use of computer simulations

[ceblair]

It seems to me that one could use computer simulations to look at a lot of bidding issues. Example: I may be the last person on the planet who thinks

1NT on a 5-card major should be a last resort. I've been too lazy to do it,

but I could take a typical such 1NT opening, say S Kx H AQxxx D QJx

C Axx, deal out the remaining cards 100 times, and compare likely results of 1NT

vs 1H. I realize there would be a lot of subjective cases, e.g., likely

competition or lack of it, but it still might be worth doing.

 

Has anyone done something like this? (Not necessarily this specific issue.)

[helene_t]

There was a guy from New Zealand who wrote in Bridge World a couple of years ago about whether you should seek a major suit fit after p opens 1NT and you have a balanced hand. The conclusion was that bidding 3NT (not using Stayman or Jacoby) was often best, especially at matchpoints.

 

That was a relatively simple problem, but even so he made a number of simplifications. For example, if one has the agreement that a game-going 5M332 never starts with a transfer, one might wonder what the sequence 1NT-2♦-2♥-3NT shows, maybe one should assign a specific meaning to it so that the sequence 1NT-2♦-2♥-3m becomes more specific.

 

As for the question of whether one should open 1NT with a 5-card major, the issue is much more complicated. For example, if partner opens 1♥ and it starts 1♥-1♠-2♣-4♣, it makes a difference whether the 2♣ bid could be based on a balanced 16-count, or it promises 4+ clubs. Even if opener has 4+ clubs, it matters whether responder knows that.

 

Simulations can surely be illuminating, but they will (almost) always be based on a number of dubious assumptions, so they will rarely convince people who are already prejudiced.

[nigel_k]

The problem is that the computer cannot accurately simulate human behaviour in complex situations. In the example of opening 1NT with a five card major, you would have to make all kinds of questionable assumptions about subsequent choices. Likewise, with bypassing Stayman or a transfer, you may reach a theoretically inferior contract but how can the computer work out whether this is offset by your gain from depriving the defenders of information?

 

I think you want to simulate situations where these factors are minimized. For example, I did some analysis of when to make a game try after 1M-2M.

[hotShot]

It is easy to run a simulation using the available software, what is not easy is to understand what the simulation is actually telling you.

 

Lets assume that you open Hands with 5332 or 5422 shape with:

1NT if they are 15-17

1m => 2NT if they are 18-20

2NT if they are 21-22

 

and lets assume that 100 simulations show that you will make 3NT more often than 4M.

 

What does that tell you?

 

I can tell you what it does not prove.

It does not prove that it is a superior treatment, because to you will have to analyze what happens to your 1M openings. They used to be 12-21 HCP with 5+M cards with a reasonable frequency. Since you took away a lot of hands the frequency dropped and they are now 12-14 or (seldom) 12-21 unbalanced. The average strength of you 1M bid drops and you will miss a lot of partscores and games just because you did not expect your partner that strong.

You also need to investigate what happens to those hands where a 4-4 fit in the majors leads to a 4M contract that is better than 3NT. Since the NT opener has a 5 card major quite often you need to play some kind of Puppet-Stayman. This will make it harder for your side to find those 4-4 fits esp. if opps interfere.

You will need to memorize some complex followups over 1NT esp. after opps interference. Remember that opps could play Cappelletti, DONT, Lionel,...

The computer is unable to simulate how often you will forget or misapply your complex system.

If I took more time I would come up with more side effects you will have to analyze.

 

 

Now to answer the question what a simulation like the one you suggested tells you:

 

Basically the answer you get is, that with a given strength it's easier to make 9 tricks than 10 tricks. With a more detailed look you might even find, that suit quality is a factor.

 

I bet you knew that without a simulation.