[Siegmund]

Great discussion points in this thread, guys. Especially the bit about 3N-vs-2N requiring a lower success rate than 3N-vs-1N never crossed my mind before.

[Siegmund]

So just for kicks I fed it to a simulator this afternoon. North as posted, South 2-4 spades 2-3 hearts 4-5 diamonds 2-4 clubs, 18-19HCP. After 3000 hands (so we have ±2% confidence),

Lose 3 imps (7 or fewer tricks): 24%
Lose 7 imps (8 tricks): 40%
Gain 10 imps (9 or more tricks): 36%

... which multiplies out to an imp expectation not significantly different from zero.

Seems to confirm that it's a pass at matchpoints, but at worse breakeven at IMPs, and probably a small winner at 3NT against human defence.

[maggieb]

Siegmund, on Dec 22 2009, 09:30 PM, said:

So just for kicks I fed it to a simulator this afternoon. North as posted, South 2-4 spades 2-3 hearts 4-5 diamonds 2-4 clubs, 18-19HCP. After 3000 hands (so we have ±2% confidence),

Lose 3 imps (7 or fewer tricks): 24%
Lose 7 imps (8 tricks): 40%
Gain 10 imps (9 or more tricks): 36%

... which multiplies out to an imp expectation not significantly different from zero.

Seems to confirm that it's a pass at matchpoints, but at worse breakeven at IMPs, and probably a small winner at 3NT against human defence.

220 is 6

[TylerE]

Siegmund, on Dec 22 2009, 09:30 PM, said:

and probably a small winner at 3NT against human defence.

Grossly underestimating the advantage declarer has in 3NT imo. Also, you're assuming the opponents always stop in 2NT, when IMO it's more likely that they bid 3 here.

How about if you assume the have superior methods, and thus bid 3NT 75% of the time when it's right, and only 25% of the time when it's wrong.

Then the table looks more like:

7 Tricks (24%): 25% of 0 (0), and 75% of -3, for -2.25
8 Tricks (40%): 25% of 0 (0) and 75% of -220 (-6), for -4.5
9 Tricks (36%) 25% of +480 (10) and 75% of 0, for (0), for +2.5

Net: -1.44

If we always pass 2N then:

7 Tricks: 25% of +3, 75% of 0, +.75
8 Tricks: 25% of +6, 75% of 0, +1.5
9 Tricks: 25% of 0, 75% of -480 (10), -7.5

Net: -1.92

What if we assume the opps are weaker, and have a 50/50 of getting it right:

If we always bid 3N

7 Tricks: 50% of 0, and 50% of -3, -1.5
8 Tricks: 50% of 0, and 50% of -220 (6), -3
9 Tricks: 50% of +480 (10) and 50% of 0, for +5

Net: +0.24

If we always pass 2N:

7: 50% of 0, and 50% of +3, 1.5
8: 50% of 0, and 50% of +6, 3
9: 50% of -10, and 50% of 0, -5

Net: -0.24

If we assume the opponents are very weak, and ALWAYS get it wrong then:


Bidding 3N

7: 0 8: 0 9: 10 Net: 3.6

Passing 2N

7: 3 8: 6 9: -10 Net: -0.48

And finally, against perfect oppoents:

3N: 7: -3 8: -6 9: 0 Net -3.12
2N: 7: 0 8: 0 9:- 10 Net: -3.6

I guess the moral of this is one of two things.... either A: it's too late for me to be doing IMP math, or B: Bidding 3N is always better, and it's fairly close (but real), except against very bad opponents where it pays off a lot.

[Jlall]

Siegmund, on Dec 22 2009, 09:30 PM, said:

So just for kicks I fed it to a simulator this afternoon. North as posted, South 2-4 spades 2-3 hearts 4-5 diamonds 2-4 clubs, 18-19HCP. After 3000 hands (so we have ±2% confidence),

Lose 3 imps (7 or fewer tricks): 24%
Lose 7 imps (8 tricks): 40%
Gain 10 imps (9 or more tricks): 36%

... which multiplies out to an imp expectation not significantly different from zero.

Seems to confirm that it's a pass at matchpoints, but at worse breakeven at IMPs, and probably a small winner at 3NT against human defence.

Siegmund, a couple of things:

One, as maggie said 8 tricks is a 6 imp loss for bidding game.

If you recalculate the EV, it is significantly +EV to bid now. However, your sim is probably overestimating the strength of partners hands; it includes 5422 19 counts (with 5 of the black suit) which likely would jumpshift to the black suit. Also, I think all 4252 18 counts would bid 1 or 2 spades, and some 2254 18's would bid 2C or 3C (some would bid 2N also). Not sure how you would simulate this. Also, I think some 5332 19 counts would be too strong for a 2N rebid and would open 2N.

So, that being said, I'm sure the EV of moving on is going down again. At the end of the day imo people underestimate declarer advantage.

On top of everything that has been said already, I think part of "being lucky" is putting yourself in a position to do so. If you are playing a knockout match against a pair who is constantly making every auction competitive, putting you to the test in the card play whenever possible, falsecarding frequently, doubling aggressively, bidding game aggressively, etc etc, you are going to be under constant PRESSURE, and you are just going to make mistakes. IMO I am "lucky" that my opponents make more mistakes against me than others because I am constantly putting pressure on them. That is a great meta to have going on, and it is part of winning bridge. Maybe some things are -EV in a vaccuum if my opps are going to react perfectly to it, but you know what, no opps I've ever played have reacted perfectly to constant pressure. If they are going to go low on marginally +EV things, and not test me with things that should be neutral EV but give me an opportunity to err, etc, I have a huge advantage because I will be less tired and less fried by the end of the match.

Sorry for that random rant, but I think it is an important part of bridge that is often overlooked, especially imp bridge.

Here's another thing that is overlooked. If I am playing and you make some stop in 2N making 2, I am not really demoralized. Even if I am conciously aware that that is going to be six imps out, eh whatever, 6 imps, and I didn't really do anything wrong. Similarly, if I go down in a thin game, no big deal, that is how I play and that's how the chips fall. Six isn't that much. This is how people think. However, if they happen to misdefend and let a game make, or even if I bid some 30 % game and it makes, oh no, thats a BIG SWING, that is demoralizing, some people feel pressured to get it back...and here we go...they're in another game... etc. A lot of times you can blitz an opponent on thin game hands if they mess one of them up and are not tough as nails psychologically, they're going to make more mistakes, etc etc. These kind of things do not happen if you are just playing 2N. Again, there is no pressure.

IMO there is a reason why playing Meckwell is just downright scary. There is just an overwhelming amount of pressure on you at all times, and they don't let up, EVER. If you look at the hands, it looks like they are in -EV contracts very often, so they should be losing to teams that are in +EV contracts, but that's just not what ends up happening. You have to play the match of your life to beat them. IMO it is a winning style to say to the opponents "Ok, if you play perfectly in the face of this blitz of fast competitive bidding and hyper aggressive game bidding, you will beat me. Go ahead and do it."

All of this is just more reason to bid game on these hands and not worry about it. The people who simulate hands miss all of this and wind up in 2N more often than they should, not understanding why their opponents seem to play well against them so often.

[shevek]

TylerE, on Dec 22 2009, 09:17 AM, said:

Red at IMPs? Why am I still thinking? I don't really expect this to make, but on a good day it will at we're certainly getting sufficient pot odds with the punt.

I see this a lot. Something is wrong when both partners look at the vul and form of scoring. There should be a "push seat" with the other bidding straight up and down. The problems occur because partnerships are never sure who is supposed to be pushing.
Yopu know the sort of thing, 1NT - 2NT - 3NT that turns out to be a "good" 15 opposite a "good" 7, with both citing "vulnerable at IMPs"
It would be good to have guidelines, which suggests a poll.
Who do you think is in the push seat in this auction?

[gwnn]

both players must bid to optimize the imp expectancy. the imp expectancy depends on partner's tendencies. of course this is a feedback system, it will need to be calibrated continually over time. the adjustments will be poisson processes (partner's catastrophic acceptance on that random 15 count ). it is not easy but it is pretty simple in theory.

[nigel_k]

shevek, on Dec 29 2009, 10:35 PM, said:

TylerE, on Dec 22 2009, 09:17 AM, said:

Red at IMPs? Why am I still thinking? I don't really expect this to make, but on a good day it will at we're certainly getting sufficient pot odds with the punt.

I see this a lot. Something is wrong when both partners look at the vul and form of scoring. There should be a "push seat" with the other bidding straight up and down. The problems occur because partnerships are never sure who is supposed to be pushing.
Yopu know the sort of thing, 1NT - 2NT - 3NT that turns out to be a "good" 15 opposite a "good" 7, with both citing "vulnerable at IMPs"
It would be good to have guidelines, which suggests a poll.
Who do you think is in the push seat in this auction?

This is a good question.

I have always thought that the first player to limit their hand should not push. They just accept a game try with an above average hand and refuse with a below average hand. So for example after a 1NT opening only responder pushes and after a single raise of a major only opener pushes. Otherwise yoyu do end up with those silly situations where a 1NT opener nearly always accepts an invite. But if neither hand is limited you can each stretch by half the total amount that is justified.

[1eyedjack]

I am wondering whether, on interpreting the simulation, you should consider the possibility that the opponents are close to doubling, and will double 3N when they would not dream of doubling 2N. I venture to suggest that if they do double then you should assume that the contract is failing.

This may point in favour of passing the 2N, although personally I play 2N rebid as 18-20 (and therefore forcing) rather than 18-19, so it is not a judgement problem for me.