Monday, 22 January 2007

DBT7: Balanced Hands Show Strength (part 2)

Some systems are very good at showing strength on balanced hands. For example, many Italian pairs bid balanced hands something like this:

  • 12-14 HCP: open 1-of-a-suit (usually 1C)
  • 15-17 HCP: open 1NT
  • 18-19 HCP: open 2D (showing exactly this hand)
  • 20-21 HCP: open 2NT
  • 22+ HCP: open 2C

Here, the strength of a balanced hand is never more than 2HCP above the minimum promised by the opening bid, even for very strong hands. This is taking the Balanced-Hands-Show-Strength principle to extremes, and of course most systems will not make it such a high priority. But even if you do not build your entire system around showing strength on balanced hands, you should still be aware of the hands where it is most important, which are those with a strength of about 15 or 16 HCP. On weaker hands there is rarely a problem, because unless you play very light openings, those hands will not be significantly better than a minimum opening bid. And once you increase the strength to 17 or 18 HCP, you start to reach hands which are usually good enough to take a second bid (though they can still be a problem if the intervention is high enough, as with the example in the previous post). So it is the hands in the middle, about 15-16 HCP, which are most likely to cause a problem.

Most weak NT systems violate Balanced-Hands-Show-Strength on these hands. Here is the very first example from part 2:

D 97
C AQ53

When we play a strong NT we can open 1NT and this is a very pure one-bid hand. When we play a weak NT we have to open 1C or 1H, and now it is a nightmare hand. Supposing for the moment that LHO overcalls 2S, what is going to happen now?

Well, if 2S gets passed back to us, we will have to pass it out. This hand is nowhere near worth a second bid over a 2S overcall. So it is up to partner to do something if he wants to be in game opposite a 15-16 HCP balanced hand. The problem is, if partner does do something, and finds that you actually have a weak unbalanced hand after all, you will not always have a good fit at the three-level. And if he does bid at the three level, are you invited to bid game with a balanced 15-count, or is it merely competitive? Responder really needs to have ways to do both things, which puts a lot of strain on your system.

This assumes that partner passed at his turn. But even if he made a bid, or perhaps a negative double, there still may not be any safety in us taking a second bid. In these situations, weak NT players often use opener's double to show a balanced hand (and at least 15 HCP). This gets the strength across, but it is rather dangerous since there is no guarantee that it is right for us to compete. If we had observed the Balanced-Hands-Show-Strength principle, and described the strength of these balanced hands already, then the double could be reserved for hands which genuinely wanted to compete.

People that do play Acol or K-S, or similar systems, will accept these disadvantages and hope to gain instead on other types of hands. The weak 1NT opening itself has many advantages, being both very descriptive and also a good pre-emptive bid. It is particularly useful in first seat, and when the vulnerability is favourable, because of the pre-emptive effect. But my feeling is that in other positions it comes nowhere near making up for failing to show strength on the stronger balanced hands.

But it is possible to play a weak NT without violating the Balanced-Hands-Show-Strength principle. In Fantunes, Millennium Club and Nightmare, a 1C opening shows a better-than-minimum hand: 15+ HCP if balanced and not much less than that if unbalanced. And so opening 1C does show the strength of hands in the 15-16 HCP range very well. It turns these "strong NT" hands back into pure one-bid hands, which is what they naturally are. Also many Strong Club systems do a similar thing (though they may promise slightly more strength with the 1C bid, which puts pressure on hands with 15 HCP). If you want to play a weak NT without giving yourself huge problems on the "strong NT" type, you should play a system like this.

Sunday, 21 January 2007

DBT6: Balanced Hands Show Strength

This is the companion to Unbalanced-Hands-Show-Shape. Showing shape is always a good thing, because fit is all-important in competitive auctions. But, as explained in that post, there are various reasons why showing shape is more important with unbalanced hands. In particular,

  • When you hold an unbalanced hand, your side's fit is most likely to be in your own long suits, so you need to tell partner what your long suits are. Whereas when you hold a balanced hand, your side's fit will be in partner's long suit.
  • Holding an unbalanced hand increases the chances of the deal being a big fit, and so increases the chances of there being a high-level competitive auction.

At the same time, it is more important to show strength when you hold a balanced hand than an unbalanced one. Before going into the reasons for this, we need to look at exactly what it means to "show" the strength of a hand.

The problem with trying to evaluate the strength of a given hand is that everything depends so much on the context - there is a big difference between "offensive strength" and "defensive strength", and both of these are significantly affected by how well the hand fits with partner's. Even so, we have to be able to speak about the "strength" of an individual hand. Without wanting to go into detail about hand evaluation (personally I think that coming up with detailed formulae is neither worthwhile nor particularly interesting), everyone would at least agree that the strength of a hand depends on both the high cards and the distribution. But the value of high cards depends much less on the context than the value of distribution, and so when we talk about "showing strength" this will mostly based on high-card strength.

Now clearly, a 1NT opening bid with a narrow range, such as the Standard American 15-17 HCP, does show strength very precisely. The minimum strength and the maximum strength are both useful pieces of information for responder. However, a bid which has a much wider range, like a Standard American 1C opening (showing 12-21 HCP or so) is still considered to be a good description of strength for hands which are near the lower end of the range. The reason is that partner will play you for a hand of minimum strength unless you do something which shows otherwise. So the fact that you have at least 12 HCP is very useful information for partner - it means that he can bid a game if he holds a similar amount of strength himself - whereas the lack of a maximum is not a problem because he was not going to play you for much more than the minimum strength anyway. This works fine, provided that there is enough space to tell partner when you do in fact have a better-than-minimum hand. (In the case of a 1NT opening bid you obviously don't have enough space to do this, so it is necessary to have a narrow range.)

So, the statement of our new Balanced-Hands-Show-Strength principle is that when we have a balanced hand, we should try to make sure that partner's idea of our minimum strength is as close to the true value as possible.

Why is this so important? Well, as mentioned above, if we have substantially more than the minimum strength we have promised, then partner will not play us for a hand that strong. This would not be a problem if we could make a rebid which showed our extra strength. However, the important thing about balanced hands is:

Balanced hands are not two-bid hands.

As was said in the discussion of One-Bid-Or-Two, the stronger and more distributional a hand is, the more likely it is to be a two-bid hand. Since a balanced hand is, by definition, not at all distributional, it would have to be extremely strong in order to be a genuine two-bid hand.

C KQ92

This hand is a full 19 HCP, but if you open 1C and the opponents reach 3S before you have the chance to rebid, you are faced with a difficult problem. Doubling or bidding 3NT could very easily be wrong, showing that this is not a genuine two-bid hand. But neither is it safe to pass. The problem is precisely that we have not shown strength with the opening bid - partner could easily have enough for game, but he would have no way of knowing.

Fortunately, if we have a very strong balanced hand like this one, high-level interference is quite rare. But still, it's extremely difficult to cope with when it does happen, if we have not already shown strength. With an even stronger hand we would probably have to take a second bid regardless - there comes a point when any type of hand is strong enough to be a two-bid hand - but the vast majority of balanced hands are one-bid hands, and for those we really want to show strength immediately.

Saturday, 20 January 2007

DBT5: Different Ways of Showing Shape

As explained in the previous post, showing shape with the opening bid is very important, particularly for unbalanced hands. The question is, what is the best way to do this? The most common approaches are either to show length in a particular suit, or to show a balanced hand (i.e. length in every suit). "Natural" systems use only these two methods, whereas other systems may show shape in more unusual ways.

It's important to remember that the reason we show shape is in order to find a fit. (Indeed, we need to be able to find out both where the fit is and how good it is, though these things generally come together.) So, the best shape-showing bids are the ones which maximise partner's chances of knowing (or finding out) what our fit is. Some bids which appear to show shape are actually not particularly helpful for finding fits. The most extreme sort of example is if your system has a bid which shows a hand of 5-4-3-1 shape, but does not indicate which suit is which. If your only aim was to reveal the complete shape of the hand then this would be a fine way to start - later bids could reveal the order of the suits - but for finding fits quickly it's absolutely hopeless.

The standard method of bidding unbalanced hands is to start by bidding your longest suit. One obvious reason why this is sensible is that, a priori, this suit is the one most likely to be your side's best fit. However not all systems work this way. An alternative is "canapé" style, where we bid the second-longest suit if holding a two-suited hand. Pure canapé is rather uncommon, but many systems vary between opening the longest and the second-longest suit depending on the hand type.

The obvious problem with canapé is, how are you going to find out whether you have a fit in your longest suit? You can't rely on partner to bid the suit, because your side might have a fit there even if he does not have a very remarkable holding. Essentially, the only way you can discover a fit in your longest suit is if you are able to bid it yourself. And this may not always be possible. Unless you have a pure two-bid hand, you might not be able to take a second bid at all.

But this doesn't necessarily mean that canapé is a bad idea on one-bid hands. Take a hand like this, for example:

S Q4
H KQ52
D 82
C AJ984

Would you rather open this 1H, showing 4+ hearts, or 1C, showing 4+ clubs? There is a lot to be said for the 1H opening. If partner will expect four cards in whatever suit you bid, then the 1H opening is no less descriptive than 1C. And major-suit fits are more important than minor-suit fits. If 1C did not promise four cards (say you were playing a more artificial system) then it would be even more clear that 1H was a better initial description of the hand. The problems, if there are any, come when you try to refine the description of your hand with a rebid: you would like to be able to show both of your long suits. But since this is a pure one-bid hand, you only expect to get one chance, and so it is reasonable to bid on the assumption that you will only get to show one suit. In that case, showing four hearts must be at least as good as showing four clubs.

So, concealing your longest suit works fairly well when you have a pure one-bid hand, provided that the bid you choose to make is still a good description (if you were concealing a five-card major, it would be much worse). The real problems with canapé come when you have a hand which may be worth a second bid, and particularly if it is neither a pure one-bid hand nor a pure two-bid hand, but somewhere in between.

S 5
H AK93

You open this 1H, LHO bids 2S and it is passed back to you. Now you certainly don't want to pass, but what do you do? Well, this would depend on your methods, and there are many different possibilities, but nothing looks particularly attractive. The natural method would be to show the club suit directly, but this gives up on playing in diamonds or defending spades. Alternatively you could double, if this is defined as take-out, but now you haven't revealed your longest suit and so your continuations will be hampered by having to find this out. You also have the difficulty of trying to distinguish this type of hand from one with five hearts, if those are also opened 1H. Indeed, it seems that your initial shape-showing bid is not helping you at all: if you had opened with a strength-showing bid like a Precision 1C, then a take-out double of 2S would describe the shape of this hand equally well as anything you can do after opening 1H.

Similarly, if the opponents instead overcall in diamonds, there is still a fair chance that we have a fit in clubs, but it is dangerous to look for one because if a fit is not found then we have nowhere else to go.

This exposes a basic problem with not showing your longest suit the first time: if you get the chance to take a second bid, then you will not know what to do. On the one hand, you would like to show your longest suit, because this is by far the suit most likely to be a fit: but on the other hand, you can't commit to playing in your longest suit, because there is no guarantee that it will be right.

And so, if you see a pair playing canapé, or other methods involving not showing the longest suit with the opening bid, they will generally do so only with
  • pure one-bid hands, where the problems with taking a second bid are not important; and/or
  • pure two-bid hands, in cases where it is almost certain that a good description can be given with the second bid.

For example, playing strong NT and 4-card majors, it is reasonable to open 1H with a hand like my first example above. But we would never open 1H on a similar hand with an extra king, because then it might be worth two bids. (In fact it might well be opened 1NT in that case, in order to turn it back into a one-bid hand.) Similarly, canapé often goes together with a strong club, where the natural suit openings show fairly weak hands, which are unlikely to be worth more than one bid. Though here, hands near the top of the range for a limited opening can still be a problem.

The more traditional method of opening the longest suit really comes into its own on these difficult "in-between" hands. If the auction goes in such a way that it is too dangerous to take a second bid, then at least we have already shown the most important feature of our hand. Whereas if we do get another chance, we now have plenty of flexibility in deciding what is the next most important feature to show. Flexibility is the key: having shown the longest suit already, we could later decide to bid a second suit naturally, but we could also choose to double for take-out, or bid no-trumps, or repeat the first suit, or even make a conservative pass. If we had bid some other suit first, then these options might still be available, but they would not all tell partner about our long suit, so we would have failed to give such a good description of the hand in our two bids, as demanded by the One-Bid-Or-Two principle.

Of course there are other good reasons why you might want to open the longest suit. For example, it increases the average length shown by your opening bids. This is part of a much bigger topic on homogeneity that I will come to later.

Saturday, 13 January 2007

Defence to Artificial 1m Openings

Most regular partnerships will have a defence to a strong 1C opening - there's a space for that on the convention card. But what about openings which don't promise such a strong hand? There are many conventions of this sort which may or may not have the suit bid, for example:

  • A "short" 1C opening (playing 5-card majors and 4-card 1D)
  • A 1D opening showing either diamonds or clubs (in a Strong Club system)
  • A multi-way (e.g. Polish) 1C opening

These conventions can cause much confusion when they come up, if opponents have not properly discussed their defence. I have a defence which applies to all of them. (It doesn't apply to conventions which promise a suit other than the one bid, such as a 1D opening showing hearts.) The defence is called:

"Treat as Natural"

And it's very simple: the idea is our bids have exactly the same meanings as if their opening was natural.

I do mean exactly. So for example, if they open an artificial 1C,

  • (1C) : Dbl = take-out of clubs;
  • (1C) : 2C = Michaels (assuming you play this over a natural 1C);
  • (1C) : 1H : (1S) : 2C = cue-bid showing heart support.

Now, the main advantage of this defence is in being confident that we know what we are doing, not just with the overcalls but with all the continuations as well. But I believe it's also a good defence in the theoretical sense. It's important not to get carried away with crazy defences to these sorts of opening bids. The main weakness of nebulous opening bids is that they allow the defenders to get into the auction easily and reach their best spot. The opening side can't do much about this, because they are relatively unlikely to be able to make a pre-emptive bid before it is revealed what opener has. (Contrast this with a weak 1NT opening - made on a hand type often included in these artificial 1m bids - which often manages to keep the defenders out of the auction completely.) So we exploit this weakness by bidding constructively. Though having said that, it is also a good idea to make pre-emptive overcalls a little more freely than we would over a natural opening bid.

Of course, if you play the "Treat as Natural" defence you cannot show a hand with length in the suit bid by opener. Or at least, you can't show it immediately. If you hold this hand, you have to pass initially and hope to be able to show it later. This is, again, exactly what you would do over a natural opening bid, although the situation does not come up so often in that case. The general principle is that if a cue-bid on the first round would have been artificial, then a pass and later action shows length in that suit. So for example:

  • (1C) : p : (1H) : p , (1S) : 2C = natural clubs, decent hand;
  • (1C) : p : (1S) : p , (2S) : X = take-out of spades, implying some length in clubs.

However if partner has bid in the meantime then the situation may not be so clear. This sort of auction is particularly important:

(1C) : p : (p) : 1H , (p) : 2C

Is 2C natural here, or is it a good raise of hearts? Both meanings are useful. I have a trick for this situation, which is that when we pass over a minor opening and partner bids a major suit, bidding opener's minor is natural, but bidding the other minor shows a good raise of partner's suit - the reason being that if we had a hand worth a bid in the other minor, we could have overcalled on the previous round.

Of course, in the spirit of the "Treat as Natural" defence, all such tricks apply equally after a genuinely natural opening. Whatever you agree over natural openings also applies to the artificial ones, and vice versa.

All of this applies when you are on opener's right, as well. Not only that, if they play a 1D response to 1C which is a sort of negative or relay, we go ahead and treat that as natural as well. Though this does mean you need to know what you play in sequences like

(1C) : p : (1H) : ...

A sensible approach is to play 2C and 2H both as natural here, and that fits in well if 1C is artificial. On the other hand, when the response is 1D there is a lot to be said for playing 2D (or 2C, or both!) as Michaels. A double traditionally shows the two unbid suits, but I like to define it as "take-out of the suit doubled", i.e. take-out of hearts in the sequence above. This makes it clear that a response of 2C is natural, whereas 2H would be an artificial cue (even if LHO makes a bid). Indeed, all continuations are the same as if we had doubled a 1H opening.

Friday, 12 January 2007

DBT4: Unbalanced Hands Show Shape

Let's restrict our attention to constructive opening bids (as opposed to pre-empts) and consider how to define these in order to describe hands as well as possible. For the moment we'll ignore the main other desirable quality for opening bids, which is that we want them to be obstructive to the opponents.

Bidding systems generally define constructive opening bids in terms of just two things:
  • The shape of the hand (ie. the lengths of each of the suits); and
  • The overall strength of the hand.

Of course, by describing only these two things we are leaving out much of the detail of the hand, since we will not show anything about how the high cards are distributed amongst the suits. But because we have so few different opening bids to choose from, it is simply not possible describe that sort of detail.

Indeed, even when we restrict attention to shape and strength, there is only a very limited amount of information that we can convey with the opening bid. So we have to decide what is most important. And for unbalanced hands, it turns out that showing shape is most important. That is, while any opening bid will generally tell you something about strength (a standard opening bid might promise a minimum of about 12 points), your main concern should be to describe shape as effectively as possible.

The reason for this is that, according to the Think-Competitive principle, we have to anticipate what is going to happen in a competitive auction. And in competitive auctions, fit is all-important. This is most famously stated in the "Law of Total Tricks": if our side is bidding one suit and the opponents have bid another, then the decision as to whether to compete over an opponent's bid should depend almost entirely on how many cards we have in our suit and how many cards the opponents have in their suit. Strength hardly makes any difference to this decision (though of course it may be important if there are other decisions to be made, such as whether to bid a game). So if we are to get these competititive auctions right, it is vital that we know how good our fit is. In order to do that, we have to show shape.

When we hold an unbalanced hand, this increases the chance that we have a big fit - and also increases the chance that there will be a high-level competitive auction. Furthermore, if our side does have a big fit then it is likely to be in one of our long suits, and so it is essential that we tell partner which suits we have. This gives us our new principle: "Unbalanced-Hands-Show-Shape".

For one-bid hands, then, the situation is very simple: we have to try and describe the shape of the hand as well as possible with the opening bid. In this respect, opening bids which promise long suits are very good: a 5-card 1S opening is an excellent description of a hand which has five spades. The difficulty is in being able to give good descriptions for as many different types of hand as possible, and for one-bid hands this should be done on the assumption that we will only get one chance.

For two-bid hands things are more complicated. We must still try to show the shape of the hand, but now we are going to get two chances to do this. Perhaps this means that we do not need to worry about showing shape so much with the opening bid?

In fact this is unlikely to be true. It will usually take two shape-showing bids to describe the shape of an unbalanced hand well. For example if we hold a two-suiter, we need one bid to show the first suit and another to show the second suit. With a more flexible hand such as

D 7
C AK64

we can show our long suit (spades) with the first bid and then hopefully double for take-out the next time (if opponents bid our short suit). Particularly with these flexible hands, a double is often by far the safest action in a high-level competitive auction, and so if we had not shown our long suit with the first bid, we would be struggling to show it later.

Occasionally, though, we come across a hand like this:

S AKJ9532
H A2
D 4

Suppose we start by showing shape on this hand with a 1S bid, and the opponents bid up to the 3-level or 4-level. This isn't a disaster, since the spade suit is easily good enough to rebid. But rebidding spades doesn't quite express how good this hand is. Look what happens instead if we are playing a strong club or multi-way club system and open this hand 1C. Again, the excellent spade suit means that we have a comfortable rebid even if the bidding comes back to us at a rather high level. But, having rebid 4S or whatever, we now feel that the hand is described perfectly: we have described the shape and the strength. So this particular sort of unbalanced hand - a strong major 1-suiter - is actually very good for strength-showing openings. This can be viewed as being an exception to the Unbalanced-Hands-Show-Shape principle, but really what is going on is that it is a pure two-bid hand and you know you will be able to show shape very precisely with your second bid. Note that a 1-suited hand with a long minor is not quite so good, because it may not be advisable to bid your suit on the next round if this would result in missing 3NT.

Thursday, 11 January 2007

The Siege 1D Opening

Playing 5-card majors, we have to decide what to open on a balanced hand outside the range for 1NT. In the most widely-played systems, these hands are shared between the 1C and 1D openings. But an alternative approach is to put all these balanced hands into 1C. This is what we do in Siege (as described on Mike Bell's system page here). The result is that 1D promises an unbalanced hand (even 5-3-3-2 shapes with 5 diamonds are generally opened 1C).

We also open 1D on any hand with both minors, even if the club suit is longer. As well as being more descriptive (showing four cards with the first bid instead of two), opening 1D on these hands makes it easier to find a rebid. It also helps the 1C opening a little, since a 2D rebid is freed up for some alternative meaning.

In fact, Siege is far from the only system which uses this sort of 1D opening. There are other very similar short club systems, such as the one played by Welland/Fallenius which Siege was originally based on. And the same 1D opening can also be played as part of a strong club or multi-way club system. It is an essential part of Millennium Club, for example, and I like to use it in Polish Club as well.

The best thing about this 1D opening is that it promises a very suitable hand for playing in diamonds - much more so than a standard 1D opening which could often be weak and balanced. Our 1D opening can be raised almost as freely as a major-suit opening. In particular, a weakish hand with 4 diamonds and a 4-card major can normally afford to raise diamonds immediately rather than bidding the major, since if a major suit does exist it will not be as good as the diamond fit (except in the rare case that opener has a 4-4-4-1 hand). It is good to have several different ways of raising diamonds available, for example:

2D = weakish with 3 or 4 diamonds;
2NT = either a very weak raise, or a game-forcing raise;
3C = invitational raise;
3D = weakish with 4+ diamonds;
3M/4C = splinter.

The continuations after a 1D opening make use of the fact that opener will not have a balanced hand. After a 1H response (which is natural), we use the spare 1NT rebid to distinguish between the various types of minor two-suiters that opener might have:

1NT = both minors, with clubs at least as long as diamonds;
2C = both minors, with diamonds at least as long as clubs.

And so we can usually find our best fit when responder is giving preference.

Over a 1S response it is more useful to use the extra bid available to show hearts, solving the rebid problem that we have holding five diamonds and four hearts in standard methods. So we play:

1NT = both minors, either suit may be longer;
2C = hearts.

The advantage of using 1NT to show both minors, rather than the natural 2C, is so that when responder is (say) 2-3 in the minors with 8 or 9 high-card points he can rebid 2C, keeping the auction alive in case opener has a strong hand. In standard methods after 1D:1S,2C you might give false preference to 2D on this sort of hand, but this is not so attractive in Siege since opener could have only four diamonds.

We also have a 2NT rebid available, and can use this to help solve some difficult problems after a 1-of-a-major reponse:

2NT = Good hand with 6+ diamonds, denies 3-card support for partner's major unless intending to force to game;
3D = Good hand with 6+ diamonds and 3-card support for partner's major (but not forcing).

Opener's raise of a major-suit response to the two-level will frequently be made with only 3-card support. For this reason, when holding a minimum hand with 3=3=5=2 shape we can choose to open 1D, intending to raise a major-suit response. In principle this is the only time that 1D can be opened on a balanced hand in Siege.

Thursday, 4 January 2007

DBT3: One Bid or Two? (part 2)

In the previous post, our One-Bid-Or-Two principle was a guide for what to open with a particular hand. But we can also write down a similar One-Bid-Or-Two principle which applies to entire bidding systems. It is, again, very simple:

You should arrange your opening bids in such a way that:

  • one-bid hands are described as accurately as possible in one bid;
  • two-bid hands are described as accurately as possibly after two bids.

To put it another way, it is perfectly acceptable to have a relatively poorly defined opening bid in your system, intending to clear things up with the rebid. But this only works if the hands which need clarification are genuine two-bid hands. Otherwise there will be trouble.

I hope that the One-Bid-Or-Two principle sounds reasonable. However, it is certainly not the only thing we need to worry about. For instance, it may be more important to use the available bids to pre-empt the opponents. This is often in direct conflict with the desire to describe our hands as well as possible. Since we only have a limited number of bids available, there will have to be compromises made. But One-Bid-Or-Two should always be one of your main considerations, and I will spend a lot of time discussing it because it so often seems to be neglected.

So far I haven't said anything about how the hands are going to be described, only that they should be described as well as possible. In the next post I'll start discussing the various ways you might try to describe your hands.

DBT2: One Bid or Two?

Suppose you deal yourself the following hand:

D 97
C AQ53

Let's say you're playing a strong no-trump opening, and so you open 1NT. Now suppose LHO bids 2S. and this is passed back to you. What do you do?

Pass looks pretty clear, doesn't it?

Suppose instead that LHO had overcalled 3S, would that make any difference? Well, no, now it would be insane to do anything other than pass. What about 4S? Again pass is obvious.

Of course the reason that you pass so happily in each case is that the opening 1NT bid has already described your hand very well. If it was correct for your side to do something, then your partner should have done it already. So I call this hand a "one-bid" hand: you're not planning to take a second bid (or make a double) unless forced to do so.

Contrast this with the following hand:

S AQ9832
D 8
C 4

You deal and open 1S. LHO overcalls 2D and it comes back to you. Now maybe your choice of bid is not completely obvious, but what is clear is that you must show hearts in some way, describing your hand as a good two-suiter. Similarly if LHO had bid 3D, you would still rebid hearts at your turn. An if LHO had bid 4d? It still must be right to bid hearts. You can probably see where I'm going with this: this hand is a perfect example of a "two-bid" hand.

Perhaps this looks like a completely trivial observation, but I've found that the difference between one-bid and two-bid hands is fundamental to just about everything to do with the theory of opening bids.

Obviously the two hands above are very "pure" examples, in that the decision as to whether to take another bid was pretty much independent of the level of the opponents' intervention. More often you will be faced with a hand which is happy to take a second bid if the opponents interfere at a low level (particularly if they bid our shortest suit), but which is not good enough to take a second bid if the intervention is sufficiently high.

Note also that whether a hand is a one-bid or a two-bid hand may depend on the bidding system. For example, if you held the first hand and were playing a weak no-trump, you would have to open it 1C (or perhaps 1H), but now you would not feel that you had really described your hand until you had rebid in no-trumps. Even so, it is often helpful to describe hands as "one-bid" or "two-bid" hands independently of the system being used. My second example is probably worth two bids no matter what system is being played. Whereas a flat 12-count is only ever going to be a one-bid hand in competition. Clearly, the important factors are the shape and the strength of the hand. The stronger or more distributional a hand is, the more likely it is to be a two-bid hand.

How should all this affect our choice of bid? I think the conclusion is fairly obvious:
  • When you have a one-bid hand, you should aim to describe it as well as possible in one bid.
  • When you have a two-bid hand, you should aim to describe it as well as possible in two bids.

Of course, this may be too simplistic in many cases. For instance, if you have one of those hands which may or may not be worth a second bid depending on what interference you get, then you will want to describe it quite well with the first bid, while still leaving yourself easy ways to refine the description with a rebid if the auction allows. Still, the basic principle is very important. I will refer to it as the One-Bid-Or-Two principle.

Here is a very simple example: suppose you have to choose an opening bid with the following hand, playing a natural 5-card major system:

H J4
D -
C KQJ972

Your choice is between 1C and 1S. Since 1S promises five cards in that suit, whereas a 1C opening only promises three, it might seem like the 1S opening is a better description. But that ignores the fact that this hand is very definitely a two-bid hand, and so you should choose the opening bid which allows you to describe the hand as accurately as possible with two bids. The way to do this is to open 1C: you will virtually always be able to rebid comfortably in spades, and this will inform partner that the club suit is longer. (If the spade suit and diamond suit were switched then it would not be so clear, since while the hand would still be a two-bid hand, it might not be so easy to give a good description with those two bids.)

Finally, is there such a thing as a three-bid hand? Well, not really. That would only make sense if the opponents interfered over our opening bid and then interfered again over our rebid. Since opponents don't tend to do that very often, it's not something which is very important to consider. Indeed, if the opponents are going to interfere then the worst case is when they reach their highest level immediately. So the important thing is whether a hand is good enough to take the second bid. Hence the "One-Bid-Or-Two" distinction.

Wednesday, 3 January 2007

David's Bidding Theory: Introduction

Bidding theory is a huge subject because there are so many different auctions possible. If you're given the first few bids in the auction, you can always ask the question, what are the best meanings to give to the various possibilities for the next call? Depending on what the auction has been, it may be that natural continuations are best; or maybe you should make use of some artificial ideas such as transfers or relays or multi-way bids - or anything else you can think of. Serious partnerships could spend a huge amount of time deciding what they should do in all these different situations.

But there is one situation which is much more important than any other. That is, how will you define your opening bids? This is the aspect of bidding theory that I will be taking about in this series of posts. The opening bids are essentially what determine your "basic system", and I will be discussing what makes some systems better than others.

Let's start with one of the commonest mistakes made by amateur bidding theorists.

When you open the bidding, you can't be sure how the auction will turn out. But there are two main possibilities: either your side will have the auction to yourselves, or the opponents will come in and you will have a competitive auction. Which of these possibilities should you be more worried about?

Well, perhaps the correct answer is "it depends", but most of the time there is no doubt: competitive auctions are the important ones. The reason is simple: it's harder to reach the right contract in a competitive auction, because you have to worry about what the opponents are doing, and you usually have less space available. So, if your system works reasonably well in competitive auctions, it's can't be too bad when the opponents are silent. The converse is not true: if you come up with a set of responses which assume the opponents will be silent, it might all fall apart if they start bidding. Also, competitive auctions tend to be faster than non-competitive ones. If the opponents do not bid then a poorly defined opening bid can be corrected by a subsequent rebid, but if the auction is competitive then by the time you get to make your rebid it may already be too late.

The only time you might be more worried about constructive auctions is when the opening bid itself consumes a lot of space - particularly if it promises a good hand, as with a natural 2NT opening bid.

Admittedly, if you look at a book which describes a particular system, you will probably find that most of it is devoted to non-competitive auctions. That is the way it should be, since you can't play a system unless you know what to do in constructive auctions. Even so, it's the competitive auctions which will decide whether the system is a good one or not. I'll call this principle "Think-Competitive".

No doubt if you're reading this you will have known this principle already. But ignoring competitive auctions is a fairly easy mistake to make. I've often seen systems suggested where a 1C opening is made on a variety of possible hands, and the idea is to reveal which hand type is held with the rebid. Such systems often violate the Think-Competitive principle: if the opponents interfere at a high level, it may be too dangerous for opener to describe his hand. It is of course possible to create a workable multi-way 1C opening, but you have ensure that you can cater for all the various hand types in competition. More on this later.

Tuesday, 2 January 2007

Alerting of Doubles

Bridge regulations vary from country to country. Here in England our Laws and Ethics Committee is particularly keen on coming up with regulations, and as a result we seem to have far more regulations than anywhere else. Fortunately the vast majority of these regulations are very sensible. But inevitably the L&E occasionally makes a mistake. I'd like to talk about what I think is the worst mistake the L&E has made recently.

The rules for alerting of doubles changed in August 2006. The main change was for doubles of natural suit bids: we now must alert these doubles unless they are for take-out. (Before August, the expected meaning of a double was penalties if partner had already made a bid.) I feel that these new rules are a disaster. Admittedly they're not much worse than the old rules - and many players might actually find them better - but the problem is that the old rules were awful as well. And now that we've had to change from one to the other, it's become apparent just how awful both sets of rules are.

The problem is that the EBU rules are primarily designed for simple auctions. Suppose we take a simple sequence like 1S : (2D) : Dbl. You will almost certainly find that:
(i) Everyone knows what their agreements are in this sequence; and
(ii) Everyone knows whether their agreements are alertable.
And so here the alerting regulations work quite well. But in more complicated auctions, both of these things might no longer be true.

People hardly ever have explicit agreements about doubles in complicated auctions. They therefore have to work out the meaning of a double from general principles or from previous experience of similar situations. This causes problems in trying to alert correctly. For instance, what do you do if you're not completely sure what partner's double means? Note that this is much less of a problem for alerting of bids, since even if you're not sure what the bid means, it's usually fairly clear whether the bid is natural or not. Deciding between take-out and non-take-out doubles is much harder.

The other problem is that even if you know what the double means, it's easy to get the alerting wrong. For a start there are some counter-intuitive positions. For example, a double of a natural suit bid is not alertable if it's for take-out, but it's easy to miss the fact that this does not apply to an auction like (1NT) : p : (2C) : p , (2D) : p : (p) : Dbl (because although you are making a take-out double of diamonds, the 2D bid was actually artificial). Also many people forget to alert doubles which are "obviously" penalties, such as (1NT) : Dbl : (2D) : Dbl or 1H : (1S) : 2H : (2S) , 3D : (3S) : Dbl. And the EBU rules make a fine distinction between "take-out" (usually not alertable) and "competitive" (alertable), which many people do not understand.

Admittedly, getting the alerting wrong does not always lead to problems with misinformation or unauthorised information. But it would still be much better if the rules were easier to get right.

Some other countries have the rule that no doubles are to be alerted. This is obviously much better from the point of view of understanding what the rule is. But it means that opponents are not warned about unexpected agreements.

So is there a better way? I think there is.

The point is that the unexpected agreements for which alerts are useful nearly all come in very simple auctions - and these are the simple auctions in which the current EBU rules work well. Whereas, in more complicated auctions there are hardly any agreements which are sufficiently unexpected that an alert is really necessary. So the plan is to use something like the current EBU rules for simple auctions, and change to the "no alert" rule for more complicated auctions. The difficulty is defining what "simple" means in a way that people would understand.

Fortunately, the simple agreements that we're interested in are really simple. In fact they can be limited to just four or five specific auctions. So what we can do is replace the four general rules in the current Orange Book with a similar number of rules each dealing with a specific sequence. Note that it's much easier to learn a rule about a specific sequence than a general rule covering lots of sequences. For example, while most people don't understand all the implications of the current rule about alerting doubles of natural suit bids, they do generally remember that negative doubles are no longer alertable.

The situations where I would have alerting are:

(i) Doubles of natural suit opening bids below 3NT. [Rule: Alert if the double is not for take-out.]

(ii) Doubles of natural 1NT or 2NT opening bids. [Rule: Alert if the double is not for penalties.]

(iii) After an opening bid and a natural suit overcall from opener's LHO below 3NT, a double by opener's partner. [Rule: Announce as "negative" if the double is for take-out; do not alert or announce if the double is for penalties; alert any other meaning.]

(iv) After an opening bid at the 1-level and a pass from opener's LHO, the double of a natural suit response (or raise) below 3NT. [Rule: Alert if the double is not for take-out.]

Note that in (i), (ii) and (iv), nearly everyone plays the non-alertable meaning for the bid, so most players would not even need to know these rules. The important rule to know would be number (iii).

[I would also have to make an exception for "anti-lead-directing" doubles (which are currently alertable even above 3NT) because these are particularly unexpected. But this is probably the only exception which is necessary.]

This was the specific suggestion that I sent in to the EBU. But what is more important than the specific suggestion is the general idea that in most auctions, only very unusual doubles should be alertable. Any regulation which did this would be an improvement on the EBU's ideas. The ACBL regulations do in fact work this way: in some sense my suggestion is simply a more precise version of the ACBL's rules.

I'm still hoping the L&E will concede that they got it wrong.