[hotShot]

based on a very small sample (200 deals) DD-analysis gives:

opener 15-17 HCP, no 5card major, 2+ cards in any suit
(chances are: 15: 44% 16:33% 17:23%)
responder 8-10 HCP, any shape

HCP    8- tricks  9+ tricks
23         78%        22%
24         75%        25%
25         42%        58%
26         25%        75%
27 *       14-%      86+%   
*insignificant number of deals

This indicates that somewhere between combined 24-25 HCP there is the break even point to bid 3NT.
Vul.@imps you can risk 35% contract to break even, non vul it should be around 43%. Playing MP you should be better than 50%.
So responder should bid game with good 9+.
If responder holds bad 8 the chances are less than 23% to have combined 25HCP and 55% to have less than 24.
So as Hannie recalled, only on a small range around 8+ and 9 an invitation makes sence.

[FrancesHinden]

While I am usually hugely wary of drawing any conclusions from this sort of analysis, if these figures are indicative of real life, then adding just one extra HCP from 24-25 makes a huge difference in whether 3NT is making or not....

This reflects my belief that for balanced hands rather than source-of-tricks hands a count-your-HCP invitation is a useful tool (where, of course, A109 K10 K109 AJ109x is a maximum and KQ AKJ J432 K432 isn't)

It's not difficult to come up with hands where that one extra jack makes all the difference. Axx opposite Kxxx is 36% for 3 tricks; Axx opposite KJxx is over 75%. Jxx opposite Qx is a stop, xxx opposite Qx usually isn't.

This also posts up an advantage of using Stayman to promise a 4-card major. After 1NT - 2C - 2any - 2NT, opener knows that responder has a least one 4-card major (exactly which one if opener responded in the major) which will help him evaluate his holding.

[david_c]

FrancesHinden, on May 23 2006, 02:15 PM, said:

While I am usually hugely wary of drawing any conclusions from this sort of analysis, if these figures are indicative of real life, then adding just one extra HCP from 24-25 makes a huge difference in whether 3NT is making or not....

hotshot did say that the sample size was very small. I believe that the results are correct in that the break even point is somewhere "between" 24 and 25 HCP, but the difference between 24 and 25 is much bigger than it should be.

[fred]

Hannie, on May 23 2006, 12:13 PM, said:

fred, on May 22 2006, 02:06 PM, said:

We have actually made a change so that:

1NT-2NT=natural invitation

Fred, I'm wondering how often you use this sequence. What would you say, once every two sessions, more, less? Would it be possible to post some recent hands where you or Brad decided to bid 2NT? Would you have done worse if you had to choose between pass or blast?

I realise that I'm asking for quite a lot, so please ignore me if you don't have the time to answer (also, the last question seems quite hard to answer objectively).


I find this subject very interesting. I recall reading a double dummy simulation that said that the natural invite is only useful with hands in a very small range, something like 8.5-9 points (how these points were counted I don't remember, and is not so relevant imo). With less, it is better to pass, with more, it is better to blast. I wonder if the results of this double dummy simulation correspond to your (and other's) real-life experience.

Once every 2 sessions sounds about right. I don't keep any kind of record of our results so it is really impossible for me to give you a more accurate assessment of this or to let you know how well this bid works in practice. My sense is that whether or not we include the 2NT invitation in our methods is not a big deal one way or the other.

Brad and I do open 1NT more than any other (strong notrump) pair I know. So having a 2NT invitation available is likely more useful for us than it would be for most other pairs.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[hrothgar]

hotShot, on May 23 2006, 03:58 PM, said:

based on a very small sample (200 deals) DD-analysis gives:

Couple comments about this analysis:

1. There has been a fair amount work trying to characterize the extent to which double-dummy hand analysis can be generalized towards actual declarer play. I don't think that there is any real consensus in this area, however, a lot of folks believe that double dummy analysis is biased in favor of the defense.

Note: this doesn't (necessarily) negate the basic point that there is some quantitative break point at which the chances of making 3NT increase substantially. However, this point might be a bit lower than this chart suggests.

2. A lot of this discussion has focused on characterizing the difference between slow informative auctions and simply blasting to the right contract. One would expect that the differences between single dummy and double dummy results would be most significant when studying this very problem.

I posted a couple threads on rec.games.bridge studying these topics. When I looked at the work that Bob Richardson was doing I was most interested in his efforts to calculate whether the difference between the double-dummy versus the single dummy spread can be correlated with auction length.

[dogsbreath]

re fred's post above..
'But you are right that I feel pretty strongly that, if you have to go through Stayman to invite in notrump with no 4-card major, that you should never use that sequence (just guess between passing 1NT and bidding 3NT).' ....

Is/should your approach be affected by who your partner is? I ask this because one of my 'live' partners routinely accepts ALL invites ..as this makes 'invitational' sequences somewhat redundant it seems sensible to find some other use for the bid.

Rgds Dog

[kfgauss]

FrancesHinden, on May 23 2006, 11:54 AM, said:

I'm confused by what's special about this.

Most people who play 2C as 'Stayman', even if not promising a 4-card major, still play that 1NT - 2C - 2M - 3NT promises 4 of the other major (what other reason can responder have had for bidding 2C?).

Whether or not you can afford for opener to show one 4-card major or the other 'at random' depends on the rest of your methods. In particular, you will find slam bidding harder. Maybe it's worthwhile for the extra 'confusion' generated when playing in 3NT, but nothing in life is free..

I think this is just another typo and 1NT - 2C - 2M - 2NT was the intended sequence. Then with 4 of the other major too you can freely bid 3/4oM when playing that stayman promises a 4 card major.

Andy

[fred]

dogsbreath, on May 23 2006, 06:26 PM, said:

re fred's post above..
'But you are right that I feel pretty strongly that, if you have to go through Stayman to invite in notrump with no 4-card major, that you should never use that sequence (just guess between passing 1NT and bidding 3NT).' ....

Is/should your approach be affected by who your partner is? I ask this because one of my 'live' partners routinely accepts ALL invites ..as this makes 'invitational' sequences somewhat redundant it seems sensible to find some other use for the bid.

Rgds Dog

I would suggest instead that you either have a talk with your partner or find someone else to play with. Accepting every invitational bid you make shows a lack of respect for your judgment.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[foo]

Blofeld, on May 23 2006, 06:51 AM, said:

FrancesHinden, on May 22 2006, 11:00 AM, said:

I repeat: if 3NT makes much more frequently than it goes down (at IMPs) then you aren't bidding 3NT enough.  (and saying 'no offense' before making an offensive comment doesn't stop it being offensive.)

Is this true?

We should certainly be bidding 3NT contracts that are considerably more likely to fail than to make. But we should also be bidding our 100% 3NT contracts. I haven't done any calculations, but my guess is that I'd expect over half of 3NT contracts to make. Of course, it also depends on what 'much more frequently' means.

Frances is 100% correct. At IMPs =bid your games=.

A game that makes > 3/8 is a win in the long run Red at IMPs.
A game that makes > 5/11 is a win in the long run White at IMPs.

Assuming your choices are -1 or make, making > 2/5 of your X'ed games is a win in the long run at IMPs regardless of the colors.

The flip side of this is that if you are not going minus more often than you are going plus, then you are not bidding game enough at IMPs.

[foo]

Side note on using 1N-2S! as mss.

The "old fashioned" way of using mss to only show strong D+C two suited hands interested in slam does not come up frequently enough for it to be batter than 4 suit xfers.

If your choices are "Strong only" mss vs. 4 way xfers, play 4 way sfers.

The "modern" way of using mss is to use it for both very weak hands where playing 3m is likely to be much better than 1N =and= very strong minor two suiters interested in slam.

1N-2S!; Opener now picks a minor or bids 2N if they can't choose.
If Opener picks a minor, =any= subsequent bid by Responder shows slam interest.
If Opener bids 2N, then 3m by Responder is To Play and 3M by Responder shows slam interest.

The simple way to play 1N-2S!;any-3M is as showing shortness in the Major. This has the pro of being easy to remember but the con of making it easy for Them to know what to lead, or X to show the suit to suggest a sacrifice, or etc.
Thus the better way to play 1N-2S!;any-3M is that it shows shortness in the =other= Major. This makes it much harder for Them to enter Our auction safely. OTOH, this "better" way can be harder on partnership memory.

In any event, I suggest examining "weak or strong" mss if you are looking for 1N response structure suggestions.

[david_c]

foo, on May 25 2006, 12:18 PM, said:

The flip side of this is that if you are not going minus more often than you are going plus, then you are not bidding game enough at IMPs.

That doesn't follow, even when vul at IMPs. If you were able to bid perfectly - that is, bid all games with a greater than 3/8 chance of success and none of the others - then you would expect to make much more than half of your games.

[hotShot]

david_c, on May 25 2006, 02:21 PM, said:

foo, on May 25 2006, 12:18 PM, said:

The flip side of this is that if you are not going minus more often than you are going plus, then you are not bidding game enough at IMPs.

That doesn't follow, even when vul at IMPs. If you were able to bid perfectly - that is, bid all games with a greater than 3/8 chance of success and none of the others - then you would expect to make much more than half of your games.

I think you are wrong David.

If i bid all games that have a 40% chance to make, i'll loose 60% of these games.
If i win more than 40%, either my estimate is wrong or most of my opps belong in a lower league.

[david_c]

hotShot, on May 25 2006, 02:16 PM, said:

If i bid all games that have a 40% chance to make, i'll loose 60% of these games.

Let's say that on ten hands your chances of making game are:

hand 1: 10%
hand 2: 20%
hand 3: 30%
...
hand 10: 100%.

Then your best strategy when vul at IMPs is to bid game on hands 4 to 10, right? So that's seven game contracts, and, of those seven contracts, on average you expect 70% of them to make.

OK?

[hotShot]

You imply hat 100% games have the same frequency as 40% games. If this would be the case you are right.
This bridgebrowser query result says that 50+% of the boards at least one side can bid and make game. (Double dummy results say 60% game can be made, so the difference can't be bid.)

I guess that means that looking at all games one has bid, there should be more games made, than down.
While looking only at the close games, going down slightly more often than making is best.

[foo]

Let's do a thought experiment. I give you a board where you are in 3N Red at IMPs and have 8 top tricks and a suit combination that has 3/8 of giving you your 9th trick. There's nothing else to the play.

If you make, you get $10. If you go -1, you lose $6 .

Now let's say you play 8 such boards in a row: You make 3 and go down in 5:
3*$10 - 5*$6= 0
You have broken even despite the fact that you went down far more often than you made.

A analogous thought experiment will show that White at IMPS you only have to make 5 of every 11 games bid to break even.


QED: If you make most games you bid at IMPs, then you are not bidding enough games.

[david_c]

hotShot, on May 25 2006, 03:21 PM, said:

You imply hat 100% games have the same frequency as 40% games. If this would be the case you are right.

I just made the numbers up, I don't claim to know what the exact distribution of probabilities is. You could do a simulation to get a better idea. I think that an average of about 70% sounds about right though, and is possibly even a little conservative - note that there are a lot of hands where game is laydown.

Quote

While looking only at the close games, going down slightly more often than making is best.

This is vacuously true: if you restrict your attention to hands which have a chance of about 3/8 of making, then it should not surprise you that about 3/8 of them turn out to make!

[bid_em_up]

david_c, on May 25 2006, 08:30 AM, said:

hotShot, on May 25 2006, 02:16 PM, said:

If i bid all games that have a 40% chance to make, i'll loose 60% of these games.

Let's say that on ten hands your chances of making game are:

hand 1: 10%
hand 2: 20%
hand 3: 30%
...
hand 10: 100%.

Then your best strategy when vul at IMPs is to bid game on hands 4 to 10, right? So that's seven game contracts, and, of those seven contracts, on average you expect 70% of them to make.

OK?

This logic is somewhat flawed as the discussion pertains to when to bid game with borderline hands.

In other words, 10 hands:

Hand 1: 40%
Hand 2: 40%
Hand 3: 40%
...
...
Hand 10: 40%.

By bidding these games, you should win approximately 40% of them, and lose 60%. It is my belief that the actual ratio of bidding a 40% and it actually making is slightly higher, since these numbers usually fail to take into account any defensive errors that might occur.

[bid_em_up]

david_c, on May 25 2006, 09:45 AM, said:

This is vacuously true: if you restrict your attention to hands which have a chance of about 3/8 of making, then it should not surprise you that about 3/8 of them turn out to make!

David, that is all you expect them to make, and all you need them to make in order for it to be profitable to bid these games.

[david_c]

Oh, this is just getting silly.

All I am saying is that foo's statement,

"If you make most games you bid at IMPs, then you are not bidding enough games."

is false. I'm afraid I can't explain this any better without repeating myself. I will now have to stop posting in this thread before it drives me insane.

[jdonn]

foo, on May 25 2006, 09:43 AM, said:

Let's do a thought experiment.  I give you a board where you are in 3N Red at IMPs and have 8 top tricks and a suit combination that has 3/8 of giving you your 9th trick.  There's nothing else to the play.

If you make, you get $10.  If you go -1, you lose $6 .

Now let's say you play 8 such boards in a row:  You make 3 and go down in 5:
3*$10 - 5*$6= 0
You have broken even despite the fact that you went down far more often than you made.

A analogous thought experiment will show that White at IMPS you only have to make 5 of every 11 games bid to break even. 


QED: If you make most games you bid at IMPs, then you are not bidding enough games.

Your percentages are essentially right, but that experiment makes NO SENSE. I'll reproduce it and show why.

Let's do a thought experiment. I give you a board where you are in 3N Red at IMPs and have 8 top tricks and a suit combination that has 1/100 of giving you your 9th trick. There's nothing else to the play.

If you make, you get $99. If you go -1, you lose $1.

Now let's say you play 100 such boards in a row: You make 1 and go down in 99:
1*$99 - 99*$1= 0
You have broken even despite the fact that you went down far more often than you made.

QED: If you make more than 1% of games you bid at IMPs, then you are not bidding enough games.


Pretty ridiculous, right?


Even though your percentages are not far off, your QED is completely 100% false.

EDIT 1: What I'm calling percentages were not actually percentages but you know what I mean, the 3/8 and 5/11.
EDIT 2: I wouldn't blame David for going insane reading some of the posts in this thread, especially your repeated incorrect assertion that you should go down in more games than you make.