[akhare]

awm, on Feb 26 2010, 09:00 PM, said:

There's also the argument that it's not just frequency but importance.

The 0-4 hands are usually partscore deals, so there are not so many IMPs to be won or lost.

The GF hands are quite often game/slam decisions, so there are a very large number of IMPs riding on those hands.

Also, in my experience, the opps almost invariably interfere on the 0-4 hands and in a way make the continuations over 1♣ - 1♠ easier (assuming responder even gets to bid 1♠).

One counter argument in favour of 1♣ - 1♦ (all positives) is that opener's reverse relay (with a limited hand) regains the lost step. If responder has a limited hand also, the auction can be terminated without revealing too much information about the hands.

Of course, structures like TOSR give the ability to allow opener to reverse relay over the positive response, but as I recall, it requires a special "10 shape reverse relay"...

[straube]

akhare said:

One counter argument in favour of 1♣ - 1♦ (all positives) is that opener's reverse relay (with a limited hand) regains the lost step. If responder has a limited hand also, the auction can be terminated without revealing too much information about the hands.

Reverse relays do not gain a step. They are still +1.

Reverse relays would regain a step if opener relayed all of his hands automatically starting with 1H.
Obviously, one would be reluctant to do so.

[rbforster]

I generally agree with Adam.

awm, on Feb 26 2010, 01:06 AM, said:

(2) The range of strength is much wider for the positive response (8+ instead of 5-7), and after you resolve shape you will need to resolve strength and honor location, which means you actually need a lot more space after the positive response than on the semi-positive auction.

I think this is the big reason to go with +0 for positives and +2 for semipositives, rather than +1 for both. Frequency issues aside, you need the space more on the positives to check for slam (safely), while the narrow range of the semipositive means you can afford to be higher since your combined strength is already pretty well known.

I made this tradeoff in designing my strong club, where semipositives don't relay and get a little less space/precision, in exchange for getting lower relay resolution for positives.

awm, on Feb 26 2010, 01:06 AM, said:

It seems like your current scheme does fine over 1♣-1♥. However, this particular unobstructed sequence is not very common, as opponents like to bid over 1♣ especially when they have some values.

Certainly it can't hurt to work out good methods if you can remember them. Generally I thought you usually got interference from weaker-but-not-broke distributional hands, rather than better ones. I've had good results just waiting with sounder hands while the precision people flounder to a poor partial after 1C-1D.

[rbforster]

Cascade, on Feb 26 2010, 04:46 PM, said:

Edit whoops I see you were discussing 15+ here are those numbers :

         15+ High Card Point    16+ High Card Point
HCP  Frequency Cumulative  Frequency   Cumulative
0  0.009913845	0.009913845	0.011278423	0.011278423
1  0.019908795	0.029822641	0.022437932	0.033716354
2  0.031543206	0.061365847	0.035111572	0.068827926
3  0.05236386  0.113729707	0.057458297	0.126286223
4  0.074091549	0.187821257	0.079893517	0.206179741

Am I reading this correctly that P(0-4|15+) = 19%, but P(0-4|16+) = 21%? Shouldn't 0-4 be getting less likely as opener gets stronger?

Edit: nevermind, too tired when I was reading that.

[Cascade]

Rob F, on Feb 27 2010, 10:25 PM, said:

Cascade, on Feb 26 2010, 04:46 PM, said:

Edit whoops I see you were discussing 15+ here are those numbers :

         15+ High Card Point    16+ High Card Point
HCP  Frequency Cumulative  Frequency   Cumulative
0  0.009913845	0.009913845	0.011278423	0.011278423
1  0.019908795	0.029822641	0.022437932	0.033716354
2  0.031543206	0.061365847	0.035111572	0.068827926
3  0.05236386  0.113729707	0.057458297	0.126286223
4  0.074091549	0.187821257	0.079893517	0.206179741

Am I reading this correctly that P(0-4|15+) = 19%, but P(0-4|16+) = 21%? Shouldn't 0-4 be getting less likely as opener gets stronger?

Why?

If partner has more then we rate to have less.

[straube]

Well, if one wanted to be +0 for the positives and +2 for the semipositives but wanted to play immediate semipositives and double negatives, one could...

1D-GF with many shapes, including balanced
.....1H-opener relays
.....1S+ reverse relays hoping to catch responder with the balanced hands
1H-any semipositive
.....1S-relay
1S-any negative
1N+ GF with distributional shapes

The difficulty I see with this is finding part scores with the semipositives.

I'm also not liking how semipositives and double negatives mess with the declaration. A lot to be said for 1C-1D 0-7.

[akhare]

Rob F, on Feb 27 2010, 04:21 AM, said:

I made this tradeoff in designing my strong club, where semipositives don't relay and get a little less space/precision, in exchange for getting lower relay resolution for positives.

IMO, not relaying over SP responses is the right approach.

The only reservation I have about the positive response structure at 1♠+ is opener doesn't have the ability to effectively reverse relay.

As I see it, having to relay out responder's shape every single time with 9-11 QPs is a losing proposition in the long run. Even assuming that the contract is right sided every single time, resolving dummy's exact shape and strength with zero slam interest will likely confer some advantage to the opening leader.

For example, with say opener holding a 15-18ish hand and responder holding a 6-8 QPs 5♠??? I would rather have the auction go 1♣ - 1♦ (GF) - 1N (reverse relay, 9-11 QPs, bal hand) - 2♥ (transfer breaking relay -- no slam interest either) - 2♠ - 3N - 4♠, concealing dummy's exact shape.

This brings up an interesting question -- what should relay breaks / reverse relays after a positive response show?

[straube]

akhare said:

IMO, not relaying over SP responses is the right approach.

I think you're almost endplayed into relaying SP responses if you play a system like Moscito. Say it goes 1C-2S (semipositive with spades), you can decide to show your own suit (nf), but if you want to establish a GF your only option is to relay.

I think I was a little misled with the Moscito structure because their semipositives are 3-5 QPs and I've seen claims that those hands represent 60% or so of all responses. I think that claim may be true but only because many of the hands that are GF hands are categorized as semipositives. It's not necessarily a bad thing to group semipositives and minimum positives together except that opener can never drop responder. For example their 1C-2S (which is nf) is problematic as responder can have KQJxxxx xx x Qxx or better.

I like avoiding such distortions. Let the GF hands GF and the semipositives not.
If this is done, however, the SP frequency drops to (what did we find out?) low 30% of hands and the GF hands are like 49%.

So if it doesn't make sense to relay the SPs then why relay them, especially at the expense of the GF hands...which are more frequent and more important, especially at teams.

[akhare]

straube, on Feb 27 2010, 05:00 PM, said:

akhare said:

IMO, not relaying over SP responses is the right approach.

I think you're almost endplayed into relaying SP responses if you play a system like Moscito. Say it goes 1C-2S (semipositive with spades), you can decide to show your own suit (nf), but if you want to establish a GF your only option is to relay.

I disagree -- even if responder holds a magical hand like KQJxxxx xx x Qxx, it's just a question of reorganizing the response structure so that responder gets more than one bite at the apple.

For example, one possible scheme might be that a 2♦ response shows a SP response with single suited hand in a major.

Over this opener can bid 2♥ (not forward going in ♥s, maybe forward going in ♠) or 2♠ (the other way around). In the case where responder holds the improbable hand that's good enough for game, but doesn't have the necessary number of QPs, responder can always rebid over the attempted signoff.

Also, if you forgo Marton's 2N - 3N positive responses (which hardly ever come up to be useful), there's plenty of room to unwind freak SP hands (but I won't hold my breath waiting for one to come up any time soon).

If opener has a forcing hand, you can play 2N as a puppet to 3♣ for a potential sign off at the 3 level and immediate bids as forcing (or the other way around).

BTW, I am curious about why Moscito forces relaying of SP hands -- are you referring to 15+ ♣ systems or any system with SP responses?

[PrecisionL]

Quote

Thanks for replying.  I've reason to believe that the 0-4 hcp hands are on the order of 20% of the responding hands to 1C.  One reason is that the bridge encyclopedia gives the odds of a random 13 cards having 0-4 pts as about 9%.  Given the restriction that one hand holds 15+, the odds increase.  Also, Atul and I looked at about 100 hands and the 0-4 pt hand came up something like 30% of the time.    Obviously that's a small sample size, but I'm still thinking that 0-4 will be about 20%. 

That's really my argument for 1C-1H as negative.  Should I destroy 20% of the auctions for my 1C (granted that these are my least important 20%)?  I'm still not sure.

But I think I agree with you in liking positives better.

Good guess / approximation!!!

My spread sheet shows 0-4 hcp response to a strong club of 16 or more hcp is 20.6%.

I also prefer strong positives. Keylime and I have a complete system which is Major Centric for bidding over 1♣ - 1♦ (0-7 hcp)! For those that use a 1♥ response for 8-11, the frequency is 65% whereas the 12+ other positive responses are only 12.5%. Is this the tail wagging the dog?

[straube]

Hi Atul, BBO wouldn't let me talk to you when I was kibitzing.

I don't think a hand like KQJxxxx xx x Qxx is that unusual. My point was that there's a lot of hands that don't satisfy Moscito's 6QP positive...hands that include KKQJ or KQQQJ or AKJ or other hands with stiff K or stiff Q (that aren't counted for QPs in Moscito) that intend on GF.

But we both agree that one can organize the response structure for a second bite at the apple. It just costs to do so.

Look at (for instance) playing that 1C-2H shows spades. This lets opener declare 2S which solves the problem of KQJxxxx xx x Qxx (which raises to 4S) but it also costs 2S as a relay bid. Now 2N has to relay and we've lost a step.

Or look at the 1C-2D you proposed to show a single-suited major. Imagine I have...
A x AKQJx AQxxxx. I am now screwed. Partner has used all of the space up through 2D and I don't even have knowledge as to which major he has.

2H-p/c? No.
2S-interest in hearts? No.
2N-Lebensohl? No. Use as relay? Not enough space.
3C-forcing? Will partner raise with three clubs? Will he show his diamond fragment (leaving me in the dark about his major? or will he show his major? He needs to show his spades if my hand is Axx void AKQx AKxxxx but I might prefer a club raise if he has hearts. He can't know what to do.

So my point is that the more space responder has taken up with his semipositive, the more we ought to already know about his hand and the more important relays are. The designers of Moscito knew this of course and designed the semipositives so that they could effectively be relayed. The less space responder has taken, the less he's shown about his hand and the better chance we have for interactive bidding (both show) or a combination (opener relays with S1 and shows minimums with other bids).

akhare said:

BTW, I am curious about why Moscito forces relaying of SP hands -- are you referring to 15+♣  systems or any system with SP responses?

I'm more acquainted with Moscito than others and was using it as my example.

I actually prefer to relay the semipositives, but I think I wouldn't mind if they lost 2 steps so that I saved space for the positives. If we played 1D=0-7, then I would look at naturalish bidding (a la Meckwell) as well as TOSR

[akhare]

straube, on Feb 28 2010, 01:08 AM, said:

Look at (for instance) playing that 1C-2H shows spades. This lets opener declare 2S which solves the problem of KQJxxxx xx x Qxx (which raises to 4S) but it also costs 2S as a relay bid. Now 2N has to relay and we've lost a step.

This is the basic fundamental disagreement. Basically, there are no relays over SP hands (and Rob F. has a similar approach also).

IMO, as long as the SP structure allows both opener and responder have more than one bite at the apple, it's sufficient. Basically, responder needs to be able to show a hand that's a little too good despite the 3-5 QP response and opener needs something that forces a rebid and / or ostensibly sets up a GF.

It's true that you may be able to resolve responder's shape with pinpoint accuracy and you may occasionally miss a slam that depends on the magical fit.

However, IMO, not relaying over SP hands is taking the practical percentage approach to the problem...

[straube]

akhare, on Feb 28 2010, 12:39 PM, said:

straube, on Feb 28 2010, 01:08 AM, said:

Look at (for instance) playing that 1C-2H shows spades.  This lets opener declare 2S which solves the problem of KQJxxxx xx x Qxx (which raises to 4S) but it also costs 2S as a relay bid.  Now 2N has to relay and we've lost a step.

This is the basic fundamental disagreement. Basically, there are no relays over SP hands (and Rob F. has a similar approach also).

IMO, as long as the SP structure allows both opener and responder have more than one bite at the apple, it's sufficient. Basically, responder needs to be able to show a hand that's a little too good despite the 3-5 QP response and opener needs something that forces a rebid and / or ostensibly sets up a GF.

It's true that you may be able to resolve responder's shape with pinpoint accuracy and you may occasionally miss a slam that depends on the magical fit.

However, IMO, not relaying over SP hands is taking the practical percentage approach to the problem...

Then why was Moscito's semipositive response structure organized so that responder's hand can be relayed? How can it make sense to opener to start describing his own hand (in a forcing way) after responder has shown say 5S/4H with a 1C-2S bid?

I was interested in knowing how you would handle the hand I gave you after 1C-2D (showing a major). I gave that as an example of why relays are necessary and why something like 1C-2D (showing a major) doesn't carry enough information to be effectively relayed. What would your continuations over 1C-2D be and how would you handle that hand?

[akhare]

straube, on Feb 28 2010, 01:34 PM, said:

Then why was Moscito's semipositive response structure organized so that responder's hand can be relayed?

I was interested in knowing how you would handle the hand I gave you after 1C-2D (showing a major). I gave that as an example of why relays are necessary and why something like 1C-2D (showing a major) doesn't carry enough information to be effectively relayed. What would your continuations over 1C-2D be and how would you handle that hand?

Well, that's really a question for Martson, but just because it's possible relay SP hand doesn't mean that it's right to do so (and in fact I would argue that it's better to eschew relays altogether over SP hands).

Handling the example hand you posted is rather easy:

Opener:

XX AQXX AQXX KJX

Responder:

KQJxxxx xx x Qxx

Say after:

1♣ - 2♦ (single suited in a major) - 2♠ (not forwarding going opposite most ♠ SP hands, accept for ♥ hands)

Now responder can autosplinter in ♦s to show the unusual hand and opener signs off.

The bottom line is that there always will be magic hands that lead to improbable slams opposite the perfect hand. However, I would rather play something that isn't perfect but gets us to playable spot most of the time than something that aims for the pie in the sky and risks ending up too high more often than not.

[straube]

I meant this hand after 1C-2D (showing a major)
A x AKQJx AQxxxx
How would you continue if 2H is p/c, 2S shows interest in H and 2N is a puppet to 3C?

I can't ask Marston, but I'd say that the answer to

straube said:

Then why was Moscito's semipositive response structure organized so that responder's hand can be relayed?

is that relays are a most reliable method of finding a fit and that responder's first semipositive response is already the answer to opener's 1C relay and that continuing to relay is the most effective way of finding fit and mesh before 3N has been passed.

Let's suppose 2S shows a 5S/4D semipositive. It wouldn't make sense after this for opener to start describing his own shape in a forcing manner. It would be changing directions midstream and it would be totally inefficient. Probably after such a sequence, it would make sense for 2N to be an offer to play, 3C to be a relay and 3D and 3H an offer to play those contracts. 3S would be invitational.

akhare said:

The bottom line is that there always will be magic hands that lead to improbable slams opposite the perfect hand. However, I would rather play something that isn't perfect but gets us to playable spot most of the time than something that aims for the pie in the sky and risks ending up too high more often than not.

I'm not focused on magic hands. After a semipositive, I'm more keen on getting to the right games and partials, but the same tools I use for these will help me also to get to sensible slams.

[rbforster]

akhare, on Feb 28 2010, 12:39 PM, said:

straube, on Feb 28 2010, 01:08 AM, said:

Look at (for instance) playing that 1C-2H shows spades.  This lets opener declare 2S which solves the problem of KQJxxxx xx x Qxx (which raises to 4S) but it also costs 2S as a relay bid.  Now 2N has to relay and we've lost a step.

This is the basic fundamental disagreement. Basically, there are no relays over SP hands (and Rob F. has a similar approach also).

Actually I'm currently experimenting with a combination of these approaches. The point here is that sometimes it's hard with SP values to separate a very good major suit (which often makes game) or a very good minor suit (which often makes 3N) from a random SP hand with only 5 cards or 6 poor ones. While most SP hands go through a natural non-relay continuation after 1C-1D, good 1-suited SP hands are shown in the same way as 1-suited GF hands. So for example:

1♣ 1♥ strong; 4+ spades unbal GF or SP with a good 6+ suit
1♠ 2♣ relay; single-suited (could be SP; other responses to 1♠ confirm GF values)

At this point, opener doesn't know whether or not responder has full GF values. If he's minimum, he can bid 2♦ weak relay, and the SP hand will answer 2♥ and natural bidding will continue with no GF established (GF hands will answer 2♠+ and relay out their shape). Alternatively, if opener wants to "super accept" over the potential SP 1-suiter, he can bid 2♥ instead of 2♦ as a strong relay. This works nicely because the shapes are the same for the SP and the GF hands, and you can relay out shape with 2♠+ in the same way, the only difference being that you can't be as sure about responder's total strength when you get to the control asking part.

straube said:

A x AKQJx AQxxxx

This is the kind of hand you can deal with by 2N (5-10, 5/5+ minors), followed by 5m or 4N.

[akhare]

Rob F, on Feb 28 2010, 04:10 PM, said:

Actually I'm currently experimenting with a combination of these approaches.  The point here is that sometimes it's hard with SP values to separate a very good major suit (which often makes game) or a very good minor suit (which often makes 3N) from a random SP hand with only 5 cards or 6 poor ones.  While most SP hands go through a natural non-relay continuation after 1C-1D, good 1-suited SP hands are shown in the same way as 1-suited GF hands.  So for example:

1♣ 1♥ strong; 4+ spades unbal GF or SP with a good 6+ suit
1♠ 2♣ relay; single-suited (could be SP; other responses to 1♠ confirm GF values)

Rob,

Actually I was thinking of taking an idea you had proposed earlier and sort of running away with it.

Here's a strawman, but the idea is to achieve TOSR+0 *and* meld SP responses:

1♦ DN OR GF H+m OR GF ♦
1♥: SP bal OR ♠+m GF OR ♠ GF
      1N: S+♣
      2D: SS ♠
     2H+: S+♦ -> two suited module
1♠: SP majors OR bal GF
      2♣: SP with major (and possibly 4441 GF)
      2D+ -> Bal module
1N: GF Majors or SS ♣
2♣: SP (Single suited major?)
2D: GF SS ♥ -> Single suited module at 2S+
2H+ GF Minors -> Two suited module

Over 1♥, opener has the option of "predicting" the bal SP hand by bidding 1N. Responder can now resume relays hwith 2♣ following the bal module (possibly relay stayman).

[straube]

Rob, your approach works for the major single-suiters (and possibly minor single-suiters) but
1) as you say, when opener superaccepts (showing strength and not necessarily fit I suppose) then responder's strength/controls have to sort out semipositive vs GF strength.
2) I don't see a plan for the semipositives with 5/4 or 6/4 or 5/5 or 7/4 shapes that also are semipositives by definition but plan to GF all the time or GF vs reasonable fits.

RobF said:

straube said:

A x AKQJx AQxxxx

This is the kind of hand you can deal with by 2N (5-10, 5/5+ minors), followed by 5m or 4N.

I don't understand. I gave this hand as an example problem hand after Atul's 1C-2D (showing an unspecified major]

[rbforster]

straube, on Feb 28 2010, 05:18 PM, said:

2) I don't see a plan for the semipositives with 5/4 or 6/4 or 5/5 or 7/4 shapes that also are semipositives by definition but plan to GF all the time or GF vs reasonable fits.

Those are handled in with all the normal semipositives (1C-1D) in my system, where you would then commence invitational auctions looking for major fits, etc. You could treat a 7/4 as just 1 suited I suppose, but most of the SP's for me go through 1C-1D. The idea is to establish a GF and show a suit early in case they interfere, and if this means you've got a 1-suited SP hand and have to overbid it a little in competition, well, that's a rare occurrence and maybe it works out .

Quote

RobF said:

straube said:

A x AKQJx AQxxxx

This is the kind of hand you can deal with by 2N (5-10, 5/5+ minors), followed by 5m or 4N.

I don't understand. I gave this hand as an example problem hand after Atul's 1C-2D (showing an unspecified major]

My point was that if you anticipate problems on some big minor 2-suiters, you can solve them by opening 2N (5/5+ minors, weak or GF). This is in contrast to opening them 1C and dealing with the frequent 2M/3M overcalls.

[straube]

RobF said:

QUOTE (straube @ Feb 28 2010, 05:18 PM)
2) I don't see a plan for the semipositives with 5/4 or 6/4 or 5/5 or 7/4 shapes that also are semipositives by definition but plan to GF all the time or GF vs reasonable fits.

Those are handled in with all the normal semipositives (1C-1D) in my system, where you would then commence invitational auctions looking for major fits, etc. You could treat a 7/4 as just 1 suited I suppose, but most of the SP's for me go through 1C-1D. The idea is to establish a GF and show a suit early in case they interfere, and if this means you've got a 1-suited SP hand and have to overbid it a little in competition, well, that's a rare occurrence and maybe it works out .

This is workable. I don't see anything wrong with straining a bit on the shapely hands as long as partner is on the joke. I also think your placing the semipositive 1-suiters in with the GF hands (at least the majors) is playable though I personally would not choose that.

I think there's a tremendous difference bidding naturally (not relaying) over 1C-1D (0-7) and bidding naturally (forcing) and not relaying over a semipositive that has already started to unwind its distribution.

I don't care for opening 2N on the giant minor-suited hand. That's definitely give-up-science and partner can't even pass with a misfit and some values.

Apart from that, opener would be in a nonideal position after 1C-2D (an unspecified major) with AKx x Axxxx AKQx. Wouldn't he like to know if responder is 6-3-1-3?