Everyone Needs a Forcing Raise

This is not really a bidding theory post, it's more of a practical thing.

In my regular partnerships I play very complex systems. But I also play with lots of different people from the club, and then we play simple versions of Acol. Now, I'm not a big fan of Acol, but if it's just for the odd evening I don't really feel I miss anything from my more complex systems ... except for one thing.

Traditional Acol has no bid which shows a forcing raise of a suit opening. If partner opens a major and you have a game-forcing hand with 4-card support, you are supposed to start by bidding a new suit, and then support partner at game level on the next round (a "delayed game raise"). With a very strong hand you might have to start with a jump-shift. (These days splinters have become very popular - widely understood even at the club - but they don't help you when you don't have shortage.)

The problem is that this just doesn't seem to work. Either you or your partner has to guess whether to go past game, and my experience is that it is very difficult to guess well. In order to have a sensible auction, you really need to tell partner that you have a game-force with support before you reach game level. So, you need a forcing raise.

For me this really became very evident in the last few weeks. I saw six hands suitable for a forcing major-suit raise - three in a beginners' class, two in the club duplicate and one in a league match. Of these, there were two missed small slams, one grand played in game, one poor slam going off and one hand played at the five-level with three losers. Only one time was the hand bid to the right contract. But in each case where it went wrong you couldn't say anyone had made an obvious mistake. And in each case the hand would have been trivial to bid with a forcing raise (apart perhaps from the grand slam, which might only reach six).

So it's clear to me now: if you have a new partnership - even for just one evening, and no matter who your partner is - you have to agree a forcing raise. Forget about defences to 1NT or other such trivial matters. You can do without those. You can't do without a forcing raise.

This applies even to beginners. Generally you would like to teach a beginner basic natural methods, leaving any unusual conventions to people who have reached a more advanced level. But for raises it's totally the opposite way round. A beginner is hopelessly lost without a bid which shows this hand, whereas it takes expert judgement to play delayed game raises. This is a rare example of how adding a convention actually makes the game much easier to play. Club players are often criticised for using Blackwood too early in the auction; but in many cases this is because they have no reasonable alternative. If they haven't been taught a forcing raise, what else would you expect?

The good news is that the EBU's teaching methods now appear to be recommending 3NT over a major-suit opening as a "pudding raise" (showing a raise to game without a shortage to splinter in). But this hasn't yet permeated through to the ordinary bridge player in the way that splinters have. A pudding raise is excellent for beginners or for a one-off partnership, though it doesn't solve all problems and so a lower forcing raise would be preferable (but requires more complicated responses). When I played with David H at Cambridge we thought we couldn't afford to give up our natural inviatational 2NT response (I might have a different opinion now!) and so we used 3C as a raise instead. We started winning IMPs every time it came up. I've never seen a convention which made such an immediate improvement to a system as this one did.

This was all assuming a major-suit opening. And indeed it is more important to have an artificial raise for the majors than for the minors. But this is only because of frequency: with, say, 4-4 in a major and a minor, if partner opens the major you need to raise immediately, whereas if he opens the minor it's more normal to bid the major-suit first. Thus a minor-suit raise is only really needed when we have a single-suited hand - but when this does come up it is no less important than it was for the majors. In a simple system I might like to use a jump in the other minor as a game-forcing raise. Admittedly, when it comes to the minors, if you're playing in a one-off partnership you might just hope it doesn't come up. That wouldn't work for the majors though.

I've heard some traditionalists say that you can get by without an artificial raise. Perhaps you can, just about, get by. But as I said, it takes expert judgement to play delayed game raises with any sort of effectiveness. And if you then go and look at the systems the experts are actually playing ... they are unanimous that an artificial raise is a good idea.

So my conclusion is not so much that it's nice to play a forcing raise - because I'm sure you knew that already - but that if you have a beginner who is learning Acol, or a new partner who plays old-fashioned methods, this is one thing you really must add to their system.

Stayman v. Keri

Keri is Ron Klinger's system of responses to a 1NT opening, as described in his book "Bid Better, Much Better After Opening 1NT". I played this in my partnership with Mark. But since then I've gone back to ordinary Stayman, and in this post I'll try to explain why.

In Keri a 2C response forces opener to bid 2D. There are then three basic sequences which take care of the hands which would have bid Stayman in standard methods:

• 1NT : 2C , 2D : 2M shows an invitational hand with 4 or 5 cards in the major.
• 1NT : 2D , 2H : 2S shows an invitational hand with both majors. (The 2D response in Keri is a transfer to hearts, but differs from a standard transfer in that in can have only 4 hearts if it is invitational with both majors.)
• 1NT : 2C , 2D : 2NT shows a game-forcing hand.

The first of these sequences is the one we need to focus on, since it is the only really fundamental part of the system. (There are variants of Keri which deal with game-forcing hands differently to the "book" version above.)

The idea is that after 1NT : 2C , 2D : 2M opener will pass with a minimum hand with 3 or 4 cards in the major. Thus a big advantage of Keri is being able to stop in 2M - particularly when this is a 4-4 fit (where a Stayman auction would have reached 3M after 1NT : 2C , 2M : 3M) or a 5-3 fit. There are plenty of examples of this in the book. The disadvantages, on the other hand, are not so clearly spelt out!

The main problem with the 1NT : 2C , 2D : 2M sequence is the ambiguity in responder's major-suit length. It could be either 4 or 5 cards - and having to cater for both is a little uncomfortable. In particular, it makes a difference to the the type of hand on which opener wants to accept the invitation. When responder has only 4 cards in the major he is mainly interested in whether opener is minimum or maximum in high cards. But when responder has a 5-card suit, fit tends to be more important. In an attempt to cater for both types, Keri uses a 3C bid by opener to show a minimum with good fit - but of course when you do this you are losing the chance to play in 2M, which was supposed to be the advantage of playing Keri. And in any case this still doesn't solve the problem that there can be a big difference between 2- and 3-card support when responder has 5-cards, but there isn't when responder has only 4.

Another issue is that you frequently play in a 4-3 fit. Opener passes the 2M bid whenever he is minimum with 3-card support. As Klinger's book points out, this often works quite well. But it does rather depend on the hand. Playing a 4-3 fit looks great when responder has a good suit and opener has a weak side-suit doubleton. It looks rather less clever when responder's suit is bad and/or opener is flat. The difference is highlighted at matchpoints where you are going very much against the field (and if 2M makes the same number of tricks as no-trumps you will score badly). Unfortunately the system forces you to play 2M every time. So you end up with some good results and some bad results. Klinger's book seems to be trying to persuade us that it is a winner on average; this wasn't my impression from playing it.

There is no doubt that when responder has a five-card suit and the invitation is rejected, you are pleased to be able to play in 2M. But here Stayman can be even better than Keri, at least when the suit is spades. Playing Stayman we can use 1NT : 2C , 2red : 2S to show an invitational hand with 5 spades. When this comes up we get all the benefits of Keri but without the ambiguity about spade length. As mentioned above, this helps opener in knowing when to accept the invitation. But, even better, it means that opener can pass with a doubleton. A 5-2 fit does tend to play better than no-trumps, particularly if responder is unbalanced. In Keri you have to bid 2NT over 2S with a doubleton spade, in case responder has only four. But when responder has a 5-card suit you would generally prefer to play 2S in the 5-2 fit rather than 2NT.

This is not available when the suit is hearts. If we play Stayman we have to start with a transfer on an invitational hand with 5 hearts. My preference is that opener should be keen to super-accept with a good heart fit, so when responder has a borderline invitation he should transfer and then pass 2H. This can also work better than Keri on those hands (since we play 2H rather than 2NT opposite a doubleton). On the other hand, a sound, fairly balanced invite with hearts is certainly better bid in Keri.

One nice thing about Keri is the continuations after a transfer: because 2NT is not needed as invitational, it can be used to improve the game-forcing sequences instead. However, this is not really an advantage of Keri since it can also be easily incorporated into a Stayman-based system. Playing Stayman-then-2S as an invite frees up 1NT : 2H , 2S : 2NT just like in Keri; and if we are playing Stayman we don't need to use 1NT : 2D , 2H : 2S for 4-4 majors like Keri does, so there is plenty of room here as well.

There are more problems if you play the "book" version of Keri. One that particularly bothers me is that while Stayman is notorious for giving away information to the opponents unnecessarily (when opener shows or denies a major that responder is not interested in), Keri is, if anything, even worse in this respect. After 1NT : 2C , 2D : 2NT opener reveals whether or not he has a doubleton, even though this information may not be needed by responder. A similar thing happens after 1NT : 2C , 2D : 2M if opener has enough to accept the invite. Against opponents who are good enough to make use of the information this is a very bad idea.

Finally, you do need to think carefully about how you will cope with interference, and the book is a little short on detail here. I remember once, after bidding 2C and hearing a 2S overcall, not being sure whether 3D would now be weak or invitational. (It may not even be possible to distinguish.) In the Junior Camrose this year a player responded 2D to 1NT on an invitational hand with 5 spades and 4 hearts, and then heard a 2S overcall! This should have gone for a four-digit penalty, but instead they had a misunderstanding and ended up in a silly contract. A transfer which only guarantees 4 cards does tend to cause problems in competition even if you are well-prepared. More generally, you'd better be sure you know how much of the artificiality applies in a competitive auction. This significantly increases the amount of work involved.

My conclusion is that while Keri is a decent alternative to Stayman, it certainly isn't "Much Better" as the title of the book claims. Perhaps the best thing about the system is that if you and your partner both know the book, you can agree "Keri" and immediately have a complex, rather effective system ready to go, without needing to develop it for yourself. But, in my opinion, if you had a book which was based on Stayman that could be even more effective.

L&E Minutes for Feb 2008

So, this post is the second instalment of my not-so-regular feature discussing what's going on with our L&E. The latest meeting is particularly interesting for me because the main thing they discussed was the issue I've ranted about more than anything else on this blog. (You know what I mean.) The minutes are on the EBU's website here.

There were some other things of interest. In particular, there is the very important job of explaining to the EBU membership what the changes are in the new Lawbook. The job has been given to DWS and Mike Amos. This is promising. I wish them luck!

It also looks like there may be some new stuff allowed at Level 4.

But let's cut to the chase.

It's fantastic that they are now discussing the problems with alerting of doubles. There seems to be a genuine enthusiasm to do something about it. Even though they failed to make significant progress at the meeting, it seems that the issue is still very much on the agenda.

The concern now is that, while they have spent plenty of time discussing what to do, they haven't yet come up with a viable, specific proposal. And there really isn't much time left if they are trying to do this by August 1st (given that it would have to be properly advertised).

But what was really disappointing - and the reason I actually felt the need to post about this - was what happened to the idea of alerting only "highly unusual" doubles. This idea is probably not optimal, but it would be workable if done properly, and it's surely miles better than what we have at the moment. And yet it was sabotaged.

David Burn had produded a paper on this idea. And the paper was pretty good. I agreed with more or less all of the points he made, and in a couple of places I felt he really hit the proverbial nail on the head. The problem was, the paper was manifestly not about alerting only highly unusual doubles. And I don't mean just in the wording - he changed from defining "unusual" to defining "expected", which ought to end up with much the same thing - I mean that the effect of his suggestions didn't have anything to do with alerting only unusual doubles.

He gives as an example:

1D : (Pass) : 1H : (1S) , Double.

which he defines as "expected" to be penalties - thus a take-out double would be alertable. The problem is, a take-out double is not unexpected. If a pair plays this double as take-out they will not believe they are doing anything "highly unusual" - because they aren't. The whole point of the idea of alerting only highly unusual doubles is that it will only affect the (minority of) players who play weird methods. Once you start defining perfectly normal methods to be alertable this is no longer the case. Mr Burn writes, "The idea is that players will alert the doubles their opponents need to know about without actually having to read the list at all, let alone memorise it." I couldn't agree more with that philosophy. But you can't achieve this if you arbitrarily define some perfectly normal agreements as being alertable.

It seems that Mr Burn fell into the trap of wanting just one agreement to be "expected" in any given situation. But in examples like the one above, a double might be take-out or penalty or value-showing: none of these things would be unexpected. So if you were actually going to base your rules on the idea of only alerting very unusual doubles, none of those things would be alertable. Indeed I would say that trying to have just one expected agreement is the mistake which causes all the problems in the current rules. This is precisely what we need to get rid of!

My point is that alerting only highly unusual doubles is a workable approach, but it isn't what the L&E was presented with. So they may have dismissed the idea of alerting highly unusual doubles on the basis of a paper which had nothing to do with alerting only highly unusual doubles. That would be a pity.

I'm certainly not saying that this would have been the best approach. There are two alternative approaches that I think are better: the first is the one on my webpage, and the second was actually mentioned at the meeting, attributed to Sally Bugden:

Mrs Bugden suggested that a possible way forward would be to not alert take out or penalty doubles ...

This really would be an excellent basis for a regulation. The problem is that it is not a "finished product", so my fear is that it could go the same way as the idea of alerting only highly unusual doubles. The main issue to be sorted out is that sometimes a take-out or penalty double can be extremely unusual. This doesn't invalidate the general rule; it just means you need to consider whether it is worth having a very small number of specific exceptions (like we have for alerting above 3NT at the moment). For example, a double of a natural suit opening bid is expected to be for take-out: in this situation a penalty double might catch people out if it was not alerted. A complete proposal would have this as an exception. It would also need to sort out what happened for doubles of artificial bids, and other techincal points. My worry is that without a complete proposal, people would see issues like this and dismiss the idea.

Systemic Opening Pass (part 2)

Following on from the previous post, let's assume we are playing a system which uses a natural 2C opener, and have agreed to pass hands of minimum opening strength with 1=4=3=5 or 4=1=3=5 shape.

The reason we might pass 1=4=3=5 hands was explained in the previous post. Hands with 4=1=3=5 shape are very similar: in particular, the singleton heart gives us a way to get back into the auction later if opponents bid hearts. The 4=1=3=5 shape is not quite so good for passing as 1=4=3=5, but the problems with finding an opening bid are the same. (In a standard "natural" system, on the other hand, it would be ridiculous to pass 4=1=3=5 hands with opening strength: there is no problem with either the opening bid or the rebid.)

Now, if a pass can be made on up to 13 HCP, we need to make some changes to the methods we use as a passed hand. In particular, there is a problem if partner opens 1-of-a-suit in third or fourth seat. Here we have the values for game unless partner has made a tactical light opening, but using traditional methods it is difficult to see how we can bid the hand. A natural 2C would normally be played as non-forcing by a passed hand, and in some circumstances a 2C response might even be played as artificial (Drury).

Since the "strong" hand types have such specific shapes, I think the solution is to use a specific response to show these hands. For example, over a 1S opening we could use a jump to 2NT to show the 1=4=3=5 type with 11-13 HCP. Normally most jumps by a passed hand are used to show hands with support for partner, but these do not come up particularly often, and with both opponents passing there is no particular need to have lots of bids which show support. So it makes a lot of sense to use some of the jumps for a new purpose.

If it is the opponents who open the bidding, there is not so much need for specifically-designed methods. If they bid one of our long suits there is no need to come into the bidding at all. If they bid our short suit, on the other hand, it is quite safe to come into the bidding even at relatively high levels, since a take-out double describes the hand very well. A take-out double does not necessarily show the "opening pass" type of hand - it could just be a normal maximum pass with ideal shape - but partner will be aware of the possibility.

The systemic pass on minimum 1=4=3=5 and 4=1=3=5 shapes makes a big difference to the 2C opening bid. The traditional Precision 2C opening shows either 6 clubs or 5 clubs and a 4-card major. Playing a systemic opening pass we can guarantee that if the hand does have only 5 clubs, it will be at the maximum end of the range. This is because, for weaker hands:

• With 1=4=3=5 or 4=1=3=5 we can pass;
• With 2=4=2=5 or 4=2=2=5 we can pretend the hand is balanced;
• With a hand short in diamonds we have an alternative opening bid available: a three-suited opening bid in traditional Precision, or a 1C opening in Polish Club.

When responding to a 2C opening this information is very helpful: it means we can investigate game with an invitational hand knowing that if opener does not have the values for game, he will always have a six-card club suit for us to fall back on. Whereas in standard Precision we might play an uncomfortable 2NT contract after 2C : 2D , 2M : 2NT, if minimum 5-4 hands are passed this sequence should probably be played as forcing, which in turn makes life easier for responder when he holds a strong hand.

Playing this agreement means that a 2C opening bid will only very rarely have precisely 5 clubs. Ideally we would like 2C to absolutely guarantee a 6-card suit. Some versions of Precision do this by putting the other hands into 1D, but a systemic pass is a very good alternative for the minimum hands. A more extreme version of the systemic opening pass would pass all hands in range for 2C if the shape was right, but now you get to the point where partner will want to allow for the strong type when deciding whether to open the bidding; once this happens you are not really playing a standard system any more. Of course, there is not necessarily anything wrong with unusual systems, and it would be amusing to see how a genuine two-way pass system would fare, for example:

Pass = 0-10 HCP any shape, or 11-16 HCP unbalanced with 4+ clubs and a 4-card major
1C = 11-13 HCP balanced / 44(41), or any hand with 17+ HCP
1D = 5+ diamonds, 11-16 HCP
1H = 5+ hearts, 11-16 HCP
1S = 5+ spades, 11-16 HCP
1NT = 14-16 HCP balanced / 44(41)
2C = 5+ clubs, no 4-card major, 11-16 HCP

Partner would be expected to open in 3rd or 4th seat with any 9+ HCP. Perhaps this sort of thing should not be taken too seriously, but it might be worryingly playable. (Except that it's almost certainly illegal, of course.)

Systemic Opening Pass

Suppose we hold this hand as dealer, playing Polish Club:

S 8
H KQ73
D AQ7
C JT653

What do we do? The normal opening bid with this shape is 2C, but the club suit doesn't really look good enough for that. Alternatively we could consider opening 1C, but then partner will expect a balanced hand, not a small singleton in spades.

I believe that by far the best option is to pass this hand.

Now, certainly, this hand is better in terms of playing strength than a typical balanced 12-count which is an automatic opening bid. But when deciding whether to open the bidding it isn't all about strength. In particular it is perfectly appropriate to consider how well our system deals with the hand. Hands with this 1=4=3=5 shape are particularly bad for systems with a natural 2C opening, so there is no need to open them on minimum values. With only 12 HCPs, passing is not an unbearable risk.

As well as being a bad hand type for opening the bidding, there are also some reasons to think that passing will work quite well. This is essentially because of the singleton spade. The singleton improves the prospects for a pass in two ways:

1. We are not so worried about the hand being passed out - firstly because the other players at the table are more likely to open if they have length is spades, and secondly because there is a good chance that if the deal is passed out, it actually "belonged" to the opponents in a spade contract.
2. If the opponents do bid spades, this hand has a good way back into the auction via a take-out double.

So if you were ever going to consider passing a hand with opening strength, a 1=4=3=5 shape is the best candidate. Indeed the same is probably true in a natural system, though there the problem is not so much in finding an opening bid, as finding a rebid after 1C : 1S.

Having observed that it can work out well to pass, we could just use this as an occasional alternative on borderline hands. But it becomes much more interesting if it is an agreed part of the system. The systemic opening pass works like this:

• We identify some specific shapes where passing seems to work well with minimum opening strength. (Let's say just 1=4=3=5 and 4=1=3=5.)
• We agree that pass is allowed with these shapes on a little more strength than normal - say up to 13 HCP.
• Now when a passed hand shows this sort of shape, partner will be aware that it can be relatively strong.
• We can also define some specific bids which show these hand types as a passed hand.

Although partner will be aware of the possibility of a stronger hand, it would be a mistake to let this affect things like our style of opening the bidding in third seat. The relevant hand types are very rare. The whole point is that we would expect passing to work well even without partner doing anything differently - at least until the passed hand gets a chance to clarify what he has. So on the first round at least, partner should just bid as normal.

Once the partnership's methods are designed to cater for these hands being passed, the pass becomes even more attractive. So we could pass hands with these shapes even when the club suit is not so bad, perhaps even with a hand like this:

S AJ96
H 5
D 753
C AQJ63

Note that these hands are still relatively bad for opening a natural 2C, because of the difficulty in investigating whether we belong in spades.

In the next post I'll go through some more implications of this agreement.

Interference over Natural Suit Openings

There is an old debate about whether a new suit by responder (e.g. 1S : (2C) : 2H) should be forcing or not. Either method is very useful when the appropriate hand type comes up - and causes difficult problems when you pick up a hand of the opposite type. If forced to choose one or the other throughout, I would prefer to play a new suit as forcing (which is standard in most parts of the world). Though non-forcing free bids do look very attractive when the suit is a major that can be shown at the 2-level.

But there is no need to choose. By using artificial tricks such as transfers we can usually show both hand types at an appropriate level. Unfortunately, different situations require different gadgets. Below are the methods I'm currently using for 2-level interference over 1D/1H/1S openings, which demonstrate all the various ideas that might be used.

1. Over a 2C overcall.

Here we use switch: that is, of the two unbid suits, a bid in one of them shows length in the other one. So:

Over 1D : (2C), 2H shows spades and 2S shows hearts.
Over 1H : (2C), 2D shows spades and 2S shows diamonds.
Over 1S : (2C), 2D shows hearts and 2H shows diamonds.

This essentially gives us both a forcing and a non-forcing bid in the higher-ranking suit, while not raising the level on the lower-ranking suit.

Note that the "switch" relies on both unbid suits being at the same level. So if you are fond of this method you could use it over spade overcalls as well, and also over 1S : (2H) and suchlike. But we actually do different things there.

2. Over 1M : (2D)

Here we free up some space by giving 2NT an artificial meaning. This allows for the following structure:

A bid of 2 of the other major is natural and not forcing.
2NT shows clubs.
3C shows the other major (stronger than the NFB).

3. Over 1D : (2M)

Again it makes sense to use 2NT as a transfer to clubs here. Unfortunately this time there is a real shortage of space, since while we would like to use 3C to show the other major, it would also be nice to have a way to show a good diamond raise below 3D. We can't realistically do both.

I currently play 3C as the diamond raise, so over 1D : (2S) we have to play 3H the traditional way, natural and forcing. Over 1D : (2H) we have 2S as a NFB and 3H showing a stronger hand with spades.

4. Over 1M : (2OM)

This is where transfers really come into their own. Again starting from 2NT we can play

2NT shows clubs
3C shows diamonds
3D is a raise. (When we have two artificial raises available below 3M, as with 1S : (2H) : 3D/3H, we use them to distinguish between 3- and 4-card support.)

Exactly the same transfers work if opponents bid 2M Michaels over our 1M opening.

The Monkey Option

Congratulations to the 94 EBU members who managed to vote for

"2-level suit openings should all be alerted"

in the EBU's online survey.

I would like to think that they were making some ironic statement about the deficiencies in the EBU's polling methods.