[straube]

Leaving aside for the moment the 5422 pattern, standard symmetric resolves 2-suited patterns thusly...

3D-5431
3H-6421
3S-6430
3N-7420
etc-7420 with extra for lower short or 7411 any with higher short

I'm fuzzy on what standard is for the 3N and higher, but it doesn't matter that much for the subject I'm posting.

The relative percentages are...

3D-5431.....64%
3H-6421.....23%
3S-6430......7%
3N-7420......2%
etc-.............4%

Others have noted the big drop off here and maybe have come up with alternatives. Probably many like that 3D allows for a 3S ask for a fragment stopper.

The other thing is that 3N by asker is almost universally played as a sign off. I think I've seen some discussion of this as well and maybe have invented ways of
preserving a relay step when the telling hand bids 3S.

So I'm wondering if there are other ways to organize these step, perhaps involving strength information. Where min and max refer to maybe the difference of 3 relay points or instead 2 controls, first try for me...

3D-5431 minimum................40%?
3H-6421 or 6430..................30%?
.....3S-asks strength first
..........3N-min
..........4C-6430 max
..........etc-6421 max
3S-5431 maximum...............24%
.....3N-asks
3N-7420 or 7411 minimum...4%?
4C-7420 max
etc-7411

I think others have worked on this. Curious what's been done or can be done.

[Free]

My first thought was "Scary!" but when you do the calculations it seems very efficient. The biggest issue is when you're at +1 or +2 with the relays, then this scheme suddenly becomes a nightmare (but you can still revert to standard in that case).

Interesting idea!

[rbforster]

It can be worth restructuring your relay answers in order to make better use of either the possible relay breaks or to allow for cheaper relays. In general, this is an ad hoc process and gives you less symmetric relays, but it can be done if you want to optimize things. A simple example which is a little more general is flipping the last two responses to full shape before zooming into strength. Consider that some symmetric relays put 7411 with 5422 under the "no shortness" branch for two suiters:

Two suiter symmetric relays:
2D low suit longer
2H equal lengths (55xx, 66xx, 4441s)
2S+ high suit longer

2S high shortness
2N no shortness (5422/7411)
3C+ low shortness

So then instead of the "normal" probability ordering for shape under the "no shortness" branch, ie

3D 5422 any
3H 7411 min
3S 7411 min+1
3N 7411 min+2, etc

you instead flip them,

3D 7411 any
3H 5422 min
3S 5422 min+1
3N 5422 min+2, etc

because after all you almost always want to ask strength after hearing a 5422 any strength response at such a low level as 3D. This swap resolves 5422 strength one step lower than the first version, since you save the relay ask step and zoom into strength. It resolves 7411 strength one step higher than before, but that's probably ok because 1) 7411 is much rarer, and 2) more extreme shapes can go higher more safely in their relays, and sometimes but not in this case 3) 3N is very unlikely to be the final contract with extreme shape so you can agree to use 3N as an ask over a 3S response sometimes.

In essence, this is still a probability optimization, but rather than optimize for probability at which level you resolve full shape, you optimize for the level to resolve both full shape and first strength response.

[rbforster]

Another example of reordering shape asks was something I did with relays resolving a balanced responding hand. TOSR doesn't tell you which suits are which until the final shape is known (crash-style, ie 4432 with 44 in same color/rank/shape shown first, then resolves which doubleton on second ask), so you can't very well relay break without missing possible major fits. Better it seemed to me was to ask for major shape first, and then resolve minor shape second, ie 44 majors, or 43 majors, or 42 majors, and then go from there. I wrote this up but never really used the relay breaks options after it.

http://www.its.calte...relays%20v3.doc

[straube]

3C-5422 or 7411
.....3D-asks
..........3H-5422, less than super-accept
..........3S-7411
...............3N-asks
..........etc-5422 super-accepting

3D-5431

3H-6421

3S-7420
.....3N-asks

3N-6430, less than super-accept

etc-6430, super-accepting

[rbforster]

Looks good, especially structuring the 3N relay asks for the 7411 shape.

[straube]

Thanks Rob and thanks Fred. One opportunity I've missed is that 3D-3H, 3S-3N is still a sign off. Ideally it should ask for something. It's possible that there's no solution for this without changing the meaning of 3D itself.