All systems working

Is there a perfect bidding system for bridge?  Chthonic (that fictional, bridge-playing automaton) believes there is.  It has something to do with using the bids to divide the space of possible hands according to the Fibonacci numbers.  But of course, it's totally artificial.  Besides, he hasn't actually published it yet.

All other practical systems have trade-offs.  Precision is a carefully designed system and is used by some of the top pairs in the world, including Meckstroth and Rodwell.  The first call (assuming the opponents haven't already bid) divides your hand, more or less, into three tranches:  0-10 (pass), 11-15 (1♦ thru 2♦), 16+ (1♣).  I say more or less because there are still other bids like weak twos, etc. but these are the constructive opening bids.  If one partner opens 1♣, responder immediately attempts to divide the partnership wealth into two tranches: part-score, game.  There's a lot to be said for knowing in two bids if we have game, definitely don't have game, or might have game.  But even Precision has weaknesses, it allows the opponents to preempt before we've designated a suit and then there's the nebulous 1♦ call which has to cover a wide range of hands, some of which are decidedly short in diamonds.

But, for the purpose of this discussion, I'm more interested in natural systems where the constructive opening bids are all at the one-level.  In the early days of bridge most people played that you needed four cards to bid a suit and, if you had a balanced 16-18 count, you opened 1NT.  Occasionally, it might be necessary to make a "prepared" bid, that's to say a bid in a low-ranking suit, preparing to show a balanced hand at your next turn, when your point range fell outside the 16-18 range.  However, this was fairly unusual, because you almost always had a biddable four-card suit.  The notrump range was reasonable, because even opposite a Yarborough, you would still have some chance of making or at least going down no more than one.  Besides, you don't get so many balanced 16-18 hands so the bid didn't come up all that often.

Then, it was realized that this was all very well for Rubber bridge where you were constantly partnering someone different and therefore didn't really have time to work out a "system" anyway.  But for "serious" tournament bridge, where bidding a game is so important (all bidding systems are tuned primarily for playing IMPs because that is considered to be "real" bridge, as opposed to matchpoints), it was felt that some adjustments were needed.  In particular, suppose partner opens 1♠ and you hold ♠A53 ♥QJT9 ♦7 ♣A9863.  You'd really love to know if partner has five spades because then game is reasonable opposite even a minimum hand.  This means that you can afford to make some bid other than 2♠ which would be a shut-out opposite a normal 12-15 point hand.  However, if partner in fact has only four spades, your attempt to get to game might easily get you too high, resulting in a negative score.

So, the idea of five-card majors was born.  It was accepted that minor suits could be bid with just four, or even three, cards because we don't tend to look for marginal games based chiefly on minor suits.  We look primarily for 3 notrump contracts or, secondarily, 5-of-a-minor based on something like 26 hcp (with some distributional help in the case of the minor-suit game). 

So, now comes the big question.  In the event that we don't have a 5-card major, but instead some other balanced hand, what should our NT range be.  The early theorists felt that there was a considerable advantage to the weak no trump (12-14).  This was because, on those occasions where you had to start with a minor suit (thus leaving the door open to major suit intervention) and which might not even be based on four cards, it was better to have something to spare point-wise (i.e. 15+).  Those balanced 12-14 hands would be opened with a somewhat preemptive 1NT.  As long as partner had distribution, and/or at least 6 high card points, nothing bad would be likely to happen.  Much thought was put into running from the dreaded double of 1NT.

Some people were not convinced by this argument and decided to switch things around.  The new system was called "Standard American", although to begin with it certainly wasn't standard, even in America.  Balanced 12-14 hands would henceforth be opened with a minor suit, followed by a 1NT rebid and balanced 15-17 hands would be opened with 1NT.  Not quite so much comfort as in 16-18 but we got to use the new systems (transfers, etc.) more often.

Because balanced 12-14 hands are so common, std-am players open a lot of hands with 1 of a minor.  It makes it very easy for the opponents to intervene now because responder has very little idea of what's going on.  Is the minor suit opened really "real"?  After partner opens 1♣ and RHO overcalls 1♠, a hand like ♠853 ♥QJT9 ♦72 ♣A986 wants to raise to 2♣ but may fear that partner opened a three-card suit with four good spades and a total of 12 hcp.  He's apt to make a negative double, especially because partner might have a good hand with four hearts, but which might very easily get the partnership too high when opener has the wrong hand.

Now, let's address the issue of missing a 4-4 major fit and playing in notrump instead.  First of all, provided both contracts make, there really isn't much difference when playing IMPs (1 imp probably).  Admittedly, there may be a big difference between 90 and 110 at matchpoints.  However, whatever system we are playing, 1 notrump will usually only be played when we have two opposing balanced hands without game aspirations.  Sometimes, given these conditions, playing in a suit will be better than playing 1NT but this is not always the case: one at least of the balanced hands has to have some ruffing potential; trump quality has to be good enough to withstand an attack on the trump suit, etc.

Let's assume that we have two hands with a combined point count of 22 and an eight-card spade fit.  Opener has a balanced hand with 4423 shape.  Responder's hand is 4243.  Let's first assume that opener has 14 hcp.  The weak-no-trumpers' auction will go 1NT all pass.  The strong-no-trumpers will go 1♣ – 1♠ – 2♠ (we're assuming a Walsh style here, otherwise the spade fit will likely be lost).  Thus the weak-no-trumpers may lose an imp at teams and perhaps earn only a 25% board at matchpoints.

Now, let's look what will happen if a Jack is passed under the table from responder to opener (a priori 15hcp is slightly less likely than 14 hcp so this scenario will happen slightly less frequently).   Now, the tables are turned.  The weak-no-trumpers will find the spade fit and win an imp or the 75% board.  In the long run, these differences are not likely to be really significant, especially at teams.

The one situation where we may legitimately miss our spade fit is when opener is 4333 and responder is 4423.  An auction which starts 1♣ (when the hand doesn't fall into our NT range) may continue 1♥ – 1NT all pass.  But, this is usually not a tragedy.

Now, why is it wrong to rebid a suit when you've opened a "prepared" club (or diamond).  Again, we're not too concerned about getting to the perfect part-score for the reasons outlined above.  Any making part-score is considered equivalent.  But what if our aspirations are game, or slam?  Let's suppose that we pick up this hand ♠A653 ♥KJT9 ♦7 ♣A983 and hear partner open 1♣.  We dutifully bid 1♥ and now partner rebids 1NT.  We know within a point or two what partner's strength is and we know that he is balanced.  If he's showing 12-14, we will want to try for game.  If partner has 15-17 we definitely want to be in at least game, but aren't sure of the strain yet.  It's possible that partner has only three clubs but in that case, his distribution is 4333.  If we had a minimum responding hand, say ♠8653 ♥KJT9 ♦7 ♣A983, we wouldn't really care and we would pass (they might take a few diamond tricks but we only have to make 7).  But with the stronger hand, we have plenty of room to find out exactly what partner has.  If he has four spades, we want to be in 4♠ (slam is possible but not likely).

What if partner rebids 1♠ over our 1♥?  Now, we know that he has at least a relatively unbalanced hand.  It still might have as few as 12 hcp but conceivably it could have up to about 21 hcp.  Most probably he has a singleton somewhere (probably in hearts) [although some hands with a small doubleton diamond might also  rebid 1♠].  So partner most probably has something like 4135 shape.  He certainly has at least four clubs.  It seems that we have a double fit in the blacks.  If our rounded suits were switched so that our hand was ♠A653 ♥A983 ♦7 ♣KJT9, we would certainly want to investigate slam.  How disappointed we're going to be if partner puts down a 4333 hand like ♠KJ42 ♥Q72 ♦AQ4 ♣Q86!  That's why it's so very important to limit your hand (by bidding notrump) as soon as possible when it's balanced.

Finally, what about Walsh?  It's designed to prevent us from finding our 4-4 major fits, right?  No.  Quite the reverse.  First of all, Walsh only applies when the auction begins 1♣ – 1♦, not 1♣ – 1♥.  Walsh bidders suppress a four or five card diamond suit when they hold a four-card major but are only strong enough for one unforced bid.  This maximizes the chance that the partnership will find a 4-4 major-suit fit.  If responder has diamonds and a major with invitational strength or better, then he bids naturally and if opener rebids 1NT, possibly hiding a four-card major, then we can still find it by using a checkback bid such as NMF or XYZ.