Monday, September 16, 2013

Super-duper double disaster

Once in a while, because of my obsession with trying to tame the meaning of Double!, I get asked to adjudicate, or otherwise comment on, mishaps at the table involving doubles. The latest was a beauty. I am protecting the identities of those involved, but they managed to create a swing of 2500 points against themselves.



I believe that South made a significant (but recoverable) error in not bidding 2 (or 2) at his second turn. After all, he has a very nice offensively-oriented hand, especially when the opponents overcall clubs. And partner is not broke – we have about half of the deck, and when we have half the deck, we should not be shy about bidding shapely hands.

Most experts are reluctant to double for penalties bid-and-raised suit contracts below game. There's just too much risk for too little gain. Let's say, for instance, that both sides have an eight-card fit and that the law of total tricks is behaving well (indeed, on the basis of our hand with a void and a solid suit, there are quite likely to be more than sixteen tricks). For the moment, let's consider only the two options of doubling for penalties or bidding on (and they will double us if we're going down). If the opponents bid to the three-level (as here) and both sides can take eight tricks, doubling will give us +200 instead of -200  (9 imps) – that's nice. Or maybe we can be +500 instead of +140 (8 imps). Doubling looks lucrative. But just suppose that there are seventeen total tricks (the most common number). Doubling gains either 2 imps or, if disaster strikes, we lose 10 imps (for -670 versus -200) – an average of -4 imps. When there are more total tricks (as might well be the case when a minor suit is overcalled and raised like this), the risk of doubling far outweighs any gain because we might actually suffer a double game swing against us. This would happen here if there were nineteen total tricks: perhaps -670 for doubling and +620 for bidding on to game (a loss of 15 imps).



On this occasion, as you can readily perceive looking at all four hands, there are 20 total trumps and a massive 22 (!) total tricks. This is often the case with pure hands where there are singletons and, especially, voids. Here, the deal is pure because we have no secondary honors in clubs, nor they in spades. There is a small amount of crossover in the red suits but not very much.

Thus, North-South are cold for 6 yet allowed East/West to make 1070 in their club contract (one overtrick was always coming, the other, at trick one, was I suspect a result of North thinking that South had a defensively-oriented hand. That's 21 imps.

And the swing in a team game might be even greater, depending on the result at the other table. Suppose the other N/S played in a sensible but slightly pedestrian 4. Getting to 6 would therefore have won 13 imps. The actual result would have been -17 imps. That's an unfavorable swing of 30 imps on one hand!

So, North's double was intended as take-out, and I am almost in 100% agreement with it. I say almost because I think that the double here shows a hand with "extras", approximately a doubleton in the enemy suit and no clear bid.  But here, North could easily have bid 3.

It is true that North had bid 1NT and it's tempting to think that 1NT should switch on penalty doubles. But it shouldn't – it's not an offer to play notrump – it's forcing. As we've seen here, it can mask a fairly decent offensively-oriented hand.

2 comments:

  1. Why aren't you criticizing South's failure to double at his second turn?

    ReplyDelete
    Replies
    1. A good question. I definitely should have listed double as an alternative (I did criticize his pass) as probably the most descriptive action. At that level, double would be reasonably safe, despite having a void. But in my opinion (and it is only an opinion) double should show a hand more like 5431 shape, or even 5422. Having a void *might* have worked out badly if partner had a hand where a penalty pass was a candidate (reasonably probable given the 1NT response). Not here, obviously.

      So, to answer your question: because I forgot to mention that option.

      Delete