Scream vs IMPrecision vs straw man

[straube]

So I tallied up the first round responses for 163 hands for SCREAM, IMPrecision, and the straw man I posted on the other thread. I would have liked to have included Moscito, but their club is weaker and their SPs are stronger and it wouldn't have been a fair comparison. Had I evaluated 1D as 0-7, this response would have fetched 57%. Definitions are approximate....

 

..........................SCREAM....................IMPrecision........................straw man......................SCREAM................IMPrecision...............straw man

1D.....................GF, bal or M..............DN or 7+ RP......................DN or H, 5+ RP..................36%....................29%..........................28%

1H.....................2-4 RP......................2-6 RP, 4+ S......................2-4 RP..............................41%....................27%..........................41%

1S......................0-1 RP.....................2-6 RP, 5+ D or bal............5+ RP, C/D or bal.............16%....................27%..........................20%

1N.....................5+ RP, H/S................2-6 RP, 5+H, no 4S............5+ RP, S, S/C.....................1%.....................12%..........................4%

2C.....................5+ RP, C/D...............2-6 RP, 5+C unbal, no M....5+ RP, S/D.........................2%......................1%...........................1%

2D.....................5+ RP, 6+C..............2-6 RP, 5+C/4H..................5+ RP, 6+C........................1%......................0%...........................1%

other......................................................................................................................................3%......................4%...........................3%

 

So a few observations....

 

1. I like to see responses to almost any opening taper off in frequency at something resembling a Fibonnaci graph. IMPrecision seems to do this best. In particular, IMPrecision makes the most out of the 1N response. I'm sure btw that with a larger sample size, SCREAM's 1N response would register a few percent.

 

2. One of my initial concerns about IMPrecision was that some of the semipositive/GF hands were overbidding. After all, what is a 2 RP hand with 4H/5C doing bidding 2D already? But if you look at it from a frequency point of view, this sort of thing happens seldom and has chances to win anyway. By far, most of the SPs are divided between 1H, 1S, and 1N. While 1N is forcing to the 2-level, there are many chances at settling in good part scores. Contrast this to SCREAM and straw man which only has one SP for all shapes. Personally, I think I prefer how IMPrecision divides the SPs in comparison to Moscito because responder can show 4+ spades and the contract can still wind up in 1N (I think).

 

3. SCREAM, straw man and Moscito have a common drawback in that SPs and GFs have to be neatly separated by RPs OR run some risk of confusion when RPs are requested and denial cue bidding starts. IMPrecision avoids this issue.

 

4. While I dislike putting the DNs into 1S, I also dislike putting balanced hands into 1S....especially the stronger ones. IMPrecision makes a compromise, putting the weaker balanced hands into 1S and the stronger balanced hands into 1D. Is that the way it works exactly, Adam? I divided them 7+ going into 1D and 2-6 into 1S, but I wonder if the 5+ could go into 1D. I haven't read through the continuations after 1C-1D, 1H yet so perhaps this isn't feasible.

 

5. Putting GF 1H into 1D along with DN's is a bit cross-purposed. I mean that if opener starts to show his pattern, then responder is forced to relay. This means that the weaker hand is captaining the stronger hand and it means (in a sense) that the information that responder had a shapely heart hand is not being utilized. Putting the 7+ RP hands into 1D gives a chance that some good balanced hands can be captain of opener's shapely hand.

 

6. I'm not very wild about concealing hearts, which both SCREAM and straw man do.

[awm]

We put all balanced GF thru 1d. Our 1S is: (1) balanced 5-8 w/o 4S or 5H, includes 5m-(332), 2-5 RP (2) 5+d without 4S or any 5+ side suit, 5+ hcp and 2-6 RP.

 

Also note: 1H includes 5-8 balance with 4S (but not GF). 2C includes 5+/5+ in the minors, but not 5m-(332). 2D includes 5+d/4+h. 6m and semi-bal and GF 4-6R bids 2N+ optionally.

[wclass___]

I like to see responses to almost any opening taper off in frequency at something resembling a Fibonnaci graph. IMPrecision seems to do this best. In particular, IMPrecision makes the most out of the 1N response.

Do you actually know how Fibonnaci numbers work?

It works like this:Frequency =%

 

%1♦=%1♥+%1♠

%1♥=%1♠+%1NT

%1♠=%1NT+%2♣

etc.

 

If you want to get something that would resemble Fibonacci graph it should be more like:

 

1♦ - 41%

1♥ - 25%

1♠ - 16%

1NT - 9%

2♣ - 7%

2♦=2%

 

It is not that simple: this works in only case where we assign each hand type the same level of importance, e.g. balanced 3-count is treated with same importance like some 10-point 5422, which is obviously nonsense. I suggest you take out opponent hands that would overcall 1♣; then recalculate new probabilities of responses. Put 1♥ as 5♠+ or 4♠&unbal+ 7+; 1♠ = balanced 8+ including 5♥; 1N=5♥ 7+ etc. and calculate probabilities. It is somewhat hard to analytically assign each hand type level of importance, so i guess one should just keep it in mind when analysing results.

 

e.g. It is normal that %1♥<%1♠+%1NT, because probability of 1♠ is somewhat artificially risen by fact that balanced hands comes up a lot, while in relay bidding they don't need that much space nor do they great slam potential.

 

Dude, i think you are wasting a lot of time with those 1♣-1♦, If you find it interesting then great, but if you are doing this just to improve system (results), then my advise would be to take 1♦ 0-7 type as best...(or close to that) and improve from there.

[straube]

We put all balanced GF thru 1d. Our 1S is: (1) balanced 5-8 w/o 4S or 5H, includes 5m-(332), 2-5 RP (2) 5+d without 4S or any 5+ side suit, 5+ hcp and 2-6 RP.

 

Also note: 1H includes 5-8 balance with 4S (but not GF). 2C includes 5+/5+ in the minors, but not 5m-(332). 2D includes 5+d/4+h. 6m and semi-bal and GF 4-6R bids 2N+ optionally.

 

 

In this case, I tallied IMPrecision's results wrong and a significant portion (5%?) ought to be moved from 1S and perhaps a few percent ought to be moved from 1H into 1D. Also, I put your 5m/5m into 1S, so move a percent or so into 2C. It becomes an even nicer distribution. Have you tallied responses yourself for it?

 

So I'm guesstimating IMPrecision now...

 

1D-37%

1H-24%

1S-22%

1N-12%

2C-2%

other-4%

[straube]

Do you actually know how Fibonnaci numbers work?

It works like this:Frequency =%

 

%1♦=%1♥+%1♠

%1♥=%1♠+%1NT

%1♠=%1NT+%2♣

etc.

 

If you want to get something that would resemble Fibonacci graph it should be more like:

 

1♦ - 41%

1♥ - 25%

1♠ - 16%

1NT - 9%

2♣ - 7%

2♦=2%

 

It is not that simple: this works in only case where we assign each hand type the same level of importance, e.g. balanced 3-count is treated with same importance like some 10-point 5422, which is obviously nonsense. I suggest you take out opponent hands that would overcall 1♣; then recalculate new probabilities of responses. Put 1♥ as 5♠+ or 4♠&unbal+ 7+; 1♠ = balanced 8+ including 5♥; 1N=5♥ 7+ etc. and calculate probabilities. It is somewhat hard to analytically assign each hand type level of importance, so i guess one should just keep it in mind when analysing results.

 

e.g. It is normal that %1♥<%1♠+%1NT, because probability of 1♠ is somewhat artificially risen by fact that balanced hands comes up a lot, while in relay bidding they don't need that much space nor do they great slam potential.

 

Dude, i think you are wasting a lot of time with those 1♣-1♦, If you find it interesting then great, but if you are doing this just to improve system (results), then my advise would be to take 1♦ 0-7 type as best...(or close to that) and improve from there.

 

Thanks for the Fibonacci breakdown. I haven't seen it in awhile. Fibonacci is just one consideration, but in general (not always) it's nice to see S1 more frequent than S2 and so on. Yes, I think I am wasting time, but this is a hobby, right? And it's not all about winning; it's about playing a beautiful system, too (though the two should go together). I think you are right in that 1D (0-7) is very playable.

[straube]

ok, so putting these side by side. IMPrecision is guesstimated and generic symmetric 1D response is 0-7 and 1H response is all of the S+ hands and 1S is all of the balanced and both minor hands....

 

.....................Fibonacci.............SCREAM...............IMPrecision.................straw man.................generic symmetric 1D=0-7

1D...................41.......................36........................37..............................28.............................57

1H...................25.......................41........................24..............................41.............................14

1S...................16.......................16........................22..............................20.............................20

1N....................9.........................1.........................12...............................4..............................4

other................9.........................6..........................6................................7..............................5

 

Again, not saying Fibonacci is everything...other considerations like which sequences are most important plus one would expect to spend more room than Fibonacci would suggest because one's other goal is to take up space before the opponents do.

[awm]

Corrected IMPrecision results:

 

1♦: 46.6%

1♥: 21.5%

1♠: 16.3%

1N: 7.3%

2♣: 4.0%

2♦: 2.3%

2♥: 0.7%

2♠: 0.6%

Higher: 0.7%

[straube]

Wow.

[akhare]

4. While I dislike putting the DNs into 1S, I also dislike putting balanced hands into 1S....especially the stronger ones.

 

5. Putting GF 1H into 1D along with DN's is a bit cross-purposed. I mean that if opener starts to show his pattern, then responder is forced to relay.

 

6. I'm not very wild about concealing hearts, which both SCREAM and straw man do.

 

It's pretty easy to fix all of the above 1♦ being a two way bid.

 

Note that I am not suggesting that the following is very good, but:

 

For example:

 

1D: DN OR Bal GF OR some S two suited GF hands

1H: H / H+S

1S....

1N....

2C....Minors

2D....Clubs

2H....

2S+...Diamonds

 

I think another way of looking the numbers you posted is compression of information.

 

The strawman unwinds GF hands faster and therefore has higher % numbers for the higher bids.

 

From that perspective, it stands to reason that the 1♦ % distribution for Adam's system (many GF including bal GF + DN) > Moscito 1♦ (almost all GF) > SCREAM 1♦ (some GF) > Strawman 1♦ (few GF+DN). By the same token, the 1♥ response in the strawman (all SP) > Moscito 1♥ (SP without 5CM).

 

 

I tend to prefer the right side of the spectrum, but that's just my subjective opinion...

Edited January 7, 2012 by akhare

[straube]

I've never quite understood Moscito's decision to show semipositives with 5 spades at the 2-level immediately. What's the hurry? They can transfer or compete later. Losing a 4-cd major or 5+ hearts is more problematic imo.

 

So what does IMPrecision do? Well, it shows 5+ hearts similar to Moscito with a 1N response...except that it gets a higher frequency use because awm's 1N is 2-6 RPs and Moscito's is 3-5. It hasn't room to show both 4-cd majors, so it shows the most important one (spades) at 1H. This lets opener rebid 1N which is nf. If responder has a fifth spade, he can always decide to transfer later. OTOH, frequently awm will find a 4-4 or better spade fit immediately. For most of the leftover semipositives, awm responds 1S which allows opener to rebid 1N nf or relay with 2C or.....

 

Dividing the semipositives between 1S and 1N doesn't make sense to me. We will get too high and opener will have difficulty suggesting a 1N contract to play.

 

I'd guesstimate this latest strawman as....

 

1D-41%

1H-14%

1S-27%

1N-13%

etc

[akhare]

I've never quite understood Moscito's decision to show semipositives with 5 spades at the 2-level immediately. What's the hurry? They can transfer or compete later. Losing a 4-cd major or 5+ hearts is more problematic imo.

 

Richard is the best person to comment on it, but it does make the 1♥ response less vulnerable to competition (less compression of information) and there's something nice about blasting to 4♠, say 1♣ - 2♣ - 4♠.

 

For most of the leftover semipositives, awm responds 1S which allows opener to rebid 1N nf or relay with 2C or.....

 

Dividing the semipositives between 1S and 1N doesn't make sense to me. We will get too high and opener will have difficulty suggesting a 1N contract to play.

 

Isn't that a contradiction?

[straube]

No, it's not a contradiction. Your last strawman divides the SPs between 1S and 1N. Adam's groups SPs and light GFs into 1H, 1S, and 1N mostly but also 2C and higher. It would be much better to put all the SPs into 1H than divide them between only 1S and 1N. So I suppose you could revise this into...

 

1D-DN, bal GF, S GF

1H-SPs

1S-GF H

etc...

 

but I haven't done the math.

[dake50]

What if Fibonacci-ing your system begins after a 1st response?

Now you'd have 'pages' to sequence per Fibonacci:

after 1C-1D, 1C-1H, 1D opener, 1H opener, etc.

[hrothgar]

I've never quite understood Moscito's decision to show semipositives with 5 spades at the 2-level immediately. What's the hurry? They can transfer or compete later. Losing a 4-cd major or 5+ hearts is more problematic imo.

 

Just to review, the MOSCITO semi-positive schedule is

 

2♠ = single suited with 6+ Spades

2♥ = single suited with 6+ Hearts

2D = 5+ Spades and 4♦

2♣ = 5+ Spades and (4+ Clubs or 4+ Hearts)

1N = Unbalanced with 5+ Hearts

1♥ = Balanced, unbalanced with no 5 card major, or three suited

 

So...

 

1. The decision to immediately show 5+ Spades doesn't impact our ability to show 5+ Hearts

2. The decision to immediately show 5+ card majors is motivated by two primary factors.

 

(a) Its dangerous to use bids like 1N/2♣/2♦ to show a 4 card major. If you have a misfit you could be in real bad shape (especially since the requirement that step is a relay cuts down on your ability to scramble). Please recall, these aren't game forcing auctions. In a lot of cases, you're going to want to be able to sign off at the two level.

(b) Its nice to be able to blast to 4M ASAP (This is the same reason that the relays over limited openings focus on showing a 5+ card major ASAP).

 

3. If you prefer to immediately show 4 card majors and shove the 5 card majors into 1♥ there's nothing stopping you.

 

FWIW, I think that there might be some merit to having a bid that immediately shows at least 4-4 in the majors (including balanced, two suited, and three suited). Maybe 1NT or 2♣

However, I never came up with a schedule that I liked more than the one listed above.

[straube]

We weren't proposing to show 4-cd majors via 1N, 2C, or 2D. I was suggesting that SPs with spades should respond at the 1-level as with IMPrecision or hidden within SCREAM's ambiguous 1H semipositive. I think being able show 5 spades immediately is nice, but I would rather use the 2-level for other things. I think it's a lot more important to show SPs with five hearts than SPs with five spades.

[hrothgar]

We weren't proposing to show 4-cd majors via 1N, 2C, or 2D. I was suggesting that SPs with spades should respond at the 1-level as with IMPrecision or hidden within SCREAM's ambiguous 1H semipositive. I think being able show 5 spades immediately is nice, but I would rather use the 2-level for other things. I think it's a lot more important to show SPs with five hearts than SPs with five spades.

 

First and foremost, as I already noted we're able to show BOTH semi-positives with 5 Spades AND semi-positives with 5 Hearts so your comment there seems misplaced.

 

Second, you were the one suggesting that it was more important to show hands with a 4 card major than ones with a five card major, so I apologize for assuming that's what you would do with the bids that show a 5 card major. I will, however, note that if you AREN'T using 1NT+ to show these hands then you are, essentially, forced to stick them into the 1H bid which is what we do...

 

I guess my next question is IF you don't want to use 1N < -- > 2♦ to show a semi positive with a 5 card major just what do you think it should be shown? I'd really prefer to see some kind of proposal rather than random, inaccurate complaints

[straube]

I've played Moscito before and I'm aware that Moscito shows both 5-cd major suit semipositives. I'll try to clarify what I meant. I think it's important to show 5+ heart semipositives and not very important to show 5+ spade semipositives. It's my opinion and I know that everyone may have their own opinion about that. My intent in criticizing a design choice in Moscito was not to complain, but to make an observation that I feel to be true and to support what I think to be an improvement. I like very much awm's decision to show 4+ spades with 1H and 5+ hearts with 1N. I can understand why you thought I was suggesting we show 4-cd major at the 2-level, and I wasn't making that suggestion. Adam uses his 2 and 3-level bids for other semipositive/light GF patterns and I think that dovetails perfectly with the rest of his system.

[Free]

I think straube has a point that showing 5♠ is less important than showing 5♥. When the auction becomes competitive, responder can always bid his suit when holding ♠ without raising the level. However, there's more to it than that. 2-suited hands also contain 6-4, 5-5 and 7-4, in which case it's very handy to show 5M immediately and act accordingly afterwards.

 

For example, suppose opps interfere to 3♥. When responder has shown 4+♠ (or "any semipositive"), he can bid 3♠, but is that a 5, 6 or 7 card suit? When he has shown 5+♠ already, it now shows 6+♠. With a 5-5 he can bid 4m (after showing 4+♠ this rather shows 4♠-6m imo).

 

Similar, when responder denied a 5 card M, he can always bid the M at any level (not too high ofcourse) with an unbalanced hand with 4M and longer m. Opener knows what to expect. Say I have a max SP with 4♥-6♣. After 1♣-(pass)-1♥-(3♦)-pass/Dbl-(pass)-? I can comfortably bid 3♥ in MOSCITO.

[straube]

I think straube has a point that showing 5♠ is less important than showing 5♥. When the auction becomes competitive, responder can always bid his suit when holding ♠ without raising the level. However, there's more to it than that. 2-suited hands also contain 6-4, 5-5 and 7-4, in which case it's very handy to show 5M immediately and act accordingly afterwards.

 

For example, suppose opps interfere to 3♥. When responder has shown 4+♠ (or "any semipositive"), he can bid 3♠, but is that a 5, 6 or 7 card suit? When he has shown 5+♠ already, it now shows 6+♠. With a 5-5 he can bid 4m (after showing 4+♠ this rather shows 4♠-6m imo).

 

Similar, when responder denied a 5 card M, he can always bid the M at any level (not too high ofcourse) with an unbalanced hand with 4M and longer m. Opener knows what to expect. Say I have a max SP with 4♥-6♣. After 1♣-(pass)-1♥-(3♦)-pass/Dbl-(pass)-? I can comfortably bid 3♥ in MOSCITO.

 

Totally agree. It has value but less value than showing 5 hearts. How much less is hard to quantify.

[straube]

So I'm wondering if IMPrecision can extend it's 2-way idea a round further.....

 

 

1C-1D......

 

.....1H-17-20 NT or hearts

..........1S-DN or GF balanced or possibly a few rare shapes

...............1N-17-20 NT

....................a system based on transfers to allow relay for a few shapes, signoffs for the DN and bal vs bal bidding

...............other-hearts

....................S1 (usually)-GF balanced

..........1N etc-relays 7+ RP shapely hands at +2 or better

.....1N-21-23 balanced or semibalanced

[awm]

So I'm wondering if IMPrecision can extend it's 2-way idea a round further.....

 

 

1C-1D......

 

.....1H-17-20 NT or hearts

..........1S-DN or GF balanced or possibly a few rare shapes

...............1N-17-20 NT

....................a system based on transfers to allow relay for a few shapes, signoffs for the DN and bal vs bal bidding

...............other-hearts

....................S1 (usually)-GF balanced

..........1N etc-relays 7+ RP shapely hands at +2 or better

.....1N-21-23 balanced or semibalanced

 

You could do this, but I'm not convinced it's better. Some of the issues:

 

(1) By far the most common 1♦ "positive" is the balanced hand. You will have some issues when opener has hearts and responder is balanced by playing this way, because you cannot make full use of your space. For example 1♣-1♦-1♥-1♠ and opener cannot bid beyond 2♥ on most hands because of the double negative possibility. Opener may have to immediately show values on some hands to avoid a pass from the double negative (awkward for relay). The step bid cannot always be the relay because the double-negative may need to be able to correct the strain (i.e. 1♣-1♦-1♥-1♠-2♦ and 2♥ probably isn't the relay). I think you are losing quite a bit on these sequences, both in complexity and space.

(2) When responder has a shapely GF, you have saved about a step.. but you have wrong-sided some contracts, reduced the information available to responder (i.e. if you have 6(322) opposite 1NT you can just set spades and cuebid but now you have to resolve shape in case opener has hearts so you do lose something here)... and arguably once you are "+2" relays are not great in any case. In fact a part of the reasoning for our methods that hasn't been discussed is to avoid "+2 relays" in uncontested auctions always!

(3) You lose some space in describing shapely negatives (i.e. 1♣-1♦-1♥-2x in my methods) which can be quite helpful when they come up. You also lose the ability to show a negative with a heart fit right away (1♣-1♦-1♥-2♥) which can help in finding shapely games on low values. Note that 1♣-1♦-1♥-1♠-2x-2♥ presumably could contain a wide range of heart holdings (not necessarily a real fit) in your method.

[straube]

Which of your balanced hands with four or more spades rebid 1S after 1C-1D? Is it 21+? 17+? 4+ spades? 5+ spades? I would think 5 spades only.

 

If you had to play the change I proposed, which hands would rebid 1H (balanced 17-19) and which hands would rebid balanced 1S? 4 spades? 5 spades? I would think 5 spades only.

 

Or perhaps I would have to give up showing balanced with spades in order to show S/H hands...or more of them.

 

If I have the time, I want to look at frequencies, but those frequencies depend on what the answers are to those questions.

 

I think you made some good points. For example, it's rather nice that 1C-1D, 1H-2H shows 4 hearts with a DN. Not always nice as sometimes opener has the big balanced hand without a heart fit and retreats to 2N. But usually nice.

 

You have a nice way to show a spade preference after 1C-1D, 1H-1S, 2C. I would definitely have to assign 2D as a relay; no getting around that one. 1C-1D, 1H-1S, 2C-2H DN no fit is bad, but no worse really than what Atul and I presently play. So I worry about the S/H hands, but I'm not overly concerned about the giant heart hands.

 

I'm not worried about 1C-1D, 1H-1S, 2D-2S relaying because we're +0.

 

I don't like +2 either, but I'm concerned about showing all of the 5431s and 6421s etc. I think if partner has 17-20 balanced and knows partner's shape (say 5431) and knows he has at least 7 RPs, that some good can come of that. And when responder has like 9 or 10 QPs, he can reject sign off attempts.

 

1C-1D, 1H-2C/2D/2S as DN 5/5s feels wrong to me.

 

Hope you don't mind poking at your system. Atul and I are pretty excited about it.

[awm]

(1) We don't really rebid 1NT with five-card major after 1♣-1♦. If responder has GF we are much better to bid 1M (where we can relay, and in fact we are at +0 on one-suiters). If responder has garbage we are very often better to play 2M (even on a 5-2) than to play 1NT. One can make exceptions for really terrible major suits of course.

 

(2) We will rebid 1NT with 17-20 and any 4432, 4333, or 5m-(332) shape. This is regardless of major suit holdings. Bidding 1M here is not great because our balanced hand relays are not that efficient (thus benefit from more limited range of strength, which we see in many sequences), and because our bidding opposite DN tends to be better if we rebid 1NT on these.

 

(3) If we have a four-card major and a balanced 21+, we rebid that major. This helps us sometimes (i.e. finding spade fits after 1♣-1♦-1♠ is easy, and usually we are willing to bid to the field 2NT if we have to) but mostly it's to reduce the load of balanced hands on the 1♥ rebid (it improves our relay structure).

 

The 1♣-1♦-1♥-2♥ auction is normally not on utter trash (i.e. usually hands with like 2-4) so we are pretty safe if opener rebids 2NT natural over this (in fact we are usually going to game opposite 21+ balanced on these hands). It helps a lot to show a real heart fit right away when opener has hearts and it doesn't really cost opposite the balanced hand because it's a big one.

 

The problem you are going to have is that if you play 1♣-1♦-1♥-1♠-2♣ the way I do, you have to cover five-four and five-five hands in three suits (opener can have any second suit); you're going to be at +2 fairly often here I think. You can reduce the pressure somewhat by offloading some hands (say 5♥/4♦) into 2♦, but then you are behind there also and have some issues because DN responder doesn't know relative suit length too. Further you're losing responder's call when he has DN with a long string of spades, which is actually a potential game contract (and very often 2♠ is the best partial).

 

If I had to play your change, I wouldn't really want to bid 1♠ with four spades and a balanced 17-19 (because I can't get to 1NT any more) and would probably just bid 1♥ on all min balanced hands.

[straube]

I'm thinking about....

 

1C-1D,

.....1H-H various or 17+ balanced

.....1S-S/m or 5S bal

.....1N-H/S

 

Now we have to rebid 2N with 21-22 or so.

 

What do you think of this tradeoff?

[straube]

1C-1D, 1N would include the 3-suited short minor, of course.

 

Another thought would be to reassign 1C-1D-2H to 4S/5H (or the reverse), but I'm favoring 1N rebid as both majors (partly because it handles 3-suited).