[DavidKok]

I should caution that the percentages are not matchpoint scores, though they are related. I simply pitted 1NT versus 2♠ and checked which score was higher, as a tally out of 5,000 or 10,000. In a field where those are the only two contracts this number is closely related to your matchpoints score, but in a more heterogeneous field there might theoretically be a dozen other scores to compare against.

[Gilithin]

DavidKok, on 2023-December-29, 02:48, said:

Here is the average double dummy bridge score of 2♠ minus 1NT

If people are interested I'd be happy to share the dealer script, seeds for the simulations, and/or discuss this in more detail.

What are your criteria for rebidding 1NT with a singleton, something accepted and even standard in many popular systems? What are your criteria for 3 card raises, something also common in some systems? It seems to me that your answers to these questions will have a massive impact on the overall assessment. If you play the French/German style, where a raise can only be made with 4 and a 1NT rebid will always be a balanced hand, then it seems clear to bid the major. For other methods I would be surprised if the answer gets increasingly unclear and at some point swings the other way. Stating your answer as definitive while basing it on a particular set of assumptions is not really kosher.

[DavidKok]

As discussed earlier in the thread, I disallow rebidding 1NT on a singleton in my simulations. Opener is allowed to make a 3-card raise with an empty doubleton outside even if balanced. If there is a preferred set of methods you wish to compare I'd be happy to take a look at that, I started with the style I prefer. I had the concerns that not only do people have different sets of rules about rebidding 1NT (maybe the singleton needs to be an honour? Maybe it depends on the quality of your five card minor? And what if your hand is outside your 1NT rebid range but was unsuitable for a 1NT opening?), also I expected those hand types to be rare and presumably not move the needle much (and you'd lose out on the possible gains from bidding 2♠ every time partner does not have your singleton hand). So I started with the simple case I'm more familiar with, while offering to run others multiple times.

Also I think you would do well to reread the thread. What I am claiming is the following:

• Personally I think rebidding the 5cM always, or at least 'almost always', is best.
  
• There is classical advice about checking for suit quality and strength of the hand and outside shape and more, and my personal experience is that you are better off disregarding all that and just bidding.
  
• In an attempt to give a bit more support for this I've run a few simulations that might give a better insight into what's going on. This is not a "definite answer" - my answer was my first claim, preceding any computer stuff.

I even gave a possible explanation for how the classical advice might have come about despite, in my opinion, being worse. Personally I find discussions like this very difficult. I have an idea and opinion, supported by my experience at the table over a couple of years, and I present it here. In an attempt to provide more evidence for the belief I already hold I run a double dummy simulation, and offer people to check my code and to do repeat the simulations for other styles or bidding approaches. In thanks mikeh implies that I misread Bird & Anthias and now you are telling me my comments are not kosher. I'd be happy to remove all mentions of computer simulations and pick up the discussion from there, but rather I think the onus is on everyone else to give some more motivation for pass being better on average.

[awm]

I'd be interested to see the breakdown of your statistics based on degree of support (3-card support and side doubleton vs. 3-card support and 4333 vs. 2-card support vs. singleton).

For another way of looking at this, here's a table of declarer's advantage vs. double dummy (based on top-level play). This matches my experience that 1NT gives one of the largest advantages to declarer (declarer takes about 0.3 more tricks than double-dummy and makes about 7% more often) whereas 2M yields much less advantage (declarer takes about 0.1 more tricks than double dummy and makes about 2% more often). Of course this depends on the hand and auction and opponents, but it does suggest that your double-dummy simulations underrate 1NT as a contract.

[Gilithin]

DavidKok, on 2024-January-02, 02:38, said:

you are telling me my comments are not kosher.

I am saying that stating definitively that always rebidding the 5 card suit regardless of agreements based on a model of a specific set of agreements that strongly favour this course of action is misleading. I do not dispute your assertion that rebidding the 5 card suit under those conditions is the generally recommended course of action, nor do I find your posts generally lacking in merit.

DavidKok, on 2024-January-02, 02:38, said:

If there is a preferred set of methods you wish to compare I'd be happy to take a look at that

See Adam's post for a set of conditions that may well produce a different outcome. I would suggest that looking there would be a good start.

[helene_t]

Gilithin, on 2024-January-02, 05:56, said:

See Adam's post for a set of conditions that may well produce a different outcome. I would suggest that looking there would be a good start.

Yes. In that case it would be good to have spelled out which awkward hands the 1NT rebid should accomodate. For example, there was a recent discussion in one of the Facebook groups where Kieran Dyke argued that something like
KJx-x-AJxxx-KJxx (I don't remember the exact honours)
should rebid 1NT after a 1♥ response. This is not a hand that has a terrible rebid problem in French style (2♣ is not a distortion), but nevertheless the 1NT rebid obviously has merits.
Also, if we swap the minor suits, some people still open 1♦, but then the temptation to rebid 1NT is bigger as you don't want partner to take a false preference for diamonds.

[DavidKok]

I repeat/second this - could you describe which hands will rebid 1NT on a singleton for this simulation? The auction I'm limiting this to is 1♣-1♠. So in particular:

• Is there a (negative?) suit quality or length requirement on the club suit?
  
• Does the singleton need to be an honour?
  
• How do you handle your 4♦5♣ hands? Or, for that matter, 4♦441 hands?
  
• Can the strength be outside the 12-14 range, e.g. a 1=3=4=5 15-count? 1=4=3=5 11-count?
  
• On the same topic, do you also rebid 1NT on 2=4=2=5 too weak to reverse? If so, what is the range of this hand type in the 1NT rebid? And what about 2=2=4=5? 1=3=3=6 with concentrated values in the red suits?

I need to identify and program a complete list of hands compatible with the auction 1♣-(P)-1♠-(P); 1NT to run this simulation. If you happen to already have one would you please share it with me?

[Gilithin]

DavidKok, on 2024-January-02, 10:02, said:

I repeat/second this - could you describe which hands will rebid 1NT on a singleton for this simulation? The auction I'm limiting this to is 1♣-1♠. So in particular:

I need to identify and program a complete list of hands compatible with the auction 1♣-(P)-1♠-(P); 1NT to run this simulation. If you happen to already have one would you please share it with me?

Adam would be the best for this but I will kick off with:
If singleton 1444 or 1(34)5, no restriction on the singleton, 1336 will usually rebid 2♣ though a weak club suit and singleton A/K makes 1NT tempting so see if that makes a difference to results.
2(24)5 will usually rebid 1nt in 11-bad14 range. Most 14s will probably upgrade.
3(23)5 with a small doubleton is probably an important hand. I think results for both 1NT and 2♠ are useful. Adam's style is not to rebid 1NT with these.
When you extend this to a 1♦ opening, you have some additional hand types to consider, such as 3244 with a small doubleton. Since this is such a common hand type, it will need a separate sim. It would be a particularly interesting and counter-intuitive result if a 1♣ opening had a different result from a 1♦ opening.

I am probably missing some points but it's a start.

[awm]

For me, the shapes for opener on the auction 1♣-1♠-1NT look like:

3334
3433
2335
2344 (although sometimes I open 1♦ with this shape depending on suit quality)
2434
2245
2425
1435
1345
1444 (although sometimes I open 1♦ with this shape depending on suit quality)

There is no requirement that the singleton be an honor (this sequence is normal for me with a small singleton).

I'm sure at the table I will occasionally make other choices (like rebidding 1NT with 1336 or 0445 and bad clubs, or rebidding 1NT with 3244 when the hearts are AQ-tight) but the above is a good enough approximation for a sim.

[DavidKok]

Thank you. I'll plug 12-14 for the shapes without nine cards in two suits and 11-14 for those with, or should that be quite different?

The results are in: double dummy passing is better in this style in all situations with a 5332.
Weak suit, 6-8 HCP: -16.2, 32.8%/25.6%/41.6%
Weak suit, 9-10 HCP: -25.7, 40.2%/9.4%/50.4%
Strong suit, 6-8 HCP: -16.4, 37.1%/21.0%/41.8%
Strong suit, 9-10 HCP: -15.2, 40.6%/5.7%/53.6%

The presence of the several hand types with a singleton or two doubletons changed the numbers significantly, maybe also because their range is wider so their frequency is boosted. And since all hands with 3-card support and a doubleton were removed the chance that responder will find a fit by pulling is now very slim. In fact, I think that 5332 facing 3433 or 3334 probably plays better in 1NT anyway, so you are guaranteed to find a misfit or a worse contract on average.

[mikeh]

DavidKok, on 2024-January-02, 02:38, said:

As discussed earlier in the thread, I disallow rebidding 1NT on a singleton in my simulations. Opener is allowed to make a 3-card raise with an empty doubleton outside even if balanced. If there is a preferred set of methods you wish to compare I'd be happy to take a look at that, I started with the style I prefer. I had the concerns that not only do people have different sets of rules about rebidding 1NT (maybe the singleton needs to be an honour? Maybe it depends on the quality of your five card minor? And what if your hand is outside your 1NT rebid range but was unsuitable for a 1NT opening?), also I expected those hand types to be rare and presumably not move the needle much (and you'd lose out on the possible gains from bidding 2♠ every time partner does not have your singleton hand). So I started with the simple case I'm more familiar with, while offering to run others multiple times.

Also I think you would do well to reread the thread. What I am claiming is the following:

• Personally I think rebidding the 5cM always, or at least 'almost always', is best.
  
• There is classical advice about checking for suit quality and strength of the hand and outside shape and more, and my personal experience is that you are better off disregarding all that and just bidding.
  
• In an attempt to give a bit more support for this I've run a few simulations that might give a better insight into what's going on. This is not a "definite answer" - my answer was my first claim, preceding any computer stuff.

I even gave a possible explanation for how the classical advice might have come about despite, in my opinion, being worse. Personally I find discussions like this very difficult. I have an idea and opinion, supported by my experience at the table over a couple of years, and I present it here. In an attempt to provide more evidence for the belief I already hold I run a double dummy simulation, and offer people to check my code and to do repeat the simulations for other styles or bidding approaches. In thanks mikeh implies that I misread Bird & Anthias and now you are telling me my comments are not kosher. I'd be happy to remove all mentions of computer simulations and pick up the discussion from there, but rather I think the onus is on everyone else to give some more motivation for pass being better on average.

David, I did not intend to imply that I thought that you misunderstood Bird and Anthias…to the contrary, I assumed that you shared my view that, while they acknowledged that their DD approach biased their results, they had not gone far enough in recognizing some of the issues. My ‘favourite’ involves the common finding, in their analysis of opening leads, that an unsupported ace did better than one would expect. In real life, most partnerships have specific carding agreements should partner lead an ace…count or attitude generally. Leader will continue based on that signal….and since in many cases the signal is based on leader also holding the King, the signal will be actively misleading. DD analysis ignores signals altogether, so leader literally cannot be mislead and can survive the weird lead.

I’m strengthened in my dislike of DD analysis on this topic by Adam’s reference to real world expert declarer play, where 1N often fares better at the table than does 2M. That definitely correlates with my experience. I’ve heard, many times, club players lament that they hate playing 1N….personally, I think it’s often a great contract to play and a horrible one to defend unless one has a clear and visible path.

DD has its place in the study of bridge but it’s crucial to bear in mind it’s limitations. In addition to those above, constraints are critical.

In the real world, if I have 5=2=3=3 with xx in hearts I’ll pull to 2M for two reasons. I don’t want the opps balancing into a probable 8 card heart fit and I don’t want them leading hearts. Of course, partner may be loaded in that suit but he probably isn’t.

But give me 5=3=3=2 and I’m more inclined (if the hand otherwise looks right) to pass 1N. Simulating this kind of subtlety is tedious but in the real world it certainly plays a role for me and, I assume, others.