[Old York]

nigel_k, on Nov 3 2009, 02:10 AM, said:

I ran a simulation on this assuming the 2NT bidder has four spades, but not five - so you use Puppet Stayman and bid 3NT over 3♦ thus missing a 4-4 spade fit but not a 5-4. Out of 10,000 hands:

5188 times 4♠ was a better matchpoint score than 3NT
4244 times 3NT was a better matchpoint score than 4♠
1611 times 3NT made and 4♠ failed
894  times 4♠ made and 3NT failed

So the 4-4 spade fit has the edge over 3NT double dummy at matchpoints. The defenders are more likely to err against 3NT, especially on the opening lead, plus looking for a 4-4 spade fit will reveal extra information about the declaring hand when it doesn't have spades. I don't know if these factors are enough to make 3NT better but I doubt it.

3NT fails less often so is better at IMPS.

Many, many thanks for taking the trouble to do this research, it is certainly thought provoking and I will look into this much deeper.
Again, this highlights a big difference between Imps and Matchpoints

Cheers, I owe you a drink
Tony
afterthought: If the make/fail is double dummy, maybe the 1611 is actually higher in reality

[gnasher]

The number of tricks available double-dummy isn't the only factor. It's harder to defend correctly after
  2NT-3NT
than after
  2NT-3♣
  3♦-3NT
or
  2NT-3♣
  3♥-3NT
or
  2NT-3♣
  3♥-3NT
  4♠
or
  2NT-3♣
  3♠-4♠

It's different if you can investigate the spade fit without having to reveal anything else about opener's hand.

Edit: Having reread nigel_k's post, I see that he'd already mentioned this factor. As you've probably gathered, I disagree with him about its significance.

This post has been edited by gnasher: 2009-November-03, 03:21