2S or 1N?

[Walddk]

I have now sent an e-mail to 50 North American experts on 2/1 with the following wording:

 

Hi all,

 

We have a thread on the BBO forums regarding a single raise or a 1NT response. I'd like to hear the opininon of 10-15 2/1 experts.

 

S: Q9X

H: Q8X

D: QT9X

C: KXX

 

PD opened 1S, right hand opponent passed; playing 2/1 with the above hand, what do you respond? 2S or 1N?

 

I would appreciate your reply as soon as possible. 2S or 1NT is OK, but if you want to comment further as to why you bid this and not that, please do.

 

Regards,

 

Roland

 

....

 

With some luck I'll get the 10-15 replies I am looking for; if I am very lucky even more. What do you expect that I get back? How many % for 2S and how many for 1NT?

[Cascade]

Here are some interesting statistics I computed on this hand:

 

Opposite a 5332, 5422 or 6322 with 5 spades 14-19 hcp

 

This hand made double dummy

 

3NT 60%

4♠ 46%

 

A random 4-4-3-2 10 count with only two spades made double dummy

 

3NT 61%

4♠ 42%

 

A random 9 count with 4-4-3-2 or 4-3-3-3 with three spades

 

3NT 52%

4♠ 57%

 

Opposite a 5431 with spades the number were

 

Our actual hand

3NT 51%

4♠ 59%

 

balanced 10-count two spades

3NT 60%

4♠ 51%

 

balanced 9-count three spades

3NT 45%

4♠ 58%

 

These number suggest opposite a 'balanced' hand this hand behaves more like a 10-count with two spades than an average 9-count with 3 spades.

 

Opposite the unbalanced hand it behaves slightly better than a 9-count with three spades in 3NT but otherwise more like the 9-count.

 

As I said earlier which it is more like overall on average might depend on a partnership's propensity to open or not 1NT with a five-card major including 5-4-2-2s.

[helene_t]

With some luck I'll get the 10-15 replies I am looking for; if I am very lucky even more. What do you expect that I get back? How many % for 2S and how many for 1NT?

I don't really care. An answer like "this hand is better described as 9 points with 2-card support" (as Wayne suggests) or "this hand is better described as 7 points with 3-card support" would be interesting. A vote for 1NT (or 2♠) without motivation would not be very enlightening.

[655321]

When I agree to play a system I never agree to throw my judgement out the window.

I am not used to playing that sort of straight-jacket system.

To me, this sort of comment reads like justification for masterminding.

 

I am fortunate in that my partners don't feel compelled to rescue me from the stresses of decision-making by taking all the decisions themselves.

[Walddk]

With some luck I'll get the 10-15 replies I am looking for; if I am very lucky even more. What do you expect that I get back? How many % for 2S and how many for 1NT?

I don't really care. An answer like "this hand is better described as 9 points with 2-card support" (as Wayne suggests) or "this hand is better described as 7 points with 3-card support" would be interesting. A vote for 1NT (or 2♠) without motivation would not be very enlightening.

Don't worry; you'll get explanations too. Six responded so far (it's still very early in NA), and they all tell when they will bid 2♠ and 1NT. I'll bring all quotes eventually.

 

Roland

[glen]

My opinion:

 

2♠ and 1NT

 

2♠ at IMPs (or if 1NT is forcing) - 2♠ will often be okay as a contract, and 2♠ will move us towards a possible game - my partnership will need to be able to investigate 3NT as a possible contract - do Meckwell still play 1♠-2♠;-3♠ as "interested in 3NT played by responder"?

 

1NT at pairs if 1NT is semi-forcing or non-forcing - we may miss game after 1♠-1NT;-2X-2♠ (opener treating 2♠ as a doubleton if 8-10) but staying low can be good at pairs, and we get to 1NT when it can be a great pairs contract

[cherdano]

Cliff notes version:

If you raise to 2♠, you will be able to show a balanced hand with 3 spades worth a good single raise. If you bid 1N, you will not be able to show a balanced hand with 3 spades.

[Walddk]

I have now sent an e-mail to 50 North American experts.

Overwhelming. I got 44 replies. Unanimous decision: 2♠ it is. Here is what the experts say:

 

Peg Kaplan:

I respond 2S, and do not think it very close. (I don't think 1NT is ridiculous; I just don't like it). Here's why. It's not good enough for a 3 card limit raise. (Too soft values; not enough shape). If partner bids a suit over 1NT, what is my response now? If I bid merely 2S, he won't think I have 3 spades. If I bid 3 spades, he'll think my hand is better than it is. If I rebid 2NT over 2 of another suit, I'm showing more than I have. Would I like to have a doubleton somewhere? Ya; you betcha. But, 2S is closer than anything else, and covers rebids better than any other call.

 

Richie Reisig:

2S. The only time I prefer 1NT when weakish (holding 3 card supp) is when I'm VERY weak. Meaning, I don't want partner to get excited and overbid. The problem on bidding 1NT forcing - is what will you bid next (over 2C/2D)...if 2S - you sound like 2 card support. Whenever you raise immmed. - opener re-evaulates based on the fit...when you deny support - you send the wrong message.

 

David Grainger:

2S. In 2/1 with the forcing NT, 1NT with 3 spades should either have a limit raise, or be less than either constructive (7+-10+), or semi-constructive (6-10-), depending on agreement (I prefer the latter). Even if playing bart, the bart sequences tend to show hands with 2 spades. So even though you may really want to suggest NTs, you 1) won't be playing 1N anyway, and 2) will be forced to show a much worse hand than you have over partner's next call, and you'll still be in spades, unless you upgrade to invitational values, on this hand which is really just garbage. (Getting to 3N opposite balanced 13-14s or 5431 14-16s with stiff H/C is not going to be pretty). I would respond 1NT holding a hand such as xxx xxx xxx Kxxx or Qxx xxx Kxx xxxx, but having the intention of never letting the hand be played in NT.

 

Debbie Rosenberg:

2S. With 3-card spade support, I'd bid 1N only with a hand either too weak or too strong for 2S. Too weak would include hands with fewer than 6 support points, a flat hand with 6 HCP, or a flat 7 with poor spots and bad trumps. Too strong would be defined as a hand with 10-12 support points, though a flat 10 HCP, with poor spots, might still bid 2S. With the "too weak" hand, my plan is to take a preference next time. Partner will expect only a doubleton trump for this action. With the "too strong" hand, my plan is to jump in spades next time.

 

Danny Sprung:

2S. Support with support. The only hands I bid 1nt with 3 card support for a major are limit raise with 3 trump, or a total sub minimum response, say 2 queens.

 

Henry Bethe:

2S; but I am biased as I play that 1NT virtually denies 3 card support, certainly constructive 3 card support. This hand is surely not good enough to bid 1N and then 3S, and is far too good to bid 1N and then 2S.

 

Phillip Alder:

I would bid two spades. This hand is not quite good enough to invite game. If I had, say, the jack of hearts instead of the eight, I would bid one notrump, then two notrump. This hand does have notrump written all over it, but partner could be short in hearts or clubs, which would not be so good. And to have the sequence start 1S-1NT-2X-2S, partner will assume I have only two spades.

 

Brad Moss:

2S.

 

Larry Cohen:

2S. Not close. I play 2S can be up to a decent 10 -- and this hand, in spite of those decent spots is 4x3 and not worth 1NT followed by 3S.

 

Jack Oest:

Two Spades. While I understand the hand looks like NT, I might never recover from a 1NT start. Assuming partner rebids his second suit, what then? My 2S bid now sounds only like a preference and partner will pass on many hands where he might have moved had he known he was facing three card support. Raising spades now does not preclude getting back to NT later.

 

Bob McPhee:

2S for me Mr Wald. This is what I have.

 

Robb Gordon:

2S. This certainly is not close to a 3 card limit raise!

 

Jill Meyers:

Since I play forcing nt over 1s I would bid 2S directly. If I bid 1nt and then correct partner's rebid to 2s my hand will seem weaker or fewer trumps and I don't have enough to rebid 3s over partner's rebid (3s being 3 card LR).

 

Fred Gitelman:

2S. The main reason is that I think this hand is too strong for spades to consider bidding 1NT followed by 2S. This would be especially true if non-vul when I would often bid 1NT on close to nothing as a "tactical psych". 1NT then 2S is an intentional distortion IMO - partner will never expect a hand with both 3 spades and this many HCP. 2S also has tactical advantages - it makes it harder for LHO to get into the auction via an overcall (unless you play that OBAR thing and, if you do, good luck to you when you actually have a real hand). For me 2S is completely normal. I try to keep an open mind about such things, but honestly it would not occur to me to be anything other than 2S.

 

Mildred Breed:

I respond 2S. Since I play F1 NT, it is clear-cut to me. If 1NT is not forcing, there is more to be said for the 1NT bid.

 

Nader Hanna:

2S. If I bid a forcing 1NT then minimally support Spades it shows a weaker hand or only two card support. The hand is not good enough to jump to 3S (limit raise with 3 card support) after partner’s rebid over 1NT. If partner makes a game try over 2S I can bid NT to suggest a maximum balanced raise with scattered values.

 

Bobby Wolff:

2 Spades - Just about right. Too strong for 1NT and then 2S and not strong enough for 3S.

 

Bart Bramley:

2S. I have no objection to bidding 1NT with this hand TYPE, but 9 HCP is too much for 1NT. However, I will reject all game tries except 2NT, which I will raise to 3NT. Despite the 9 HCP, the bad shape and lack of aces and kings makes this a minimum raise. I assume we are playing 5-card majors. If I’m playing a style that includes frequent 4-card spade suits I would bid 1NT.

 

George Jacobs:

2 Spades. No reason to mastermind. You are probably not planning to jump to 3 Spades to show a limit raise later.

 

Mark Feldman:

2S.

 

Matt Granovetter:

2S.

 

George Holland:

2S. I am prepared for all game tries by partner.

 

Andy Stark:

2S, like a shot. Too weak to bid 1NT, then raise to 3S. Too good to bid 1NT, then rebid 2S which implies a doubleton spade.

 

Nikolay Demirev:

2S for me with the most developments. I trust partner to offer 3nt later and I see no particular need to make sure my hand will declare. However, if playing semi-forcing or forcing NT and 1S-1nt-2C-2D!-2S! showing 9+ with a doubleton in pairs I will lean towards responding 1nt, especially if we tend to open the major with 15-17 on certain hands. The implied alternative 1S-1N-2x-3S is not in consideration for me. I am a whole red K short for that.

 

Jan Martel:

I thought 2S when I read the BBOF thread, but I think it's really a partnership question more than a standard 2/1 question.

 

Kit Woolsey:

My philosophy is to always raise with 3-card support regardless of the hand. Just because I am balanced doesn't mean that partner is. The raise is more preemptive, blocking a 2-level overcall, and establishes a trump suit quickly. In addition, when I bid 1NT partner knows that I don't have support (unless I have a 3-card limit raise which I can show next turn). Thus, when I later take a preference to 2S he will know it is a doubleton and can act accordingly.

 

Bruce Gowdy:

I bid 2 spades 6-9 points 3 card support. I like using the "8 missing AKQ count for limit raises"--thus on this hand you have an 8 count (there are two missing AKQ in each suit), but without an ace I reduce the value of the queens-thus for practical purposes this is a nine or even a ten count.. Therefore I cannot bid a forcing nt and then jump to 3 spades showing a 3 card limit raise. If the club King was the ace, now I bid 1nt forcing and jump to 3 spades next. Note that 1spade-1nt-2 spades- 3 spades is not invitational in a sense, but still just a limit raise-inferring that you would have bid 3 spades over any two level response.Further on this holding, if it went 1 spade-1nt-2spades, I would raise that to 3 spades as the presence of a 6 bagger in pards hand increases the value back to limit raise status.

 

John Stewart:

2S, no choice for me.

 

Jon Wittes:

2S, clear cut. I would only respond 1NT on hands that are too weak for a constructive raise, or hands that I intended to show as a balanced limit raise. This hand is clearly in neither of those categories.

 

Valerie Westheimer:

2S. I wld play that 1NT, then choicing back to 2S over partner's expected minor-suit rebid is much weaker than the hand I have here. Set trumps, support when you have support, keep partner happy :)

 

Jeff Rubens:

My view on your problem is that this is clearly a two-spade response in Bridge World Standard, too strong for one notrump + two spades.

 

Adam Wildavsky:

2S. Too good for 1N followed by 2S, not good enough for 1N followed by 3S. Note that the hand's not as good as the raw point count suggests, even with all those spots. A CCCC evaluation gives 7.30:

http://www.gg.caltech.edu/~jeff/knr.cgi?ha...9X+Q8X+QT9X+KXX

 

John Swanson:

2S. If I bid 1NT with queens and spot cards I would want the values to rebid 2NT.

 

Steve Robinson:

2 spades. No aces, only one king makes this hand not worth a three-card limit raise.

 

Eric Kokish:

2S. Not close to 1NT, whether forcing, semi-forcing, or NF.

 

Jim Tritt:

2s - I play that if I bid 1nt and then preference to 2s, that would show a hand that would not accept any game try. Direct 2s is better, will accept some game try. This hand not good enough for 1nt followed by 3s.

 

Roy Hughes:

2S, definitely. 1N would have to be followed by 2S, an underbid which usually contains only 2 trumps, or 3S, which is a big overbid.

 

Zia Mahmood:

2 spades 100/100..1nt 0/100.

 

Mike Hargreaves:

2S. Frankly, this is a non-problem in 2/1, whether one plays constructive raises or not. The hand strongly suggests a notrump contract, but responding 1NT, ironically, does very little to get us to notrump... it is, after all, forcing. While I don't mind downgrading really soft hands, such as this one is, I can't not treat it as constructive.... I do have 9 hcp and some spots, and even tho to me it is worth maybe a decent 7 count, that is enough for a constructive raise.

 

Walter Johnson:

2S and not close.

 

Gene Saxe:

I have a strong preference for 2S. At the same time, my style is sound invites, aggressive tries/acceptances. If partner has AJxxxx, Kx, x, AJxx where game is about 50%, and you bid 1NT, opener will bid 2S with this hand most of the time, and now, you will be overbidding to raise to 3S. If opener rebids 2C instead, and you preference 2S (why would you jump to 3S?) how can opener raise and try again? However, after a 2S raise, you will have a play with 2-2 trump and Kx of clubs, so you would try. Any red A in addition (or the SK) would make game a very good bet, with Kx or Kxx of C.

 

Harmon Edgar:

I would bid 2S. Its a GOOD 2s bid but one of the cards is probably going to be useless and you are 4333. 3S is too much of an overbid after 1N.

 

Marshall Lewis and Jim Gordon also replied. Both vote for 2♠.

[lexlogan]

From Cascade's stats, if partner is at all likely to have a relatively balanced hand worth a game try, 1NT may be a big winner. Now what if opener is 5431? The bidding will begin 1S-1NT-2x-2S. With only 14-15 hcp, opener may give up. If the 59% for game applies to all 14-19 hands with this shape, I'm going to assume 14-15 with this shape is only about 50% or less. Perhaps Cascade can simulate that more specific case. (And did you mean any 431 in the side suits?) With 16+, I assume opener will make another try, and it won't be in notrump. Once he bids his third suit (or rebids his second suit with 55 shape), we can leap to 4S, unless perhaps his singletoon is in clubs.

 

To summarize: when opener has a minimum hand, we land at 2S, without the opps realizing we have a fit. This may discourage them from balancing.

 

When opener has a balanced, game invitational or better hand, we land at the superior 3NT contract, played from the correct side and with the spade fit concealed.

 

When opener has an unbalanced, borderline game try, we stop at 2S, which appears to be a winning contract.

 

When opener has an unbalanced, stronger game try, we reach the same 4S contract we'd get to via 1S-2S.

 

I'm always skeptical of simulation results but it looks like 1NT may be a huge winner.

[lexlogan]

Here is the main reason why bidding 1N then 2S is harmful to your constructive bidding:

 

Partner will play you for a doubleton spade and will pass with hands that would have game tried over a 2S raise. Hands like 5431 15 counts and 55 14 counts that are strong hands when they find an 8 card spade fit but not that great otherwise....

From Cascade's simulations as well as simply loooking at this hand, discouraging partner from making a game try with 14-15 unbalanced appears to be a winning, not losing strategy. The difference between 2 and 3 card support is typically 1/3 of a trick, i.e., one point. Hands simply don't gain or lose a lot of playing strength based on 2 or 3 card support when dummy has no ruffing value. There's a much bigger swing between 3 and 4 card support than between 2 and 3.

[jdonn]

I have now sent an e-mail to 50 North American experts.

Overwhelming. I got 32 replies within 5 hours. Unanimous decision: 2♠ it is.

Thanks for posting it Roland. I stand by 1NT being 100% wrong. If that doesn't prove it, then what else can I say...

[Finch]

I'm surprised this hand has caused so much passion.

But to play devil's advocate against your summary:

 

To summarize: when opener has a minimum hand, we land at 2S, without the opps realizing we have a fit. This may discourage them from balancing.

 

I'm not entirely sure why you want to discourage them from balancing. If partner has, say, a 13-14 balanced hand I think I'd quite like them to be playing at the 3-level. Preferably doubled.

 

When opener has a balanced, game invitational or better hand, we land at the superior 3NT contract, played from the correct side and with the spade fit concealed.

 

Responding 2S does not preclude getting to 3NT. It might end in 3NT by either hand depending on methods, but it's not at all clear at this point what the right side actually is.

 

When opener has an unbalanced, borderline game try, we stop at 2S, which appears to be a winning contract.

 

Possibly true. You do of course have to weigh that against games you miss because partner doesn't realise you have 3-card support.

[pclayton]

I too am amazed at the amount of replies this thread getting. Frankly I consider it a non-problem.

[Walddk]

I too am amazed at the amount of replies this thread getting. Frankly I consider it a non-problem.

I don't think any problem is a non-problem. If in doubt, it's silly not to ask if you want to be enlightened.

 

Roland

[Cascade]

From Cascade's stats, if partner is at all likely to have a relatively balanced hand worth a game try, 1NT may be a big winner. Now what if opener is 5431? The bidding will begin 1S-1NT-2x-2S. With only 14-15 hcp, opener may give up. If the 59% for game applies to all 14-19 hands with this shape, I'm going to assume 14-15 with this shape is only about 50% or less. Perhaps Cascade can simulate that more specific case. (And did you mean any 431 in the side suits?)

Yes it was any 431 on the side.

 

Here are some more specific simulations:

 

14-15 hcp 5431

 

Stiff heart

 

4♠ made 41%

3♠ failed 18%

 

Stiff diamond

 

4♠ made 38%

3♠ failed 15%

 

Stiff Club

 

4♠ made 32%

3♠ failed 22%

[Cascade]

I have done a single dummy simulation of this situation using GIB.

 

This is not completely ideal as GIB plays some sort of constructive raises.

 

It describes the simple raise as "3+ S; 7-10 points".

 

However the route through 1NT it describes as a wider range "2-3 S; 6+ HCP; 11- points"

 

My experience playing with GIB in the robot race money tournaments is that in practice GIB voluntarily raises with hands weaker than 7 hcp.

 

The simulation was based on 100 hands. Each hand was played twice once with the forced start of 1♠ Pass 2♠ ... and once with the forced start 1♠ Pass 1NT ... to the auction.

 

The simulation was across the range of hands that would open 1♠. I also constrained the second seat hand to not have a clear cut bid. I varied the vulnerability according to board number by the normal rotation as I don't think that was specified in the orginal problem.

 

Over the 100 hands the simulation was very close. Responding 2♠ lost to responding 1NT 166-167.

 

I have done some analysis of where the wins and losses occurred.

 

Non-Competitive Part-score - no swing

 

22/100 hands

 

Non-Competitive Part-Score - lose IMPs

 

4/100 hands - lost 7 IMPs

 

0-7 (26 hands)

 

Non-Competitive Part-Score - won IMPs

 

7/100 hands - lost 7 IMPs

 

7-7 (33 hands)

 

Competitive Part-Score - no swing

 

3/100 hands (36 hands)

 

Competitive Part-Score - lose IMPs

 

6/100 hands - lost 20 IMPs

 

7-27 (42 hands)

 

Competitive Part-Score - win IMPs

 

11/100 hands won - won 53 IMPs

 

60-27 (53 hands)

 

Game - Same Result

 

12/100 hands (65 hands)

 

Game - Lost IMPs

 

2/100 - lose 13 IMPs

 

60-40 (67 hands)

 

Game - Won IMPs

 

1/100 - won 11 IMPs

 

71-40 (68 hands)

 

Different Game - no swing

 

2/100 (70 hands)

 

Different Game - lose IMPs

 

1/100 - lost 13 IMPs

 

71-53 (71 hands)

 

(There were no hands where we got to a different game and won IMPs by bidding 2♠)

 

Game/Partscore - lose IMPs

 

7/100 hands - lost 37 IMPs

 

71-90 (78 hands)

 

Game/Partscore - win IMPs

 

6/100 hands - won 49 IMPs

 

119-90 (84 hands)

 

Game/Partscore - no swing

 

1/100 hands (85 hands)

 

Competitive Game - lose IMPs

 

2/100 hands - lost 18 IMPs

 

119-108 (87 hands)

 

Competitive Game - won IMPs

 

1/100 hands - won 5 IMPs

 

124-108 (88 hands)

 

Doubled Game - lose IMPs

 

2/100 hands - lost 10 IMPs

 

124-118 (90 hands)

 

Doubled Game - won IMPs

 

1/100 hands - won 5 IMPs

 

129-118 (91 hands)

 

Double Game/Partscore - won IMPs

 

1/100 hands - won 9 IMPs

 

138-118 (92 hands)

 

Opponent's Game - no Swing

 

1/100 hands (93 hands)

 

Opponent's Game - lost IMPs

 

1/100 hands - lost 11 IMPs

 

138-129 (94 hands)

 

Opponent's Game - won IMPs

 

1/100 hands - won 11 IMPs

 

149-129 (95 hands)

 

Doubled Partscore - lost IMPs

 

1/100 hands - lost 16 IMPs

 

149-145 (96 hands)

 

Doubled Partscore - won IMPs

 

1/100 hands - won 15 IMPs

 

164-145 (97 hands)

 

Game/Slam - win IMPs

 

1/100 hands - won 2 IMPs

 

166-145 (98 hands)

 

Defensive Card Play - lost IMPs

 

2/100 hands - lost 22 IMPs

 

166-167 (100 hands)

 

All of this proves little as this is clearly a small sample except that in a 2/1 type system with the knowledge that partner might bid 1NT with three spades as GIB appears to be programmed for that it is likely to be close between bidding 1NT or raising to 2♠.

 

Many of the swings against 1NT were in the part-score battle where it was easier (lower) for 4th hand to compete over 1NT than over 2♠.

 

1NT seemed to gain though on a higher proportion of the bigger swings.

 

I noticed that two of the swings against the 1NT response came from very odd 'GIB-type' actions. On one hand after bidding 1NT it doubled a 2♦ overcall presumably for penalties - in practice if I responded 1NT I would never double a two-level contract without showing my support for partner. On the other hand it passed opener's 2♣ rebid - again 2♠ seems automatic. These resulted in a making doubled contract (game as the opponents ran from 2♦ into some other contract) and a missed 4♠ game respectively.

 

Quite likely there were other odd GIB actions and maybe these evened out but these two seemed to stick out like the proverbial sore thumb.

[Walddk]

I am impressed, Wayne. A lot of work for you. We are dealing with a 'trivial' single raise or a 1NT response and get all this as a bonus. Bridge is fascinating.

 

Roland

[Edmunte1]

I too am amazed at the amount of replies this thread getting. Frankly I consider it a non-problem.

I don't think any problem is a non-problem. If in doubt, it's silly not to ask if you want to be enlightened.

I completely agree with that, Roland, dubito ergo cogito :)

[Cascade]

I am impressed, Wayne. A lot of work for you. We are dealing with a 'trivial' single raise or a 1NT response and get all this as a bonus. Bridge is fascinating.

 

Roland

You too did a lot of work.

 

I think the reality is that we don't know the answers to a lot of these questions. Or even if we do we don't really have any idea how good the alternatives are.

 

The real answer to a question like this will depend on many factors and it is unlikely that it will be the same for everyone. Not least of which is partner's style and judgement - exactly which hands will partner open 1♠ on? This is far from a non-trial question.

 

Indeed bridge is fascinating. And a question like this sparks my interest in simulations which don't give answers but in many situations help to give me a handle on how different actions work.

[mikeh]

Wayne, I am impressed by your hard work.

 

I personally don't put much store in simulations in which the outcome is assessed on a double dummy basis. I also note that your simulation constrained 2nd seat from having a clear overcall (which introduces further subjectivity into the analysis.. I don't have a problem with that since I don't like any simulations that ignore the opps) but did not discuss what to me is a far, far more salient point: 4th seat.

 

One huge advantage that 2♠ holds over 1N is that 2♠ will often shut 4th seat out, where 4th seat had a relatively easy 2-level overcall or even a light takeout double (standard practice is to use the double of 1N as takeout of spades).

 

A 4th seat overcall creates a horrible scenario. Opener, not knowing of the fit, will often be forced to pass. We can survive that, sometimes, by having responder bid 2♠ next... but this assumes that he is allowed to.. what if the bidding is at the 3-level by the time it comes back?

 

Further, and perhaps finally, the choice between 2♠ and 1N cannot be made in isolation. Single raises of major openings are part of the fundamental structure of the overall method. I note that, for example, your GIB program permitted a 1N on 2-3 spades, 6-11 hcp. Whatever that method is called, it is NOT normal 2/1.

 

Since GIB is designed, in later bidding, to accept that a 1N response may include a 3 card raise with 9 hcp, it is not surprising that it may go right more often than most of us players of a mainstream 2/1 would expect... we do NOT include this hand-type in our 1N response, so will never cater to it in later bidding.

 

As I read your simulation, it suggests that IF you design your methods to permit 1N with this hand, it is playable. But that was not the question.. the question was whether, in the context of a normal 2/1 approach, 1N is acceptable.. and it is apparent that the collective judgement of a lot of very experienced players (including some notable theorists) is that it is not.

[Cascade]

Mike

 

This simulation was single dummy not double dummy. Single dummy is how you and I play every hand we play. There is no double dummy component to this simulation except that GIB internally simulates hands double dummy and then chooses the best option in the card play.

 

There was no constraint on 4th seat because GIB's bidding engine took over and bid all hands after the three bids of the original problem.

 

Indeed the biggest losses were in competitive partscore situations where bidding 1NT lost 20-53. However those losses in this simulation were picked up over a variety of situations at higher levels where bidding 1NT initially worked out better than bidding 2♠ in this particular simulation.

 

I am not an expert on normal 2/1 whatever that is. I know though that some versions use a direct constructive raise. GIB appears to play 2/1 - 2/1 bids are forcing to game and it plays a forcing 1NT. In that context it plays raises as 7-10 with three spades. I would assume that Matt Ginsberg consulted some 2/1 experts when constructing this system for GIB. Maybe standard 2/1 is more rigid than systems and styles that I am used to but I would be surprised if there is a standard that is as rigid as some have suggested in this thread.

 

I do not know enough about the internal structure of GIB to know precisely how its subsequent bidding is influenced by it failure to make a 7-10 three-card raise. It is not entirely clear what GIB's style is but it seems more flexible than that of others in this forum who have argued that this is not a matter of style and judgment.

 

The conclusion that can be reached is that playing GIB's 2/1 style based on this simulation we cannot conclude that bidding 1NT is worse than bidding 2♠ with this particular hand.

 

It is possible but I would be surprised if a subtlely different 2/1 system would produce significantly different results (unless of course this particular simulation is aberrant - which is possible it is just a small sample).

 

I usually play a quite different structure in which 1NT is not forcing and 1♠ could be four. Although 90+ % of the time (maybe higher - I don't have the numbers here) 1♠ shows five or more. So much so that our normal practice is to raise 1♠ to 2♠ with three trumps. It is very much the exception to bid 1NT with three-card support. This is the sort of hand where I would bid 1NT. When encountering the problem in a different context of 2/1 and forcing NT I naturally considered my experience from using other systems and could not immediately see why 1NT would necessarily work out badly - it would act as a brake which could be good given that we will show soft cards and poor distribution for a potential spade contract.

 

I would have thought that if 2♠ was so clearly indicated in 2/1 that even given the constructive raise style of GIB that the immediate raise would perform better in this simulation. Maybe it would in a longer simulation. Perhaps I will extend the simulation.

 

For now I would like to suggest that raising to 2♠ on this sort of hand may not be as clear-cut as expert opinion suggests.

[cherdano]

Why did raising to 2♠ lose 22 IMPs on defensive card play? This seems pretty random. I would expect to lose IMPs on defense when bidding 1N with 3-card support.

 

I am surprised there were so few instances where bidding 1N or bidding 2♠ led to a different game. But then, there is no reason to assume that GIB would get the subtleties of the COG auctions right after either start.

[Cascade]

Why did raising to 2♠ lose 22 IMPs on defensive card play? This seems pretty random. I would expect to lose IMPs on defense when bidding 1N with 3-card support.

 

I am surprised there were so few instances where bidding 1N or bidding 2♠ led to a different game. But then, there is no reason to assume that GIB would get the subtleties of the COG auctions right after either start.

Its not that easy to read but here are the two hands that GIB defended differently and that cost.

 

Enter S hand: AJT54.J763.K53.Q

Enter W hand: 7.A954.AJ764.J96

Enter N hand: Q92.Q82.QT92.K32

Enter E hand: K863.KT.8.AT8754

Enter dealer: S

Enter vul (none, N/S, E/W, both): N

1S.P.2S.3C.P.3S.P.4C.P.5C.P.P.P

I play SA

I play S7

I play S2

I play S3

 

I play SJ

I play C9

I play S9

I play S8

 

I play H5

I play H2

I play HK

I play H7

 

I play S6

I play S4

I play C6

I play SQ

 

I play CJ

I play C2

I play CA

I play CQ

 

I play CT

I play D3

I play D4

I play CK

 

I play D9

I play D8

I play DK

I play DA

 

I play D7

I play DT

I play C7

I play D5

 

I play C8

I play H6

I play H4

I play C3

 

I play HT

I play H3

I play HA

I play H8

 

I play H9

I play HQ

I play C4

I play HJ

 

I play C5

I play ST

I play DJ

I play D2

 

I play SK

I play S5

I play D6

I play DQ

 

Result: N/S -400

1S.P.1N.2C.P.2S.X.P.P.3D.P.3S.P.4C.P.5C.P.P.P

I play CQ

I play C6

I play C2

I play CA

 

I play D8

I play D3

I play DA

I play D2

 

I play S7

I play S2

I play SK

I play SA

 

I play HJ

I play H4

I play H2

I play HK

 

I play HT

I play H3

I play HA

I play H8

 

I play D4

I play D9

I play C7

I play D5

 

I play S3

I play S4

I play C9

I play S9

 

I play H5

I play HQ

I play CT

I play H7

 

I play S8

I play ST

I play CJ

I play SQ

 

I play H9

I play C3

I play C8

I play H6

 

I play S6

I play SJ

I play D6

I play DT

 

I play S5

I play DJ

I play DQ

I play C4

 

I play C5

I play DK

I play D7

I play CK

 

Result: N/S +50

 

Enter S hand: KJT865.K73..AT64

Enter W hand: A743.6.J86.Q9875

Enter N hand: Q92.Q82.QT92.K32

Enter E hand: .AJT954.AK7543.J

Enter dealer: S

Enter vul (none, N/S, E/W, both): N

1S.P.2S.X.4D.P.4S.5D.X.P.P.P

I play SJ

I play SA

I play S2

I play CJ

 

I play H6

I play H2

I play HA

I play H3

 

I play H5

I play H7

I play D6

I play H8

 

I play S4

I play S9

I play D7

I play S5

 

I play HJ

I play HK

I play DJ

I play HQ

 

I play D8

I play DQ

I play DA

I play C6

 

I play DK

I play S6

I play C7

I play D2

 

I play D3

I play CT

I play C8

I play DT

 

I play D9

I play D4

I play C4

I play CQ

 

I play SQ

I play D5

I play SK

I play S7

 

I play HT

I play ST

I play C9

I play C2

 

I play H9

I play S8

I play S3

I play C3

 

I play H4

I play CA

I play C5

I play CK

 

Result: N/S -550

 

1S.P.1N.2S.3S.P.4S.5D.P.P.X.P.P.P

I play CA

I play C7

I play C2

I play CJ

 

I play SK

I play SA

I play S2

I play H4

 

I play H6

I play H2

I play HA

I play H3

 

I play H5

I play H7

I play D6

I play H8

 

I play S4

I play S9

I play D7

I play S5

 

I play HJ

I play HK

I play DJ

I play HQ

 

I play D8

I play DQ

I play DA

I play C4

 

I play DK

I play S6

I play C5

I play D2

 

I play D3

I play CT

I play C8

I play D9

 

I play DT

I play D4

I play C6

I play S7

 

I play C3

I play D5

I play SJ

I play CQ

 

I play H9

I play ST

I play S3

I play SQ

 

I play HT

I play S8

I play C9

I play CK

 

Result: N/S +100

[awm]

To summarize the play issues, GIB lost 22 IMPs on defensive card play after 1♠-P-2♠ because it lead a spade after the raise when this was not the best lead. The 1♠-P-1NT auction that ended in the same contract for opponents resulted in GIB leading something other than a spade.

 

In one case GIB lead a spade from AJTxx setting up declarer's king; in the other GIB lead a spade from KJTxxx but declarer was void, and the immediate pitch on dummy's ♠A allowed the contract to make. Both hands were in game contracts declared by the opponents.

 

Especially considering that the spade raise did include the queen, there are probably plenty of hands where a spade is a fine lead. The fact that two hands where GIB lost in this way appeared in the sample (and that both were defending game contracts) is probably random luck (and small sample size). One could imagine hands where leading from ♠KJxxx is in fact necessary to beat the contract and is more likely after the raise than after 1NT.

 

While Cascade clearly put a lot of work into this, I think his results are unfortunately somewhat random. Very frequently the difference is on the order of one hand out of a hundred. That could just be luck. The only thing that seemed at all significant to me was that bidding 2♠ wins consistently in competitive partscore auctions (at least for GIB vs. GIB). This was over a fair number of hands and by a fairly wide margin, implying to me that "hiding the fit" is less effective than "obstructing the opponents" in a competitive sequence.

 

The other statistics besides competitive partscores generally involved only one hand either way and/or scored out very close to even. The margin lost by 1NT on the competitive partials is mostly made up by the two "misdefended opponent game" hands where GIB made a bad opening lead after the raise (but not after the 1NT response).

[Cascade]

I have extended the simulation to 1000 hands played single dummy by GIB.

 

Each hand is played twice once forcing the auction to start 1♠ Pass 2♠ and once forcing the auction to start 1♠ Pass 1NT.

 

GIB took over and bid and played the hands after those two starts. As I stated earlier GIBs methods are the 2♠ raise shows 7-10 hcp and a simple spade preference after a 1NT response does not seem to deny three spades. In practice in my experience GIB when left to its own devices seems to make the 7-10 raise with a wide range of hands (even with fewer than 7 hcp). I am not sure how frequently GIB actually bids 1NT with three-card support and a weak hand (6-9 or so hcp).

 

Over 1000 hands the 2♠ raise lost IMPs compared with the 1NT response. The IMPs score was 1974-2021.

 

That is an average IMP advantage for the 1NT response of 0.047 IMPs per board.

 

The standard error was 0.183172987 IMPs per board.

 

Statistically this means the difference between the two bids was not significant. A sample approximately 100 times bigger would be needed before we could hope to get a definitive answer to which is the better bid.

 

For the 1000 hands (actually the original 100 hands reported earlier and 900 new hands) I did not go through and analyze where the IMPs were won and lost for each call. There are 189652 lines of GIB output to wade through.

 

The frequency of IMPs won/lost (from a 2♠ response) is shown below:

 

15 2

14 4

13 8

12 17

11 13

10 53

9 5

8 8

7 31

6 40

5 25

4 14

3 12

2 26

1 72

0 308

-1 86

-2 22

-3 18

-4 15

-5 53

-6 44

-7 23

-8 8

-9 16

-10 27

-11 17

-12 14

-13 16

-14 1

-15 0

-16 2

 

Again the only conclusion that can be drawn from this simulation is that the results from GIB playing its version of 2/1 do not support the hypothesis that there is an IMP advantage to responding 2♠ rather than 1NT to 1♠.