[Gerben42]

Hi, I'd be interested to know what is the average score for balanced hands when opening NT or not, i.e.

12 HCP: average score of those who opened 1NT
12 HCP: average score of those who opened 1x

for 12 - 17 HCP.

Then one can see if it pays to play a weak NT or not

My prediction is that all HCP counts do better when opening NT than when not opening NT. Which basically tells you it's the other way around

[inquiry]

In the "old" days of bridgebrowser what you had to do for this kind of analysis was first search for balanced hands with 12 points, then use bid analysis to see how opening 1NT does compared to 1C and 1D. Then do a new search with 13 hcp, etc.

So this would be 6 different searches. About 6 months ago (maybe a little longer) bridgebrowser added a new feature to bid analysis: plotting.

Now you can search for "balanced" or "semi-balanced" hands with 12-17 hcp then use bid analysis, click on the 1NT opening bid, and display the number of hands with 12, 13, 14 etc hcp. Simply by clicking on the bar for 12 hcp, you will then see the average imps and mp, repeat for each hcp. Then click on 1C or 1D and repeat the plotting step. This is a huge time saving, I will run this analysis later this AM.

[inquiry]

I did a quick search looking for 1C to 1NT openings with balanced hand (at most four cards in either minor and 4432, 4333, or 5322 distribution) with 10-17 hcp. You see a dramatic schewing of the results for 15-17 1NT in the data.

I think for the moment, this data is inconclusive, simply because the number of weak Notrump opening bids are too low to draw a reliable conclusion. I will redo the test with more focused criteria to find higher frequency of bids. It is relatively unfair the way I searched.

For 1NT openings (number of hands in bracketts), and 1C openings
10 hcp = (6)+2.22 imps, (2) 50.00%//(28)+0.51, (23)25.83//
11 hcp = (13) +2.38, (15) 52.15//(253)-0.26, (166)49.99//
12 hcp - (34) +0.05, (15) 55.00%//(1276)+0.04, (867)50.87//
13 hcp = (47) +1.50, (49) 43.42//(1242)0.00, (903)50.83//
14 hcp = (230) +0.94, (90) 53.69//(1219)-0.09,(511)48.45//
15 hcp = (1176) +0.13, (471) 49.97//(166)-0.35,(70)48.97//
16 hcp = (1492) +0.05, (845) 51.56//(95)0.37, (52)49.56//
17 hcp = (884) +0.11, (448) 50.47//(63)-1.07, (40)50.40//

[Flame]

13-15 is the winner, 10-11(maybe 9-11) looks good too.

[awm]

As others have realized (but not fully explained) this isn't the sort of data you can read much of anything into. Besides the fact that there are not many instances of weak NT openings, there are causality issues. Basically:

(1) In the main bridge club, a huge advantage accrues to regular, established partnerships. This is because they tend to be on the same wavelength in bidding and on defense, and because they are usually playing more seriously. It's also possible that they tend to be better players in general.

(2) Who is opening 1NT on 10-12 points? You don't normally miscount your high cards by a king or more, and this is not a typical psych or tactical bid if your agreement is 15-17. There aren't really any "cookbook" systems with 10-12 notrump that you can sit down and agree with a pickup (in fact the only 12-14 I can think of that's remotely common is ACOL). So virtually all the 10-12 NT openers will bne established partnerships, who will be winning huge amounts of imps anyway (see 1).

(3) Also note that there are surprisingly many 1♣ openings on 15-17 hcp. If this stuff was all based on "agreements" you'd expect that the people who play weak notrumps might (but also might not) open 1♣ on the strong hands, and since weak NT openings are substantially more frequent than strong NT openings, not to mention that (playing ACOL or the like) some strong NTs might open 1♦, you would expect to see a lot more 1NT opens on 10-14 points than 1♣ openings on 15-17. However, the two numbers are roughly equal. This indicates (to me anyway) that a fair number of the 1♣ openings on 15-17 are beginners, who often "forget" to open 1NT especially with 5332 patterns including a minor. Obviously these folks are unlikely to score well (for other reasons, and because it's problematic to fail to open 1NT when you're balanced in your agreed range).

Anyways, while bridgebrowser is an interesting tool, I don't think you can really use it to evaluate conventions in this way, exactly because of this kind of bias. If you could restrict it to "serious" tournaments (which perhaps HomeBase will eventually allow you to do) then it might be possible, but based on main bridge club results it really isn't.

[han]

I agree with Adam, it is very hard to draw reliable conclusions from bridgebrowser for questions like this. On the bridgebrowser website Tim Bourke claims that bridgebrowser tells him that "natural weak" is the best use for a 2D opening. I think that it is impossible to make a reliable test that gives a result like this, but perhaps Tim knows something that I don't.


I started a thread on the homebase forum about this problem, but unfortunately nobody reacted to the point I was trying to make:

http://forums.homeba...p?showtopic=170

[awm]

I seem to recall a similar "result" indicating that a weak 2♥ was a terrible preempt, because people who opened it (on average) lost imps. But the interesting thing about this was, people who passed the potential weak 2♥ openings also lost IMPs, in fact even more so than the people who opened. Who was winning IMPs? The people who opened 2♦ multi!

So okay, maybe 2♥ openings aren't necessarily bad, but obviously it's better to open 2♦ multi than 2♥ weak right? Well again, no evidence to support this. People playing 2♦ multi are almost always established partnerships (and probably decent players) whereas those opening 2♥ weak are more of a mixed bag....

[CarlRitner]

awm, on Mar 24 2006, 03:05 PM, said:

(3) Also note that there are surprisingly many 1♣ openings on 15-17 hcp.

Excuse my ignorance, but is BridgeBrowser able to tell you which of these are due to a strong club system? It seems if one cannot separate systems in a structured query, the data becomes much less valuable.


Carl

[joshs]

All of these questions are complicated to answer.

Let me give an example. You may not agree with the claims in my example, but I hope you will understand the point that I am trying to make that is independant of this particular example.

Claim 1: In a good field you will lose imps on average when you open a multi 2D

Claim 2: In a good field playing multi 2D will gain you imps on average

You might read these and say, what the heck is he talking about? How can 1 and 2 both be true?

This comes from the following claims:
A. Multi 2D loses slightly compared opening thos same hands weak 2's
B. Multi 2D gains compared to passing with those weak 2's
C. Having freed up 2H and 2S for other use, you gain when you open those bids relative to a pass, or relative to including those hand types in some other bid.

So a bid might lose imps on average and still be a winner, since it fills a hole in your structure such that pass would be a bigger loser.

[inquiry]

It is possible to do a different kind of test with bridgebrowser.

For instance, you can look for hands with 6D and 6 to 10 hcp or 6 to 11 hcp and see what bids works out best.

Or a balanced hand with 12 hcp and see which one works out best (search by hand type, not opening bid). You can restrict this further by seat (is it better to preempt in 1st seat or 3rd or 2nd, etc), and things like vulnerabilty.

So I did a quick search with the following criteria, Dealer had 6D at least KJxxxx or better and 6 to 11 hcp. I then found all hands played in BBO tournament between Jan 1 and Jan 11 2006 meeting those requirements, and found out what the different opening bids (or pass) was worth. Here are the results:

Pass (2153) -0.05 imps, (788) 47.97%
1D (717) +0.12 imps, (303) 50.49%
2D (570) +0.17 imps, (249) 53.29%
3D (238) +0.22 im[s, (64) 58.09%

75% of the 1D opening had 11 hcp and 22% had 10 hcp... Only 8% of the 2D opening bid had 11 hcp, and about 5% of the 3D opening bids had 11 hcp.

So what this says is BIDDING is better than passing. These were all on the same hands with exact requirement (dealer, hcp, quality and lenght in diamonds) and I ignored hands that could open other suits (like hearts)...

In second seat the odds chage slightly...
Pass was all to the negative...as can be seen by the bids that were positive
1D (433) +0.57, (549) 51.24%
2D (707) +0.24, (405) 53.60%
3D (139) +1.09, (72) 47.77%

[hrothgar]

inquiry, on Mar 25 2006, 01:18 AM, said:

So what this says is BIDDING is better than passing. These were all on the same hands with exact requirement (dealer, hcp, quality and lenght in diamonds) and I ignored hands that could open other suits (like hearts)...

Personally, I don't like to make claims of statistical significance without including little things like confidence intervals...

[hotShot]

joshs is right. THe andvantage of playing a strong club system is, that every other opening bid is limited. This is a big advantage in the battle for the partscore.
Note:
You can only make one bid, but there are usually plenty of bids you did not make. Each of these unmade bids, excludes handtypes and pointrages.
So you gain more information about partners hand, if you look at the bids partner did not make.

[joshs]

inquiry, on Mar 24 2006, 05:18 PM, said:

It is possible to do a different kind of test with bridgebrowser.

For instance, you can look for hands with 6D and 6 to 10 hcp or 6 to 11 hcp and see what bids works out best.

Or a balanced hand with 12 hcp and see which one works out best (search by hand type, not opening bid).  You can restrict this further by seat (is it better to preempt in 1st seat or 3rd or 2nd, etc), and things like vulnerabilty.

So I did a quick search with the following criteria, Dealer had 6D at least KJxxxx or better and 6 to 11 hcp. I then found all hands played in BBO tournament between Jan 1 and Jan 11 2006 meeting those requirements, and found out what the different opening bids (or pass) was worth. Here are the results:

Pass (2153) -0.05 imps, (788) 47.97%
1D (717) +0.12 imps, (303) 50.49%
2D (570) +0.17 imps, (249) 53.29%
3D (238) +0.22 im[s, (64) 58.09%

75% of the 1D opening had 11 hcp and 22% had 10 hcp... Only 8% of the 2D opening bid had 11 hcp, and about 5% of the 3D opening bids had 11 hcp.

So what this says is BIDDING is better than passing. These were all on the same hands with exact requirement (dealer, hcp, quality and lenght in diamonds) and I ignored hands that could open other suits (like hearts)...

In second seat the odds chage slightly...
Pass was all to the negative...as can be seen by the bids that were positive
1D (433) +0.57, (549) 51.24%
2D (707) +0.24, (405) 53.60%
3D (139) +1.09, (72) 47.77%

Give me Axx xx AQxxxx xxx and I bid m googles. What is m googles? That is a good question, since you are disassociating the meaning of the bid from the bid.

If you play Roth Stone and open 1D on this hand, I am sure you will get a bad result.

If you play KS and open 1D on this (promsing a weak NT or a sound opener with diamond length) I expect you to get a bad result.

If pass for you means " 8-10 HCP some 6 card minor". You will probably get a good result.

If you open 1S on AQxxx Axxx xxx x I am sure you will get a better result on average if you are playing precision than if you are playing SAYC.

If you play roth stone with 16-18 NT and pass a 4333 12 count, I expect that you will get a reasonably good result. If you pass the same hand playing 10-12 NTs, I expect a bad result.

In general, I expect you to get better results if your hand resembles what your partner expects from your bidding than if it does not.

Now, to answer a question like how valuable is a weak 2 bid in diamonds. You can't just say here are candidate hands, lets see how people did with them.
What you need to compare is performance by a pair playing a weak 2 in diamonds, with the performance of a pair not playing a weak 2 in diamonds.
This involves comparing results on all bids effected.

Example: You have no weak 2D bid, and you never promote those hands to 1D but occasionally you open them 3D. Thus the bids effected are: P, 3D. So you have to compare how the pair playing 2D as something else did with the 6 diamond hands, and how they did on hands with 7 diamonds that is a normal 3D opening bid, and how they did with passes in general
Also you need to do the opposite comparision for the bid or bids gained by freeing up 2D and the other bids effected by that.

For instance if you want to compare a weak 2 in diamonds with flannary you need to look at:
P (overloaded by the flannary pair)
1h Opening With and Without flannary
Each of the respective 2D openings
3D opening

Also, if playing flannary effects what you open with 5S and 6H minimum, you might want to include 1S

Doing all this is complicated. For instance the gain playing flannary is principlly in your 1H opening not in the 2D opening.


If you can partition the hands into: plays flannary vs plays weak 2 in diamonds you can add the +/- imps from the 4 bids effected above for the two sets of partners and see the result. This of course assumes that partnerships playing flannary are on the same level on average, and have the same level of partnership on average as the pairs palying a weak 2D, and I doubt that is true, so probably you need to control for that...