[pbleighton]

Ben:

How difficult would it be to see the results when the 5cM has 7 hcp or better?

Peter

[inquiry]

pbleighton, on Aug 13 2006, 07:43 AM, said:

Ben:

How difficult would it be to see the results when the 5cM has 7 hcp or better?

Peter

This is not hard at all. One can pose the questions anyway that you like. By any variable that you like.

The main problem you have to watch out for is to make sure the population of hands you examine are the same. There are two ways to do this. The forst is to describe some type of opening hand based upon hcp and shape and look at the results. For example, describing the dealer as 15-16 with 5332 and a five card heart suit. There is no selection based upon what the final results would be. ALL hands that are that shaped are examined and you only look at the final outcome (opening 11-13 hcp balanced hands 1NT seems to work out better than any other option, for example).

One way you could do this is to look for all 4H and 3NT contracts with an 8 card heart fit. But, the population of hands would not be the same. The 4H hands would include sacrafices, for example, or contracts where one partner has 8 hearts and the other a void. The program will happily give you the average tricks and average results. This select the same population of hands is very important if you are going to draw conclusions. One way to do this is to describe a combined team hcp range, and find all the contracts of interest. Then find the hands with 5-3 fits, then look at the quiality of the five card hand (make it 5332), then examine how the different responding hands opposite it did. All this can be done with one search.

Your question is a little different. I think you are asking how different hands did with from 7 hcp upwards (and 5332 with five hearts). I assume you mean in 3NT versus 4H. Of course, some of the 5332 hands will be opposite 6 hearts and a void, and will always be played in hearts, and many of them will be opposite a doubleton or less in hearts and essentially never be played in hearts. So we need to refine this search a little better. The solution ot this problem is actually quite easy but requires a slight refinement in the process.

First, describe one hand (Dealer is fine) as 5332 with 5 hearts. And give him from 7 hcp to 29 hcp. Then you might describe the team total hcp as something not outrageous. The reason to limit team is if you oare going to examine 4H versus 3NT, having 38 combined hcp would present some problems. You don't have to cap the team HCP, but I will. I will make them from 23 to 29 combined hcp, thus this defacto limits dealer to 29. Now this search will not restrict you to heart fits, so the initial search will only look at hands where opener side plays in 4H. In doing this search, I will create what is known as a "board file". All this board file is only a file with the board numbers on which A) dealer has from 7 to 23 hcp, 5-3-3-2 patter with five hearts, and C) one pair at least played in 4H's. Then, I will use this board file to examine the effectiveness of 3NT versus 4H (as other contracts as well, if we like). This includes looking at the effect of 5-3 major fits versus other fits.

Next post will be the results of such a study.

[ABadPlayer]

Wow....great work Ben. Very intersting thread indeed.

[whereagles]

ABadPlayer, on Aug 14 2006, 09:12 PM, said:

Wow....great work Ben. Very intersting thread indeed.

Indeed. Ben, you should put this in a more easy-to-read fashion and publish in a magazine

[awm]

While this is definitely interesting, I believe there was an article in the bridge world a short time ago exploring the merits of 3NT holding a 5-3 major fit. The methodology was pretty similar to Ben's and the result was that 3NT is "usually better" than 4M especially when holding two balanced hands (no singletons/voids). It seems like while the initial question was "should we open 1NT with five card majors" the answer basically came out as "yes because you're more likely to miss a 5-3 major fit and play 3NT instead, and this is better" and some of the subsequent questions have just been exploring how much better it is with different hand types.

[inquiry]

This is a study of contracts with 5-3-3-2 distribution in one hand, with five hearts. The other hand has from 3 to 5 hearts, and played in 4H at least once. Further, the pair of hands had a combined 25 to 27 hcp. There was no control of which side opened the bidding, how the 25 to 27 hcp were distributed between the two partners (one could have 25 the other zero, for instance). There was 862 unique deals that met this requirement in the largest online BRBR database. Further, of these there were 57,704 unique auctions (average nearly 67 plays per board ... this was from okbridge in 2001).

Without any further division of the hands (by distribution or hand opposite 5332), with from 8 to 10 card heart fit, the blanked results (without breaking down for distribution or degree of fit) were:

There was 15044 imp auctions where openers side ended up in 4H, and those hands averaged 0.24 imps, and another 14,663 auctions in 4H at Matchpoints, averaging 51.14% The non-opening side also played in 4H on these fits with 24 to 26 hcp. They ended up winning an average of more imps and more matchpoints, presumably due to the fact that few bid the games in competition I say this because the average hcp for both was 24.84 (on all the hands playing 4H, rather opener or responder.. but remember, I forced the hcp to be from 24 to 26 hcp, so hands with more or hands with less are not considered). The imps for the non-opening side in 4H was 0.84 and the matchpoints was 53.35.

There were a lot fewer hand in notrump (given I forced an 8 to 10 card heart fit, this is not surprizing). 3NT averaged +1.54 and 55.73 when this side opened, and 1.23 and 60.21 when the other side opened. And this is BEFORE controling for shortness in the hand opposite the 5332 hand. This doesn't mean all hands are created teh same, however. The average hands for 3NT was just over 25 hcp, while as noted above, the average hand for 4H was 24.84. Also there were more than Since all these hands were played in 4H at least once, I decided to further define the hands. I forced all the boards to have been played not only at east once in 4H, but also at least once in 3NT. Thus any 4H hand for which there is not at least one 3NT comparison and vise versa will be eliminated from the study. By forcing at least one hand on each board to be played in 4H and in 3NT, the number of matching hands decreased to 667 boards (down from 862 hands). And the number of unique bidding sequences reduced by 10,000 to 45,704. Of course the 3NT finall contract stats didn't change. All the reductions were in 4H contracts were no one played 3NT.

The number of 4H contracts at imps (from opener side) was reduced to 11,121, and the average result decreased to 0.05 imps, and 50.59 MP. From non opening side, 4H decreased a bit to 0.50 imps and 53.44%.

Playing with a 10 card heart fit,
at 4♥ averaged 1.03 imps and 55.44%
at 3N averaged -0.14 imps, 67.53% (but only 9 and 11 hands -- note this means only 3 actual deals in one case and two in the other. The data is not real)

Playing with 9 card heart fit,
at 4♥ averaged 0.33 imps and 52.68%
at 3NT averaged 1.6 imps and 55.55%

Next time data on 8 card heart fits, and some hcp ranges for the 5332 hand. As well as stats on hands where there is singleton or void in one of the hands (not the 5332 with hearts, of course).

[ABadPlayer]

I'm curious as to what difference if any it makes when using a smaller NT range (as I prefer to). I'd assume that the difference between 3NT & 4M with no ruffing values remains conistent, but what about everything else? Is it still an overall positive or do you end in enough inferior partials to negate that? How big of a difference in terms of pre-emptive value does 1N really have over 1H/1S and does that play a significant factor?

[kenrexford]

This post has gotten way too long to read through the entire thing each time, so please excuse me if these Q's have been asnwered.

The decision between 1NT and 1♥/1♠ as an opening, for me, has been the ability to reach games. I have been concerned about opeing 1♠ with a 15-count and hearing a 10-count 1NT and missing game, for example. Has this issue been addressed already?

In other words, I'd be curious at the frequency of a 1NT opening resulting in a 3NT/4M contract off "X" tricks, compared with the frequency of a 1M opening ending up in 4M/3NT contract of "X" tricks, compared with a 1NT opening bid to a partscore when 400/420+ was available, compared with a 1M opening bid to a partscore when 400/420+ was available.

[inquiry]

Data for 5-3-3-2 hands with 25-27 combined hcp, when the partnership holds exactly a 5-3 heart fit. Again for this data, someone had to play 3NT at one table and someone else had to play 4H at another to be included. There where hundreds of matching hands with thousands of plays. This allowed furhter division based on other shape issues.

Lets start with when the hand opposite the 5332 is 4333. The stats were...
4333NT    IMPs        
61     Tricks     Score     IMP     HCP
Avge    8.6    263    0.34    25.01
Std Err    0.17    41    0.6    0.1
                
4333 NT    Matchpoints    
67     Tricks     Score     MP %     HCP
Avge    8.58    201    48.3    24.91
Std Err    0.17    41    4.49    0.1

You read this from left to right. the key is the average imps (bold) and matchpoints (bold). How did this compare with the contract on the same hands in 4♥? Let's see.

4333    5332♥    IMPs                
1805     Tricks     Score     IMP     HCP    Trumps    HP+Dist
Avge    9.46    210    -1.08    25.15    8    26.15
Std Err    0.02    7    0.11    0.01    0    0.01
                        
4333    5332♥    Matchpoints            
2014     Tricks     Score     MP %     HCP    Trumps    HP+Dist
Avge    9.36    175    44.54    25.09    8    26.09
Std Err    0.02    7    0.64    0.01    0    0.01


You may notice that in the 5332 versus 4333 with 8 card fit, the average number of tricks was higher to play in the 5-3 fit versus NT, but the net result for both matchpoints and imps favor NT. This paradox was reproducible in many different datasets.
5332 opposite 4432 (5-3 fit), played in NT (below) again had the same paradox when compared to played in hearts...
4432-5332NT    IMPs        
37     Tricks     Score     IMP     HCP
Avge    8.51    266    0.61    24.67
Std Err    0.23    56    1    0.1
                
4432-5332NT    Matchpoints    
42     Tricks     Score     MP %     HCP
Avge    8.76    254    50.42    25.04
Std Err    0.24    49    5.48    0.12


4432 [h]'s -5332    IMPs                
1366     Tricks     Score     IMP     HCP    Trumps    HP+Dist
Avge    9.48    236    0.23    24.81    8    26.81
Std Err    0.02    9    0.13    0.01    0    0.01
                        
4432 [h]'s -5332    Matchpoints            
1762     Tricks     Score     MP %     HCP    Trumps    HP+Dist
Avge    9.67    270    51.76    25.12    8    27.12
Std Err    0.02    8    0.69    0.01    0    0.01

The 5431-versus 5332 with 5-3 fit, the statistics swing completely into favor of the 5-3 major fit. Here is the data.
5431 – 5332NT    IMPs                
20     Tricks     Score     IMP     HCP    Trumps    HP+Dist
Avge    9.05    201    -0.44    24.75        
Std Err    0.36    64    1.14    0.15        
                        
5431 – 5332NT    Matchpoints            
13     Tricks     Score     MP %     HCP    Trumps    HP+Dist
Avge    8.38    111    40.04    24.3        
Std Err    0.52    115    7.96    0.67


5431 – 5332♥    IMPs                
720     Tricks     Score     IMP     HCP    Trumps    HP+Dist
Avge    9.61    226    0.09    24.7    8    27.7
Std Err    0.03    11    0.17    0.02    0    0.02
                        
5431 – 5332♥    Matchpoints            
333     Tricks     Score     MP %     HCP    Trumps    HP+Dist
Avge    9.69    251    48.86    25.02    8    28.02
Std Err    0.05    18    1.55    0.05    0    0.05

A couple of issues, these hand include 13 opposite 13, 5 hcp in the 5332 opposite 21 in the opposite hand and the like. That is, the fitting hands examined held 25 to 27 combined high card points, a 5-3 heart fit, ane the hand with five hearts was any 5332 distribution.

[Oren Goren]

inquiry, on Aug 9 2006, 08:03 AM, said:

The next study will examine the hands that would start 1H-2H at tables where 1 H was opened.

This may be too premature, given that I haven't finished the thread, but IMHO restricting the bidding on the heart openers to 1♥-2♥ sequences doesn't allow the full comparison.

Given that I (mis)understand the hands to be largely duplicates:

If the BridgeBrowser allows it the hand selection should be (1st and 2nd seat?) those that at some tables are opened 1♥ and at other tables 1NT - at least once each - without regard to how the bidding proceeded, including the final contract.

Certainly, the requirement of a 2♥ raise over 1♥ is not really relevant. What is it about the basic issue of ♥ versus NT that requires that limitation? What is it about the basic issue that enforces the final contract?

Is there anything in the study so far that says 1♥ is not superior in NOT reaching game?

Oren Goren

[Tcyk]

My favorite system of all time opened 1NT with a 5-card major but there were restrictions.
1. Our no trump range was 12 to 14 high card points.
2. The 5-card major was weak, perhaps as strong as QTxxx.
3. We did not have two doubletons
4. Opening 1M promised five cards that were better than QTxxx.

We also had an elaborate system of responses designed to handle these hands. Transfers were invitational or stop bids. Forcing hands began with Stayman followed by 2S over 2D or 2H and 3C over 2S. The advantage of the system was that we could expect a good suit after an opening bid of 1H or 1S and when a 5-card major was disclosed after opening 1NT we knew it had poor suit quality.

Anyway, it would be interesting to see the search results with hands that were in the 12 to 14 high card point range.

Like so many studies of this type, results are skewed by individual partnership methods. I am convinced that the good results obtained by mini-no-trump opening bids are because partnerships using these openings know how to handle them. Virtually every partnership can handle strong balanced hands without a 5-card major but how many of them actually have methods for those with a 5-card major?

Charlie

[awm]

Let me try to sum up the information I see from these bridge browser tests:

(1) In general opening 1NT with a 5332 pattern when in range (including five-card major) tends to fare better than opening 1M with such a hand.
(2) While such claims are often biased by the skill level of the players involved, this one seems unlikely to be so. This is because, when the 1NT and 1M openers end up in the same contract, the results are similar. The difference in results for opening 1NT and 1M appears to be because different final contracts are reached.
(3) The difference in results for opening 1NT and 1M appears to come primarily from game contracts, rather than one opening or the other reaching superior partials.
(4) If we have a 5-3 major fit, we will normally take more tricks in the major than in NT. However, when both hands are balanced this number of additional tricks is, on average, less than one. Since the game level is one higher in the major than in notrump, 3NT is more likely to make than 4M. Correspondingly, 3NT tends to score substantially better (especially at IMPs) than 4M in the 5-3 when both hands are balanced.
(5) When the short hand has a singleton, 4M will often play two or more tricks better than 3NT and will tend to score better at all forms of scoring.

So when bidding or selecting methods, what conclusions can we draw from this:

(1) When in doubt, open notrump with a 5332 pattern when in range. Of course there is always room for judgement, but probably the notrump opening is "usually right."
(2) Puppet stayman is overrated. It may be better not to play this convention. If you are playing it, don't use it to look for 5-3 fits when responder's hand is balanced. Finding these fits will generally not improve the contract.

[catatonic]

interesting thread with a simple conclusion .......play 4 card majors !
clearly this doesn't preclude opening with a 5 card major if you have a 5-4 shape , and using a 2/1 which promises 10 plus and therefore a rebid 4 spades and 5 hearts is easily dealt with when partner bids a 5 card minor ahead of a 4 card spade suit in response to 1 heart [ 1h/2c/2h/2s ]

so what is supposed to be the advantage of playing 5 card majors anyway if it is , as appears to be the case , better to open 1nt with single suited hands

[1eyedjack]

catatonic, on Sep 19 2006, 05:34 AM, said:

interesting thread with a simple conclusion .......play 4 card majors !
clearly this doesn't preclude opening with a 5 card major if you have a 5-4 shape , and using a 2/1 which promises 10 plus and therefore a rebid 4 spades and 5 hearts is easily dealt with when partner bids a 5 card minor ahead of a 4 card spade suit in response to 1 heart [ 1h/2c/2h/2s ]

so what is supposed to be the advantage of playing 5 card majors anyway if it is , as appears to be the case , better to open 1nt with single suited hands

I don't know about you, but to me it seems a bizarre distortion of the evidence presented in this thread, that you should conclude from it that opening 1M with a 4 card suit is suggested by a body of evidence all of which points to opening 1N with as many as 5 of them. If anything, a 4 card major is more NT-orientated than a 5 card major.

[HeartA]

1eyedjack, on Sep 19 2006, 12:42 AM, said:

I don't know about you, but to me it seems a bizarre distortion of the evidence presented in this thread, that you should conclude from it that opening 1M with a 4 card suit is suggested by a body of evidence all of which points to opening 1N with as many as 5 of them. If anything, a 4 card major is more NT-orientated than a 5 card major.

ditto.

[catatonic]

ok , let me put it another way ; what is the advantage of opening majors only with a 5 card suit ? unless both hands are 4333 the 44 fit generally brings in an extra trick , wheras with 5-3 the ruff is most frequently in the long hand , gaining nothing

I think it is fair to say that the abiliity to find 44 major fits is an important factor in virtually all bidding systems ; being able to open a 4 card major makes this much easier , so why not do so?

[1eyedjack]

catatonic, on Sep 19 2006, 03:03 PM, said:

ok , let me put it another way ; what is the advantage of opening majors only with a 5 card suit ? unless both hands are 4333 the 44 fit generally brings in an extra trick , wheras with 5-3 the ruff is most frequently in the long hand , gaining nothing

I think it is fair to say that the abiliity to find 44 major fits is an important factor in virtually all bidding systems ; being able to open a 4 card major makes this much easier , so why not do so?

Ok I understand your point now, and have some sympathy with it

[1eyedjack]

A comparison that might be worth modelling is to see if the position of opener with respect to dealer affects the gain v loss of opening 1N v 1M.

I have on occasion opened 1M opposite a passed hand when I would otherwise normally open 1N, since I can then pass responder's new suit response, when I would be stuck for a rebid opposite a non-passed hand change of suit. But I am not sure whether that makes it winning strategy. This study could clarify that.