his mind.

Btw, I just realized, in the scheme I gave, you don't need 2♣ as 5440 for opener. These hands are already included in the 1♠ and 2♦ responses. So basically we can lower the 2♦ and higher calls 1 step.

[straube]

 Free, on 2011-April-27, 07:48, said:

However, dividing opener's shapely hands is dangerous because you're never sure responder will bid 1♠.

But see, it isn't dangerous. If the bidding goes...

1C-1D, 1H-1N+

then responder is showing a shapely hand. You would usually prefer that responder show shapely hands whether opener is balanced or not.

[Crunch3nt]

I am very late to this topic, and nothing revolutionary to add but for the record here is our latest thinking:

There are three major conflicting principles when designing the shape showing part of your relay system:

1) You want the hand having its shape shown to be dummy
2) You want an unbalanced hand to show rather than ask
3) You want the strong hand to do the asking, rather than a weak hand

I consider number 1 to be most important, and 2 to be next important & 3 least important (but still important).

After 1C strong, 1D art GF: we play that 1H is relay, and 1S+ is reverse relay. The only only shapes shown by opener in the reverse relay are limited because of space to just those with exactly a singleton (not void) thus all 5431, 6421, 4441, 6331, 5521 shapes only. 76 shapes in total. We currently play these as unlimited. All shapes come out by 3H allowing a range probe at the 3 level.

The 1H relay includes all balanced hands, 6322s, 5422s, 7xxx, 74xx, 5530 and 65xx single suiters.

The crucial 1NT reverse relay is a little overloaded in stealing the NT declarership in that it shows S&H or S&C - but as we all know, its all about trade-offs...

[straube]

 Crunch3nt, on 2011-May-11, 20:49, said:

I am very late to this topic, and nothing revolutionary to add but for the record here is our latest thinking:

There are three major conflicting principles when designing the shape showing part of your relay system:

1) You want the hand having its shape shown to be dummy
2) You want an unbalanced hand to show rather than ask
3) You want the strong hand to do the asking, rather than a weak hand

I consider number 1 to be most important, and 2 to be next important & 3 least important (but still important).

After 1C strong, 1D art GF: we play that 1H is relay, and 1S+ is reverse relay. The only only shapes shown by opener in the reverse relay are limited because of space to just those with exactly a singleton (not void) thus all 5431, 6421, 4441, 6331, 5521 shapes only. 76 shapes in total. We currently play these as unlimited. All shapes come out by 3H (except 5 x 6421s) allowing a range probe at the 3 level.

The 1H relay includes all balanced hands, 6322s, 5422s, 7xxx, 74xx, 5530 and 65xx single suiters.

The crucial 1NT reverse relay is a little overloaded in stealing the NT declarership in that it shows S&H or S&C - but as we all know, its all about trade-offs...

Hi Michael,

Ours gives 1) a lot of consideration, but 2) has the most.

What's your whole structure? I've sometimes regretted that our 5422s and 6322s are not included with the balanced hands and yours appears to do this.

After 1C-1D, 1H-1S you have the opportunity to spin off opener's voids and more shapely hands...which is especially useful if responder's 1S includes all of his balanced hands. Is that what you do? If not, I think you might look at it because you don't want hands with voids doing the asking.

[Crunch3nt]

 straube, on 2011-May-11, 22:33, said:

Hi Michael,

Ours gives 1) a lot of consideration, but 2) has the most.

What's your whole structure? I've sometimes regretted that our 5422s and 6322s are not included with the balanced hands and yours appears to do this.

After 1C-1D, 1H-1S you have the opportunity to spin off opener's voids and more shapely hands...which is especially useful if responder's 1S includes all of his balanced hands. Is that what you do? If not, I think you might look at it because you don't want hands with voids doing the asking.

According to Roy Hughes analysis, having the known hand as dummy rather than declarer is worth on average 1/3rd of a trick per hand. That is a serious negative to overcome, and why 1) is the most important.

At this stage, I am still playing a 1D response to 1C as 8+ Balanced GF, not any GF. 1H is the negative (0-4 unbalanced with a major, 0-7 balanced or unbalanced no major) and the rest are unbalanced semi-positive or stronger with a major or GF unbalanced no major as we discussed when you were in NZ. I still think it is more important to show shape immediately and sort your range out later.

After the auction 1C-1D; 1H, responder goes into exact shape and opener always relays. Relay breaks by opener are undefined at this stage - something to think about!

The reverse relay structure is new and unique. It is not symmetric and is a pain to remember. I can email if you wish.