Wednesday, 28 March 2007

Polish Club: Minimum Balanced Hands

One of the main issues to be resolved in a Polish Club system is whether to open 1C or 1D on minimum balanced hands with four diamonds. In WJ05 these are opened 1D (though there is some freedom to choose: you might open 1C if the diamond suit was weak). I think that the decision is close but prefer to open 1C on all such hands.

Showing the strength of these hands is not a problem in either method. Both 1C and 1D promise a minimum of about 12 HCP, and this is a fine description of the strength of hands of up to 14 HCP. The real issue is showing shape. If you open 1D then you show 4+ diamonds, whereas if you open 1C you show a balanced hand.

I feel that showing a balanced hand is marginally more useful. As has been said before, the weak NT hand type is the dominant hand type in the Polish 1C opening, and so responder will play for that until proven otherwise. This makes it easy for responder to bid 5-card suits, particularly major suits. Since it is very important to show 5-card suits as soon as possible, and majors are more important than minors, I find that this gives the advantage to the 1C opening. 4-4 fits can often be found later with a take-out double.

Of course, this decision has an effect on the 1D opening as well. It will be opened on a 4-card suit only when the hand is unbalanced. This is a helpful inference in many auctions, in particular making it easier for responder to raise. The resulting opening bid is essentially the same as the Siege 1D opening, and I like to play the Siege 1D agreements as part of Polish Club. This works particularly well with my preference to open 1D on hands in the 18-21 HCP range, since Siege includes sequences to deal with that.

However, the Siege 1D opening also requires you to open 1C on balanced hands with five diamonds (since after a 1D opening bid, opener's 1NT rebid is artificial). Sometimes this is a good thing (when it enables partner to bid a 5-card suit in competition), but sometimes it is not. From the point of view of description and homogeneity, I would probably rather open 1D with the 5-3-3-2 pattern - again, it is usually best to show 5-card suits. In fact, it is sometimes possible to get away with opening 1D on these hands in Siege. With precisely 3=3=5=2 shape, opening 1D works because we can raise a major-suit response to the 2-level. With 3=2=5=3 shape, you might open 1D intending to rebid 2C after a 1H response, but obviously this is best only if the hand "looks" like a minor two-suiter rather than a balanced hand. With 2=3=5=3 shape you are forced to open 1C unless you fancy opening and rebidding in diamonds. So most of the time you will have to open 1C. But fortunately this does still give you the benefits of showing a balanced hand, and freeing up the 1NT rebid after a 1D opening is a great help.

Interestingly, the decision to open 1C on balanced hands with 4 diamonds also often helps when partner makes a negative double. For example, suppose you picked up a minimum 3=3=4=3 hand and were playing a system where this was opened 1D. Now if LHO overcalls 2C and partner doubles, you are a bit stuck for a rebid. Consider the same hand after a Polish 1C opening - if LHO still overcalls a natural 2C and partner doubles, you now have an easy 2D response. You actually benefit from not having shown a suit immediately. If this sounds paradoxical, the explanation is that partner now needs to have support for three unbid suits in order to double, rather than just two. So he can't actually make the double so freely. But then again, the 1C opening makes it easier for partner to bid a suit of his own, and he may be able to do this instead of doubling. Obviously you do lose sometimes - for example if responder has 4=2=4=3 shape after a 2C overcall and would have been able to raise diamonds - but we knew that when we decided to show the balanced hand rather than the diamond suit.

Tuesday, 27 March 2007

Polish Club

Polish Club is one of my preferred systems, and it demonstrates many of the things I've talked about in my series on bidding theory. It features a multi-way 1C opening bid fairly similar to the simpler example, Swedish Club, which was discussed there. But there is an additional natural hand type that is included in the bid, making it essentially a three-way bid. The definition of the Polish 1C opening is usually something like this:
  • 12-14 HCP, balanced (usually without 4 diamonds) or 4=4=1=4;
  • 15+ HCP with a real club suit;
  • 18+ HCP with any shape.

The other opening bids in the system are mostly natural. The 1D opening promises 4+ diamonds, but may have a longer club suit. 1H and 1S are natural 5-card major openings. 1NT is 15-17. And 2C shows a minimum opener with 6+ clubs, or 5 clubs and a 4-card major, "Precision-style". Higher opening bids are pre-emptive.

Some slight variations on this are possible. The above is what seems to be the standard, as described in "WJ05".

One variation which is a significant change is to discard the Precision 2C opening, and open all natural club hands with 1C, freeing up the 2C bid either for another pre-empt, or to show some very strong hands. This makes a huge difference to how the 1C opening works, and I would hesitate to call such a system "Polish Club", though the name does still get used. Furthermore, I think this version is inferior, and I won't be considering it here.

But WJ05 is not perfect either, and I would prefer to make some changes. Most of the changes would be to some of the more detailed sequences later on in the auction, but some other changes are to do with the opening bids. In the rest of this post I'll explain why I like the opening bids in Polish Club, and how I would try to make things even better than in WJ05.

First of all, the natural 1D, 1M and 2C bids are an excellent way of showing shape, particularly the 5-card major openings which are very homogeneous. With 1D promising 4+ cards, this is clearly better than in systems like Standard American where a 3-card suit is allowed with 4=4=3=2 shape.

A strong NT opening is, as we saw, an excellent way to ensure that you show strength on balanced hands. In Polish Club, the strength of any balanced opening hand of up to 17 HCP will be adequately shown by the opening bid. For stronger hands, the strength is not shown immediately, so these hands are treated as two-bid hands. This is sometimes a problem, though of course it is a problem shared by most standard systems. In competition at high levels, opener's double tends to show this sort of hand.

The advantages of the 1C opening itself are similar to those of the Swedish 1C opening. It is important to remember that, when compared to the 1C opening in standard systems, the Polish 1C opening does a much better job of describing minimum balanced hands. In competition, responder will initially assume the weak NT hand type, and this means can bid his suits much more freely than he would be able to opposite a standard 1C opening since he can expect some support. Essentially, the advantage here is that we are showing the shape of minimum balanced hands, as well as the strength. While strength is usually more important for balanced hands, being able to show the shape as well is a huge advantage over standard systems, particularly since these hands are so frequent. Of course, playing a weak 1NT opening would do this just as well, but by playing Polish Club you can describe the shape and strength of both the weak NT and the strong NT hand types.

The main problem with the Swedish 1C opening is the lack of flexibility: the "strong" hand types have to take a second bid (or else risk not showing their strength), which does not work well if those hands are not pure two-bid hands. Polish Club is slightly better in this respect since it is a more conservative system. In Swedish Club, the strong hands typically start at 17 HCP, whereas Polish Club requires about one point more than that. This may not sound like much of a difference, but it does tend to make the strong types a little more pure.

I like to modify this so that the strong hand types are even purer. As I've said, I don't believe that playing limited openings is an end in itself: rather, it is important to look at whether a natural opening bid would be a better start to the auction on some types of strong hands. One particular hand type to look at is where diamonds is the longest suit:

S 9
H AQ52

My feeling is that it is much more effective to open hands like this with 1D, compared to 1C. They are difficult to describe after a 1C opening because (amongst other things) a 2D rebid is usually defined to be artificial. Also, in competition, it will be difficult to describe the shape of this hand in one bid: you might be able to make a take-out double of spades, for example, but this wouldn't tell partner about the diamond length. And if the opponents bid some other suit (say they compete to 3C) it is a very difficult hand to get across. Opening 1D makes it much easier to describe the shape of the hand, usually either by reversing into hearts or making a take-out double of spades. If you change the hand to be a diamond single-suiter then it is less clear, but I still prefer to open these 1D in order to clarify the rebids after a 1C opening.

So in my version of Polish Club the 1D opening is not limited: it can be anything up to 21 HCP or thereabouts - only with a game-forcing hand would the opening bid be 1C.

When the longest suit is a major, things are different. The 1C opening is an excellent description of strong major single-suiters. So we definitely want to continue opening 1C with these. But with a more flexible hand, or a two-suiter, a natural opening might work better in competition. My preference, for hands with exactly five cards in the major, is to open 1H holding hearts but 1C holding spades. So 1H is not a limited opening (though it denies a single-suited hand with 18+ HCP), but 1S is. The reason for the difference is mainly that you are much less afraid of competition when holding spades. Holding the boss suit you can outbid the opponents on any level. Also, on a flexible hand you have the option of doubling and correcting partner's response to spades. This would often not be possible if you held hearts, since correcting a spade bid to hearts would mean raising the level of the auction. By opening 1C you get the advantages of showing strength. Note also that contructive bidding is slightly easier over 1H than over 1S (you have more space) so there is a little bit of extra room to fit in the additional hands which are opened 1H.

By moving many of these strong hands from 1C into 1D and 1H, particularly the "flexible" hands with 18-20 HCP or so, the remaining hand types in 1C are more likely to be pure two-bid hands, or at least are easier to describe with the second bid. This is what we wanted from our multi-way 1C opening.

Of course, getting in the way slightly are the "medium club" hands, which are the main difference between Polish Club and Swedish Club. The natural option of the 1C opening bid contains hands of 15-17 HCP which are almost certainly not pure two-bid hands, and so these can be a problem. These hands are also present in the 1C opening in standard systems, but there you have the option of rebidding in a new suit - this is not usually possible in Polish Club since such a bid would show the strong variant. Generally, if the bidding is at a very low level (certainly at the 1-level), you are better off than in standard systems since when you rebid clubs partner knows you must have at least 15 HCP. But if the bidding has reached the 3-level then life is much more difficult. Perhaps, though, the most important auctions to consider are when the bidding has reached the 2-level. This can happen because of an opponent's 2-level bid, but it will also frequently be due to partner bidding at the 2-level "to play opposite a weak NT". Unless opener has support for partner's suit, his usual action in this situation will be to rebid 3C. But this obviously requires a decent club suit. So you have a problem with a hand like this:

S AQ54
H 4
C KJ643

There's not really much you can do about it: this is precisely the hand you do not want to hold playing Polish Club. The biggest worry is that the auction might go 1C : (1S) : 2H and now you're hopelessly stuck. I don't offer any easy solution to this nightmare hand, but one thing I am certain of - you should aim to open 1NT as often as possible. Change the hand slightly to

S AQ54
H 42
C KJ643

and now I would say a 1NT opening is clear despite the weak doubleton. This works best when you are at the minimum end of the range for 1NT, since there is less chance then that your extra shape will cause you to miss a game. Hands with six clubs in a 6-3-2-2 pattern can also be opened 1NT if the club suit is weak, but single-suiters are less of a problem for a 1C opening since they are generally happy to rebid 3C if necessary.

There is one more big change that I like to make to the opening bids in Polish Club, which I will explain in the next post.

Sunday, 25 March 2007

DBT14: Conclusion

I've now finished all that I wanted to say on general bidding theory. It has only covered one particular aspect of bidding theory - the opening bid. And indeed it is even more specific than that, because I've been concentrating on what I would call "protecting against opponents' pre-emption", or, to give a shorter term, "description". There are other factors to consider when choosing your system of opening bids, which will often conflict with the desire to describe hands as well as possible. I'll just mention two of these things here, though of course they really are huge subjects in their own right:
  • Pre-emption. As well as trying to bid our own side's hands as effectively as possible, we also want to make life difficult for the opponents. Some opening bids will be explicitly defined as pre-emptive, usually at the 2-level or higher. These pre-emptive openings tend to be fairly independent of the constructive part of the system - except of course that the more pre-emptive bids you want to use, the fewer bids you have available for constructive hands. But also, the system's "constructive" bids can have a pre-emptive effect, and this aspect has to be considered at the same time as all the issues to do with describing hands. A weak 1NT opening, for example, is an extremely good bid from the pre-emptive point of view, but weak NT systems can have descriptive problems as we have seen.
  • Accuracy in uncontested sequences. As was said right at the beginning, competitive auctions tend to be more important, certainly for 1-level opening bids. But obviously there are many deals where opener's side will have a free run in the auction, and these need to be considered too. What happens in an uncontested auction will depend to a large extent on the continuations used, rather than the opening bid itself. But the definition of the opening bid can still make a huge difference to how things turn out. Most importantly, the accuracy of the continuations will depend on how much space there is: there must be enough room to look for the important information without going past the side's best contract. So again there is a conflict with trying to describe hands as well as possible - while a good description is still desirable in uncontested sequences, there is also the need to leave plenty of space for further investigation, which means that the cheapest bids will need to be used more.

Still, while you can never say that one idea is more important than anything else - it is so much more complex than that - description is the main thing that I look for in a bidding system. And it is all too easy to overlook description if you are not careful: while pre-emption is generally easy to spot, and leaving holes in uncontested sequences will also be easily picked up, protecting your side against opponents' bidding is a much more subtle problem.

Let's end by listing the main principles we've come across:

  • Think-Competitive. When deciding how to arrange your opening bids, it is most important to consider what will happen in a competitive auction.
  • One-Bid-Or-Two. Your system should ensure that one-bid hands are be described as accurately as possible in one bid, while two-bid hands are described as accurately as possible in two bids.
  • Unbalanced-Hands-Show-Shape. You must show your suits, in order to find fits.
  • Balanced-Hands-Show-Strength. The strength of a balanced hand should be bounded from below as accurately as possible.
  • Homogeneity. In any particular bid, the one-bid hands should be homogeneous.

How often have you looked at your hand and thought, "I haven't described my hand as well as I would like, but I don't have a good bid available." This problem is exactly what we are trying to avoid. It doesn't require lots of artificial bidding - indeed one of the main themes is that standard natural systems are very effective. But whether your preference is for natural systems or for lots of artificiality, by following these principles as far as possible your hands will be easier to bid.

DBT13: Limited Suit Openings

Playing a Strong Club (or Strong Diamond) system means that natural opening bids can be limited to a maximum of 15 HCP or so. Since these opening bids do show shape, their maximum strength is useful information. (Even the nebulous 1D opening found in some versions of Precision at least carries the negative inference that opener does not have (or is very unlikely to have) a 5-card major, and this is a useful type of shape information since it limits how good a major-suit fit the partnership can have.) Indeed, it is often said that having limited opening bids is the real purpose behind playing a Strong Club system.

However, I would not say that it is such a great advantage. For simplicity, let's consider just the major-suit openings. The hands which are removed from 1M by the Strong 1C opening are strong and mostly unbalanced, and so these hands are usually worth a second bid in competition (though they may not be completely pure two-bid hands). So whether these hands are opened 1M or not does not really affect the homogeneity of the 1M opening bid. Responder will initially be trying to cater for opener's minimum hands, and so excluding hands of 16+ HCP from the opening bid rarely makes a difference to responder's bidding on the first round. On later rounds, opener will have had the chance to show his strength more precisely.

Of course, knowing that opener is limited is not completely useless to responder. In particular, when responder has a minimum game-forcing hand, the limited opening may be sufficient to rule out the possibility of a slam. This means that responder can take a direct route to a game contract, rather than having to leave room in case opener had a very strong hand and wanted to make a slam try. The most well-known example of this is responder's direct raise to 4 of opener's major. In natural systems this bid would only be made on a weakish but distributional hand, based on excellent trump support. But in Precision it is also possible to raise directly to game with stronger hands: those which are worth game based on high-card strength, maybe 13 HCP or so with 3-card support. By reaching your best contract in two bids you avoid giving information away to the opponents, and if an opponent has a hand which might be worth competing over the game bid, it will be more difficult for him to judge correctly now that the jump does not promise a big fit. However, responder is aiming at quite a small target here. Even opposite a very limited opening, there are not very many hands that can be certain of wanting to play in game but also fairly confident that there is no slam available. Opener may be limited in high-card strength, but he is not limited in distribution, and there are very often some well-fitting distributional hands which would make slam good. And also, if the opponents do compete, it is now opener who is disadvantaged by not knowing responder's hand.

In any case, the immediate raise to 4 is rather a special auction. There are not many other situations where the limited opening bid is immediately useful to responder. For example, if opener's LHO overcalls 3C, then responder will have to bid 3NT (say) on almost exactly the same hands as he would if the opening bid was unlimited. Of course, when holding a good hand for the bid, responder will be happier opposite a limited opening because this reduces the chance that a slam will be missed. But the point is that when opener has a weaker hand the bidding will be the same after a limited opening as it would have been after an unlimited one. When this is the case, in order to compare different systems we only need to look at the stronger hands, and ask whether they are better opened with a natural unlimited opening or an artificial strong opening.

You see the real advantage of limited openings when opener gets to make his rebid. Playing unlimited suit openings, opener's rebids have to cover all the strong hands. With limited openings these strong hands are ruled out, and so opener's rebids do not cover such a large range and can be more descriptive. But note that it's not actually necessary to remove all the strong hands in order to do this. If the natural opening contains a restricted number of strong hand types, this still makes the rebids more descriptive than a standard unlimited opening: it doesn't matter that the bid would still be wide-ranging in terms of pure high-card strength. It might make sense, therefore, to use a strong (or multi-way) opening only for hands which are particularly easy to describe by starting with a strength-showing bid (such as strong major single-suiters), and use natural openings for other strong hands.

All in all, it is not the immediate definition of maximum strength which is the advantage of limited openings. Rather, limited openings are a way of re-arranging the meanings of opener's rebids, in the hope that hands will be better defined after the second call (and with easier continuations) than if you were using unlimited openings. But of course there are very many other ways you might try to do this, and it should be better to analyse the various hand types more closely, trying to find the best way to show shape on unbalanced hands and strength on balanced hands.

Pairs who play limited openings often open lighter than is standard. That is, while they have removed a large number of strong hands from the natural bids, this is compensated by opening more hands at the minimum end of the range. However, while there is a lot to be said for opening light, I feel that the argument that light openings go naturally with a limited opening system is largely fallacious. As was said earlier, removing the strong hands from a bid does not really affect its homogeneity, since the hands removed are mostly two-bid hands. But opening light does affect the homogeneity of the bid - the hands that are added are definitely one-bid hands. So it doesn't matter whether you decide to play limited openings or not, opening light makes your bids less homogeneous. Again, limited openings may help with the rebids, but not with the initial description of the hand. So by all means play light opening bids, but you don't need a limited opening system to do so. You would only see an effect if you were prepared to play a strength-showing opening which took out significant numbers of one-bid hands, and that would mean it would have to start at about 14 HCP - maybe even less. This might be necessary if your opening bids start really light (an 8 - 13 HCP opening bid is playable whereas 8 - 17 HCP is probably not), but including lots of unbalanced one-bid hands in a strength-showing opening is not such a good idea, as we have already seen.

Saturday, 24 March 2007

DBT12: Strength-Showing Openings

Nearly all opening bids give some information about strength, but here we want to look at bids which show only the strength of the hand, giving little or no information about shape. The most important example here is the Strong 1C opening in systems like Precision. Also, "natural" systems often use 2C as a strength-showing opening, but because of the low frequency of the bid and the fact that all hands opened 2C are two-bid hands, it is something of a special case and does not really define the system in the same way that the 1C opening is fundamental to Precision. We will be more interested in 1-level strength-showing openings.

The two main problems with strength-showing opening bids have already been discussed in earlier posts, but are worth repeating.

First of all, there is a danger of violating the Unbalanced-Hands-Show-Shape principle. On one-bid hands in particular, it may turn out to be impossible to say anything about the shape of the hand if you start with a strength-showing opening bid. But even on two-bid hands, while you will get the chance to say something about shape, it might not be possible to give such a complete description as you would if the opening bid had already begun to show shape information.

The second problem is that it is difficult to give a good definition of the strength of a hand in isolation, since its power depends a lot on how it fits with the other hands at the table. So, information about strength is most useful if the shape is also well defined. For this reason, if a system has a bid which limits strength very precisely, that bid will usually show shape precisely as well - as for example with a natural 1NT opening. A bid which showed a 3-point range of high-card strength without showing anything about shape would be very suspect - the upper limit on strength is almost useless, since hands can become much more powerful than expected if there is a big fit. Limited openings (ie. those where the maximum strength is very restricted) only really make sense if they show something about shape.

Information about the minimum strength is still useful, of course. As was said in "Balanced Hands Show Strength", if our hand has close to the minimum strength permitted for the opening bid that we choose to make, we can say we have "shown" the strength of the hand even if some much stronger hands are opened with the same bid. So one of the main advantages of strength-showing bids is that they satisfy the requirements of the Balanced-Hands-Show-Strength principle, for balanced hands near the minimum end of the range. For unbalanced hands the information about minimum strength is also undoubtedly helpful, but not as much as it would be if we had shown something about shape as well.

Looking specifically now at a "Strong" 1C or 1D opening bid, this typically promises a minimum of about 16 HCP, maybe slightly more or slightly less depending on how aggressive you want the system to be. Obviously, playing a Strong 1C or 1D opening has a huge effect on the other bids in the system as well, but for now we just want to look at the strength-showing bids themselves.

Such bids are unusual in terms of the proportions of one-bid and two-bid hands that they contain. Most 1-level bids are dominated by one-bid hands. But in a strong opening bid there are very few one-bid hands. Certainly, balanced hands of 16-17 HCP fall into this category. But that is about all. An unbalanced hand with this strength looks more like a two-bid hand: it may not be a pure two-bid hand, but the system really treats it as if it is, because having not shown shape with the opening bid, there is a lot of pressure to do so later. And once you start looking at even stronger hands than this, they might not like to take a second bid in competition, but they don't like to pass either. There is certainly a lack of pure one-bid hands. So these strength-showing openings have essentially the opposite problem to the very limited opening bids discussed in a previous post: while 8-12 HCP opening bids wasted space because they hardly include any two-bid hands, strong openings waste space because they hardly include any one-bid hands.

The small number of one-bid hands also gives responder an unusual problem. Suppose that he has to deal with a low-level overcall (somewhere between 1H and 2S, say). When he has a "positive" hand, good enough to force to game opposite opener's known strength, things are generally fairly easy. But more interesting is when he has a slightly weaker hand, a "semi-positive". These are not good enough to force to game immediately, so they have to be bid carefully, not going past the best part-score. Much of the time, it will be best to pass and wait for opener to describe his hand. However, since opener can occasionally have a one-bid hand, passing may result in the overcall being passed out. So responder is forced to act on a semi-positive hand if he wishes to compete for the part-score opposite a one-bid hand.

Now, if opener actually turns out to have extra values, any action from responder effectively commits the partnership to game. So the range of strength for the semi-positive hand types needs to be very narrow - not good enough to force to game immediately, but happy to play in game if opener has any extras. This does not seem to be very efficient: you are using an awful lot of system (the semi-positive responses and their continuations) to cater for a very small number of hands (opener's one-bid hands). This takes away space that could be used for more common hand types. Well-designed systems can use transfers or suchlike to combine the semi-positives and the positives into a single bid, but you still see the problem with semi-positives when responder makes a double, or where there isn't room for transfers, or when opener is prevented from showing his hand by having to cater for responder's possible minimum.

You can contrast this with the multi-way opening bids discussed previously. The equivalent of "semi-positive" hands for Swedish Club are those hands which want to compete opposite a weak NT hand, with the auction 1C : (2D) : 2S being a classic example. This has a much wider range. At the lower end, it only needs to be good enough for game opposite the strong option, so perhaps 6+ HCP. This is almost the same as opposite a strong opening bid. But the upper limit is determined by whether it is good enough to force to game immediately, which is much higher for the multi-way opening. So these semi-positive responses, which use up most of the available space, are much better used after a multi-way opening.

Some Strong Club systems go so far as to make responder's pass forcing over certain overcalls. In a sense this avoids the problem of having a small number of one-bid hands to deal with, by requiring opener to always take a second bid. But of course this takes away one of the main advantages of strength-showing openings, which is that they describe the strength of their minimum hands (particularly balanced hands) without opener having to take a dangerous second bid.

There are also a few systems which remove minimum balanced hands from the strength-showing opening completely, perhaps putting them into a strong NT opening instead. The idea is to increase the purity of the two-bid hands remaining in the strong opening, which works well when you do actually hold one of those hands. But again, you are losing the very thing that strength-showing openings are best at, which is describing the strength of balanced hands at the minimum end of the range, according to the Balanced-Hands-Show-Strength principle.

Friday, 23 March 2007

L&E Minutes for March 2007

The minutes for the meetings of the EBU's various committees are published on their website. I always take a particular interest in what the L&E are doing. Here are some thoughts on the latest set.

2.2.3. Use of the term "self-serving".

In the previous minutes, the comments of a player who was deemed to have fielded a psyche were called "self-serving" by a committee member. That was completely inappropriate. I don't like the term at the best of times, but its usual meaning is for a statement a player makes about their own methods or style or thought processes which would help their case if true but is not backed up by evidence, and so tends to be ignored by a TD or AC unless it is particularly credible. This instance was completely different, since it was a player's honest attempt to put his side of the case on the basis of agreed facts. It is totally out of order to dismiss this as "self-serving" - how can you have a problem with someone arguing their case like that?

The committee added a new comment to say that they didn't find the arguments convincing - fine, I agree with that - but the original comment still sickens me.

3.2. Announcements for "Acol Two"-like openings which fail to meet the definition of "strong".

This has been a known problem since before the new rules were introduced - I certainly asked about it, and I can't believe I was the only one. The problem is that neither "strong" nor "intermediate" really describes the agreement accurately - it falls between the two. I like the approach that the committee decided upon, announcing as "strong or intermediate", but there remains the question of whether the regulation is actually going to be changed to put this right. Or, more generally, what should happen to two-level openings which do not fall neatly into any of the announcement categories. It seems like they've ducked the issue a bit.

[Update 17/7/07: I should have had more faith: it has now been revealed that they actually went ahead and made a change to the OB, coming into force this summer, to address this point. So in fact this is another job well done in my opinion.]

3.5. Correspondence on announcements and alerts in general

Yes, the introduction of announcements has been a PR disaster. The committee really needs to realise that it's not just enough to make the right decision, you also have to communicate it properly to the players. I'm really disappointed in the way it's been handled. A belated consultation about it seems like a reasonable start in trying to put this right, trying to get back the support of the membership. The problem is that this consultation has to offer (or at least appear to offer) the possibility of discarding announcements completely, which would be a huge step backwards of course. All in all, very depressing that it has come to this, though we can hope that the results will be positive.

3.6. Minimum strength for "strong" openings

Brilliant work from the committee here, in the most important decision of the year so far. It's such a difficult subject - some definition of "strong" is essential, but it's so hard to express precisely where the line should be drawn. There was no doubt that the committee had previously got it wrong, making the requirements too restrictive and disallowing many hands which are normal strong openings for many players. The new definition is not exactly a thing of beauty, and people might complain that it is too complicated, but it seems to come pretty close to allowing the right things.

Perhaps even more encouraging than the decision itself was the fact that they've clearly listened to the complaints, sat down and thought about how to do better, and come up with something. You'd think that was an obvious thing to do, but it seems that very often known issues are just ignored because no-one wants to think about what to do about it, or to listen to people that have thought about it. But this time they have. (Eventually.) More of the same please.

The only slight disappointment was the decision to wait until August 1st to implement the new rule. Once you've admitted you're wrong, why let people continue to suffer? In my opinion, the right time for the change is as soon as it can be published in English Bridge (I suppose April would be unrealistic, but how about June?). We saw this with announcements - they were publicised nearly two months before the rules actually changed, and that was too much time for people to remember and keep up their initial "enthusiasm". Particularly for a relatively minor change like this, it should be immediate as soon as it is publicised. Note that the announcement of the change went up on the website immediately and people would be entitled to ask why they should have to wait another five months.

3.12. Moscito

Evidently the L&E has received some complaints that Moscito is allowed in the NICKO (since this competition is now level 4). Well, I agree with the complainants. As I've said before, I feel the standard level for tournament play should be somewhere between levels 3 and 4, and Moscito is just the sort of thing I would leave out. I've played it myself, and I know the amount of disruption it causes. The whole of level 4 would be appropriate for most team events, but the NICKO is intended to be accessible to non-experts and so I do not feel that this is appropriate.

Now, the L&E has decided that "it was a Tournament Committee matter". Very amusing. While this is strictly true, it is the L&E which has caused the problem by getting the "levels" wrong and by recommending that EBU events should be run at level 4. To be fair, there's not much they can do about it now, but it would be nice if they admitted responsibility.

Monday, 5 March 2007

DBT11: Breaking Homogeneity (part 2)

The main idea of the previous post was that in an opening bid, the one-bid hands should be homogeneous, whereas the two-bid hands need not be. But there is still an interaction between the one-bid hands and the two-bid hands, even if all the hand types are very pure.

Responder has to choose a call before it is known what hand type opener has. In making this choice, he will be mostly concerned with catering for opener's one-bid hand types. But inevitably, his choice of call will also affect how the auction goes when opener has a two-bid hand type. For a start, it may affect the amount of bidding space available. But also, and more importantly, responder's call provides information which hopefully will be of use when opener has a two-bid hand. In order to have the most effective methods, we would really like to arrange things so that the information responder provides in trying to cater for the one-bid hands also happens to be the information that would be most useful to opener when he has a two-bid hand.

The Swedish 1C opening is a very good example of how this can be achieved. Suppose that opener's left-hand opponent overcalls 2D. Now responder will initially play opener for the dominant weak balanced hand type, which means that the methods used will be very similar to if the opening bid had been a weak 1NT. He will therefore be bidding or doubling on the following types of hands:
  • Any hand strong enough for game opposite a weak balanced hand;
  • Any hand with a good long suit, wanting to play a part-score in that suit;
  • A hand suitable for a double. If double is take-out this means a hand with decent values, short in the opponents' suit.

He will not be bidding on mediocre hands without a long suit: these will be passed unless they are suitable for a take-out double.

This information is exactly what opener needs when he holds a two-bid hand. When opener has a strong hand like this, it is usually more important for opener to describe his hand, rather than have responder describe instead. So opener does not want responder to bid in front of him unless he has something particularly useful to say. The main thing that would be useful for opener to know about is if responder had a good suit of his own. And as we said above, this is indeed one of the hand types that responder will be bidding with. The characteristic auction for Swedish Club is something like 1C : (2D) : 2S. Here the 2S bid simply shows a hand which wants to play in 2S opposite a weak balanced hand, but the information that responder has at least five spades is also very useful when opener is strong.

This only works because the one-bid hands in the 1C opening are so well defined. In a standard system where 1C could be a weak unbalanced hand, responder would need a rather better spade suit in order to bid safely at the two-level. But in Swedish Club any half-decent five-card suit will do, such as KJxxx. If we had to pass with this sort of holding, then not only would we have difficulty in getting to a 2S part-score when that was right, but it would also be much harder to describe the hand accurately on the next round if opener turned out to have the strong option. Note also that if responder has a hand good enough to compete over an overcall, it is almost certain to be good enough for game if opener has a strong hand - this makes the continuations very easy. In summary, the weak option and the strong option of the Swedish 1C opening go particularly well together because in each case opener wants to know when responder has a moderate hand with a good suit. Of course, responder will also be bidding with strong hands of any shape, or with a hand suitable for a take-out double; again this is useful information if opener happens to have a two-bid hand. But really you see the biggest advantages of the two-way 1C opening when responder is able to show a suit.

We can contrast this with a similar-looking two-way 1C opening which shows either a minimum hand with 4+ clubs, or a strong hand of any shape. It would be very easy to build a system around this bid. And the one-bid hands here are nicely homogeneous, so responder does not have any immediate problems. However, it does not work so well, the reason being that the hands that responder will take action on have changed. In particular, he is going to be bidding on hands with moderate club support. Knowing that responder has four clubs (say) is really of very little use to opener when he has a strong hand - he is much more interested in 5-card suits, particularly major suits. And because opener's one-bid hand types no longer promise tolerance for the majors, responder will be much less willing to introduce a major suit into the auction. While we were happy to bid 1C : (2D) : 2S on a spade holding of KJxxx opposite a Swedish 1C, we would have to pass or double if the weak option just showed clubs (unless the hand was strong enough to force to the 3-level). So opener will get much less information about the majors: information is skewed towards the club suit.

The 1C opening showing "either clubs or a strong hand" is not particularly widely played. A more popular convention is a 2C opening bid which shows either a "weak two" in diamonds or a very strong hand. This has a very similar problem. If responder plays for the dominant weak hand type, then he will want to be raising diamonds on many hands with 3- or 4-card diamond support. However, if opener has a strong hand he is not interested in 3-card diamond suits at all. Indeed, a diamond "raise" may take away space that opener wanted to use to show his strong hand. Admittedly, on any given deal responder is likely to be able to guess whether opener is strong or weak. But there is much more uncertainty than you might imagine, and the price for getting it wrong is high. Furthermore, even if responder is able to work out what is going on, the partnership agreements still have to cater for both hand types, and this reduces the amount of space you have. Really, the two hand types are not particularly compatible, with information that is extremely skewed towards the diamond suit being of little use to a strong opener.

More generally, if the dominant hand type (the weak option) shows length in a particular suit, opener is going to get lots of information about responder's length in that suit, which is only going to be of use if opener's strong options have length in that suit as well. So again we see that simple natural bidding gets this right. For a standard natural 1H opening, responder is initially trying to cater for minimum hands with hearts, but if he supports the heart suit then this is still helpful to opener even when holding a much stronger hand. A more exotic example would be a 2H opening bid which shows either a "weak two" or a "strong two" in spades: here opener would be delighted to hear about spade support no matter which hand type he has (though these transfer pre-empts are problematic for other reasons, and clearly an immediate raise could make slam tries difficult with the strong type).

If you do want a wide variety of two-bid hands, then you are much better off making the weak option nebulous, or showing a balanced hand as in Swedish Club. While clearly the Swedish 1C opening has its flaws (the main one being a lack of purity in its two-bid hands, as discussed earlier), in terms of the interaction between one-bid and two-bid hands it's really as good as you can get.

Sunday, 4 March 2007

DBT10: Breaking Homogeneity

Some systems have bids which are explicitly defined as multi-way bids. Perhaps the archetypal example of a multi-way opening bid is the "Swedish" 1C opening, which shows either a weak balanced hand (11-13 HCP, say), or a strong hand of any shape (typically played as 17+ HCP). Natural systems do not use this sort of bid; however, many opening bids still contain a wide variety of possible hand types. For example, playing strong NT and 4-card majors, a 1H opening could be anything from a minimum balanced hand to a very strong 1- or 2-suiter. Clearly these are very different hand types, so the bid can be thought of as a multi-way bid in much the same way as the Swedish 1C is. The only difference is that the hand types in the natural bid are not separated by such a clear dividing line.

In either case, the important thing is to make sure that the various possible hand types go well together.

The One-Bid-Or-Two principle is very relevant here. If opener has a pure two-bid hand, then he will be able to reveal which hand type he has with his rebid. But if he has a one-bid hand, he may be unable to add anything to the initial description except to deny a hand suitable for taking a second bid. So, although responder's bids must cater for all the possible hand types opener can have, it is opener's one-bid hands which you need to be most concerned about. The opening bid must provide sufficiently good information about the one-bid hands that responder will know what to do. This does not necessarily mean that responder must be able to place the contract immediately, but he must at least be able to control the auction.

The easiest way to do this is to make the one-bid hands homogeneous. That is, while the opening bid as a whole may be a multi-way bid with a variety of different options, it is a good idea for all the one-bid hands to be covered by a single hand type - or perhaps by a small number of homogeneous hand types. The reason is simple: responder needs to know immediately what he should do opposite a one-bid hand, and a homogeneous opening bid is the best way of providing immediate information. All of the ideas about what makes a good homogeneous bid apply here.

For the two-bid hands, homogeneity is not so important: it is possible to have different rebids showing different hand types. The Swedish 1C opening is an excellent example of this. The one-bid hands are very homogeneous, consisting of a single hand type (11-13 HCP balanced), whereas there is much more variety in the two-bid hands: any shape is possible for the strong option. Many other well-designed multi-way bids work the same way: they have homogeneity in their one-bid hands, but it does not extend to their two-bid hands.

One problem with this sort of opening bid is a lack of flexibility. The different hand types are very clearly separated. So, if opener has a hand which is supposed to be treated as a two-bid hand, it is important to follow up by actually making the rebid which shows the hand. Unfortunately, as we know, not all hands are "pure" examples of one-bid or two-bid hands. A balanced 17-count is certainly not a pure two-bid hand. But if you open a two-way 1C, then this hand is treated as belonging to the strong variant. So if you are unlucky enough to have LHO overcall 2S, say, and responder raise to 3S, you will be in a difficult position. If you pass then partner will play you for a weak balanced hand, but there is no safety in making a free bid either. The lack of homogeneity between the different hand types is the problem here.

So this sort of bid works best when the "strong" hand types are genuine two-bid hands - as pure as possible. Ideally, they should either have overwhelming high-card strength (20+ HCP should ensure that the hand is a two-bid hand) or be very distributional - though in the latter case you need to be careful that you can actually describe the shape well with your rebid. (Single-suited hands are best for this.) In practice, most systems will have to allow some dubious hands into their strong options because of a lack of better places to put them, but clearly the strong option in our favourite example (17+ HCP, any shape) is already rather light, and anything weaker than this would be really asking for trouble.

Typically, if the one-bid hands are going to be homogeneous, this would mean that there is a particular hand type which responder will play opener to have until proven otherwise. We would call this the dominant hand type. Obviously the dominant hand type in the Swedish 1C opening is the weak balanced hand, but the term also makes sense for natural bids. For example, playing a standard 5-card major system, the dominant hand type in a 1S opening bid is a minimum hand with precisely 5 spades. More generally, the dominant hand type would usually be the most frequent one-bid hand type that opener can hold. We can state a general "homogeneity principle" in these terms:

All the one-bid hands in an opening bid should be as
close to the dominant hand type as possible.

The idea from a previous post that the "average" length (or strength) promised by an opening bid should be as close as possible to the minimum length (strength) is really just a special case of this. But we can now see that the average should be taken over only the one-bid hands. Two-bid hands do not need to be close to the dominant hand type, provided that they are pure two-bid hands. With an "in-between" hand you have a slight problem - it may work adequately well to treat them as two-bid hands, but it would also be helpful to have them be fairly similar to the homogeneous one-bid hands, in case the auction prevented opener from showing the precise hand type.

Bids which have a wide range of one-bid hands, or where there is no single dominant hand type, are much more problematic. But they do exist, with the most obvious example being the multi-2D opening bid. A multi-2D shows a weak hand with either hearts or spades (possibly with other strong options which are genuine two-bid hands). Although the strength promised is the same no matter which major suit is held, clearly in terms of distribution this is not a homogeneous bid. In a case like this, you have to find particular reasons why the bid "works" - why should responder be able to cater for both hand types at once? In general there is no guarantee that it will be possible.

There are a few reasons why the multi-2D is an effective bid. In particular:

  • It is a pre-emptive bid, and will hopefully cause even more problems for the opposition than it does for opener's side.
  • It is highly likely that responder will have at least as many cards in the other major as he does in the suit that opener actually holds. If opener has hearts and responder has 3-card heart support then the partnership should probably be playing in at least 3H; provided that responder has at least three spades as well, he will know that it is safe to bid 3H because his support for a spade suit would be just as good.
  • If this is one of the rare deals where responder is actually shorter in the other major, then the opponents will almost always have a fit there and probably ought to be playing in that suit. If they do bid that suit, then this will let responder know what is going on.
  • Because opener has a long major, if the hand belongs to opener's side, the most likely denomination is opener's suit. So responder will not often need to find out much more information than what opener's suit is. If opener's suit was a minor there would be much more information needed, and this would be difficult to get since you are already a step behind by not knowing what opener's suit is immediately.
But the important thing to note is that these points are very specific to the multi-2D bid. If you were to think of a different multi-way bid which did not have homogeneity in its one-bid hand types, you would have to try to find different ways to justify it. In many cases, the bid simply does not work. If you ignore the homogeneity principle you have to be very careful indeed.