Join Bridge Winners
All comments by Kit Woolsey
You are ignoring the author of this comment. Click to temporarily show the comment.
It is clear that there must be some kind of time control on bridge events, as otherwise a player could think for an hour about a single play which is obviously unacceptable. The question which needs to be answered is the following:

Do we want the time control to be an intrinsic part of the game which may affect the result of a match, or is the purpose of the time control simply to eliminate prolonged and unnecessary delays.

In chess, the time control is definitely part of the game. If your flag falls, you lose. Your strategy can be affected by the clock. It is quite common for a player to adopt a complicated although unsound variation if his opponents is short on time in order to force the opponent to make a mistake due to the time pressure.

If the time control is part of bridge, then it is to your advantage to cause an opponent to use time. The strategy of not claiming but forcing an opponent to sweat out discards becomes part of the game. There are other situations where one can take advantage of the time control.

Suppose you are on defense and your opponent leads up to a KJ in dummy when it is known that he has a doubleton in his hand. With no time control, if the AQ are both onside the defender will take his ace and get the hand over with. If the AQ are both offside, the defender will just show declarer his hand and again get the hand over with. But if there is an advantage to having an opponent use time on the clock, LHO will play small and the opponents will say nothing when both honors are in the same hand, letting declarer waste time working it out when the defenders know the result of the hand doesn't depend on declarer's play.

Suppose pair A has been taking most of the time during a quarter, and the table is on the verge of exceeding the time limit. If there is time control and penalties, it is to the advantage of pair B to slow down the play for the last couple of boards and put the table over the limit. The way things are done now there is never any penalty if the table finishes in time, but if the table exceeds the time limit then pair A will be penalized and pair B will gain.

Here is an situation I faced several years ago in a trials match where chess clocks were in use. I was dummy. My screenmate had just won the trick. After a minute or so, it became clear to me that he wasn't thinking about what to play – he didn't realize he had won the trick. If there were no time control, obviously I would remind him that it was his play and the game would go on. But there was a time control. It would be to my advantage to say nothing. Regardless of what you might think is the ethical thing to do, should this sort of ploy be part of the game?

It is to be noted that in all these examples the time control which allows a pair to gain when their opponents are slow actually has the effect of prolonging the match, since there is an advantage to force your opponents to use time unnecessarily.

So, I ask again: Do we want the clock to actually be a part of the strategy of bridge matches? We need to answer this question before we can sensibly discuss how to implement time controls.
Nov. 6, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

Your statement that slam is less likely when the opponents intervene might have been true many years ago, but in this day and age of aggressive opponents who know to interfere with 1C auctions on any excuse I wouldn't agree that slam is less likely.

We have found that getting a lid on responder's strength is most important. Establishing a game force as quickly as possible is obviously valuable, and similarly limiting responder's hand so opener can put on the brakes is also important. Structures such as you propose don't seem to work out as well, particularly if fourth hand raises, since the combined strength isn't defined.
Nov. 4, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
The sample size from this article is very small. 281 deals is not very many when trying to reach a significant conclusion. To make matters worse, it is likely that on over half of these deals the contract was either completely cold or had completely no play, and on these deals there would never be a distinction between making and not making. For example, there might be a hand which makes 9 tricks at notrump double-dummy but at the table declarer makes 10 tricks due to a favorable lead. If the contract were 4NT this would be meaningful, but with the contract being 3NT it comes out equal. Thus, probably only a little over 100 of the deals even had the potential for a different result as far as making vs. going down is concerned.

I ran a similar but more meaningful test a few years ago. I took all the deals I played at a fall nationals, and compared them to the double-dummy results for the same strain from the same side of the table. There were over 400 deals. Also, since the fall nationals involves only matchpoint or board-a-match events for the major championships I played in, every trick counted. Thus, every deal could be considered important.

My overall results were that on average declarer beat out the double-dummy result by about .2 tricks per hand. I admit that my sample size isn't large enough to be truly significant, but I believe these results are a lot more meaningful than the results presented in the article.

The analysis which Rainer mentions is much more meaningful. The sample size of 30 million deals is certainly large enough. The overall results are as follows:

actual double dummy difference

level 1 7.43 7.06 0.37

level 2 8.12 7.97 0.15

level 3 9.01 8.87 0.14

level 4 9.84 9.85 -0.01

level 5 10.12 10.07 0.06

small slam 11.42 11.51 -0.09

grand slam 12.05 12.14 -0.09

It is to be noticed that on balance declarer beats the double-dummy par on low-level contracts, while declarer falls below par on the higher level contracts. This makes sense. On the higher level contracts declarer has considerably more decisions to make than the defenders, thus more likelihood of doing worse than the double-dummy result. On the low-level contracts the defenders have a lot more decisions to make.

The flaw in the okbridge analysis is the quality of play. Clearly the average okbridge player is well below expert level. At first glance it might appear that the mistakes will balance out since these mistakes will be made by both declarer and defender. However, declarer has more decisions to make, particularly at the higher level contracts, hence more opportunity to make mistakes.

I would predict that at a higher level of play the okbridge results would not be the same, and that noticeably more contracts would make that shouldn't than vice versa. Torrey's methodology is fine – only his sample size is too small. If one were to use Torrey's methodology over perhaps 5000 deals from IMP expert competition (which would certainly be available on the Bridge Base vugraph archives), the results of such a study would either support or nullify my prediction.
Oct. 11, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

I play the double of 2NT shows a maximum with defensive orientation, a hand which would be happy to hear partner double 3. Axxxx x KJx AKxx for example. South is pretty much known to have short hearts when he says anything, otherwise he wouldn't be bidding. It is important to be able to punish the opponents when they try to push you around at the wrong time.


I don't remember the enemy carding, but I don't think it makes much difference. The opponents both know that declarer is short in hearts, probably void, and looking at the dummy it is pretty obvious that a passive defense is called for. Thus, there isn't much that can be read into anything they signal.

5-5 in the West hand is possible. However, most players are reluctant to overcall at the 2-level vulnerable on a 5-card suit. It certainly isn't clear, but I believe I should have gotten it right.
Oct. 9, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
It is exactly because the South hand is limited that North's 2 call is fine. He doesn't have to worry about South bidding too much on a 17 or 18-count, since South doesn't have that 17-18 count. In Precision the responding hand is captain, and North's 2 call sets the contract. South is expected to pass unless he has unusual distribution (which South did). And if South does have that distribution, maybe game won't be so bad. If North passes he risks selling out to 2 when 2 is making or a raise might push the opponents higher.
Oct. 9, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

I didn't say my solution was perfect. In fact, if West truly has 65432 in spades and the queen of hearts he might well choose to lead the queen of diamonds anyway, fearing that the danger of the missed cashout is greater than the missed promotion. But with 10xxxx of spades and queen of hearts West should certainly lead the jack of diamonds.

The point is that West can and should express his preference. If he leads the jack of diamonds, that means he has the queen of hearts and thinks that the third round of clubs is the best defense. If he leads the queen of diamonds, that means either he doesn't have the queen of hearts or his spades are so bad that he judges going for the cashout is best. East can combine this information with his own hand and probably come up with the best defense.
Oct. 4, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
I believe that if West has the queen of hearts he should definitely lead the jack of diamonds at trick 3. There is zero chance that declarer will duck with xx of diamonds. The point is that declarer will assume that if West isn't underleading the ace the jack lead is honest and East has AQ, so ducking can't possibly do him any good. In fact, declarer is more likely to find the tricky duck with xx if West shifts to the queen of diamonds, since now declarer can picture how the duck could fool the defense.

Once it goes jack, king, ace, small, East can put his thinking cap on. If the jack lead is honest that would mean declarer has Qx. It would be extremely unusual for declarer to play the king from dummy holding Qx, and if declarer has done so the only hope would be to promote a trump trick with the third round of clubs anyway. Therefore East will assume that West has the queen of diamonds, and it won't be difficult for East to figure out why West led the jack. Holding 10xx of hearts, East would have a trivial club return. If East had a weaker heart holding he would have to determine whether the danger of South having solid spades and a second diamond (and South guessing to play West for the queen of hearts) outweighs the possible gain from promoting a trump trick.

Since West did, in fact, lead the queen of diamonds, the conclusion is that he doesn't have the queen of hearts. Therefore, East should continue diamonds.
Oct. 3, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
Joshua says:

There is actually a different reason the whole thing could be wrong. LHO could play the J from J9 and likely not lose a trick (if he plays the J declarer will assuming singleton J or KJ but not J9, and play a low card on the second round). The whole exercise was based on the assumption the LHO must play the 9 from J9, therefore the T from T9, therefore the J from JT. If LHO can safely play the J from J9 at least some of the time, it all changes.

This is a good point, but I don't believe it applies here. The idea is that playing the jack from J9 doubleton gains when declarer doesn't have the 10 (since declarer will certainly go wrong rather than have a 50-50 guess). However, it loses any chance of declarer going wrong when declarer does have the 10 (even if declarer believes the jack is singleton or J9 doubleton, it can't hurt declarer to play low to the queen next, since if the suit is 4-1 he can cross back and lead up to the 10). My suggested strategy reduces declarer to a 50-50 guess, and I don't think any other approach can make declarer go wrong more than half the time.

A purer example of this suit combination which doesn't involve a possible falsecard from J9 doubleton would be:



where declarer has adequate entries and needs to take 5 tricks in the suit.

If declarer starts by leading the ace East can never afford to play the jack from J9 doubleton, since it can't gain and it loses when declarer has A10 doubleton.
Oct. 1, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

Perhaps he should. But maybe South won't have an entry, or maybe there are just 9 top tricks in notrump. On the actual hand, 4 has no play either.


You don't specify whether you are talking about opener or responder having the game-forcing hand.

If responder has a balanced game force opposite the 1 opener, we bid 1NT. Yes, this runs a risk of wrongsiding the contract, but other artificial approaches which try to avoid this have problems also.

If opener has 25-27 balanced, we make the 2 artificial rebid, and then bid 3NT over responder's 2 call. Yes, this is ugly and standard players do better since they have Kokish available, but the frequency of this hand is very small and most of the time any game will make on brute force so it won't matter.
Sept. 24, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

Having the 6 be the highest priority when encouraging is pretty arbitrary. The philosophy is: High is for suit-preference high, low is for suit-preference low, and middle is encouraging. The 10 is the highest spot, and the 2 is the lowest spot, so it feels right to us for these to be the strongest suit-preference signals. The six is the most middle card, so we call it the strongest encouraging signal.

No, we signal suit-preference on king leads vs. suit contracts (unless contract is 5-level or higher). On ace leads we show attitude.

The reason we give standard signals at trick 1 is that we have found that there are more situations where technical necessity prevents giving the desired signal when playing upside-down than when playing standard. A couple of examples:

Partner leads the ace vs. a suit contract, presumably from AK. Dummy has 10xx, and you have J92. Clearly it is necessary to play the 2 in case declarer has Q8x. If playing upside-down signals at trick 1, partner is likely to misread this. Coversely, if you have jack-doubleton you can afford to play the jack.

Partner leads the queen, Rusinow (or king if playing standard honor leads) vs. notrump. Dummy has a singleton, and you have A10x. If partner has KQ9xx and declarer J8xx, it is technically necessary to unblock the 10 in order to run the suit. If playing upside-down signals at trick 1, partner is likely to read the 10 as discouraging and switch.
Sept. 24, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

After 1-1, our 2NT rebid is 20-21 (maybe a good 19), which is basically like a standard 2NT opening. With 22-24, we rebid 2, which is an artificial call covering a variety of very strong hands (we don't need 2 as natural, since 1 is forcing). Responder bids 2 to ask which, and with the balanced 22-24 we rebid 2NT, which is the same as 2-2;2NT in standard. We lose on the 25+ balanced hands since we don't have Kokish available, but this is a relatively small loss due to the very low frequency of such hands. Keep in mind that responder has made a negative response, so unlike with standard 2NT opener or 2 then 2NT it is unlikely that we have a slam.

We gain when we have one of the strong balanced hands and responder has a positive response. This is a potential slam auction, and we have established a game force and responder has started describing at a low level so the strong hand can take control. Our slam auctions are far more efficient than after a standard 2NT opening, since we start at a lower level with more description and the strong hand rather than the weak hand is captain which is clearly desirable. So yes, we would be happy to give up opening 2NT in general.

We choose to use our 2 opening to show a good diamond preempt. We do this to take some of the burden off the 1 opening. Also, since responder has both 2NT and 3 to work with we have more room for a good constructive auction. We find we don't need to have a call for a good club preempt, since if we judge the hand too strong for 3 we can open a light 2 and usually survive. Hence, we use 2NT to show both minors less than an opening bid, which seems to work pretty well.

Sept. 24, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
On board 8 one would probably want to be in game since it is around 50%, although since the overcall puts more clubs in North's hand that makes it more likely that the queen of spades is offside. However, Adam found a great mesh in his partner's hand. Give East the jack of hearts instead of the jack of spades, for example, and while East will like his hand better game will have no play.

Since Adam is apparently playing Precision, I assume that the pair routinely opens 11-counts. I imagine Adam's thinking was: What is the point of opening this hand 1NT when partner doesn't figure to have more than a 10-count so we probably don't have a game.

Adam's instincts are right. The truth of the matter is that playing this style 14-16 is the wrong range for the opening 1NT in 3rd and 4th seat. Game is unlikely if you have a 14-count, so there is no need to make an opening bid which could encourage partner to get you overboard. Fred and I have found that 15-17 is a much better range for the 3rd and 4th seat 1NT opening.
Sept. 23, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
In my view, passing a forcing bid is no different from any other action. If I believe on the available information that it is the percentage call, then I have no problem doing it. The important thing is that we aren't having a misunderstanding, and that we both know that the bid is forcing.

As an illustration, a while ago I held xx AQxxx KJxx Qx. At favorable vulnerability I opened 1, and the bidding proceeded: 1-2-2-P;?

2 was a normal free bid, 100% forcing by agreement. Yet, I chose to pass. My hand was minimal to begin with, and the auction made it considerably worse since I had a bad fit for partner and my queen of clubs might be worthless. In addition, any bid I made was flawed. Even if partner has a good opening bid such as AKJxx xx AQx xxx, any game is poor, and he could have quite a bit less for the 2 call.

Could I have been wrong? Of course. If he has something like AKQxxx Kxx Axx x, which he could very well have, game is laydown and slam is pretty good. My judgment, right or wrong, was that over the range of hands the bidding would go this way that passing 2 would be the percentage action.

As it turned out, I was right. But suppose I had been wrong and we had missed a laydown game. Would partner not trust me, and be worried the next time he picks up a 16-count and makes a 2-level free bid that I might pass? Of course not. He knows that I know it is forcing, and if I choose to pass that is my problem.

If you don't agree with my choice, that is fine. But that would have nothing to do with partnership confidence. I expect my partner to make the bid and play he thinks is best, and he expects the same out of me. If passing a forcing bid is that action, so be it.
Sept. 22, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

Excellent job! One can, of course, question the choice of parameters, and different parameters would lead to different results. However, if I held those 10 hands and were asked the question: Assuming partner has a balanced 10-count, would I prefer to be in 1NT or 3NT – the results shown are very close to what my gut instinct would say on all 10 hands.
Sept. 18, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

That is exactly what I am implying. After the opening lead, I believe the defense does better (compared to double-dummy) than declarer.

Perhaps I am wrong. But speculation and “logic” isn't going to demonstrate this. Hard data is needed. If I am wrong, prove me wrong. Go through several hundred hands where the best opening lead was made, and see what the results are.
Sept. 17, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

My 1/4 of a trick estimate was based on live action results. I went through a full fall nationals tournament (everything matchpoints or BAM so every trick potentially mattered), and compared my at the table results to the deep finesse double-dummy results. The overall declarer advantage was around 1/4 of a trick.


Yes, the opening lead swing tends to be greater than 1/4 of a trick on balance. However, after the opening lead the “advantage” shifts to the defense, since it is declarer who has to guess those queens and jacks which the double-dummy engine always gets right.
Sept. 16, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

Very interesting analysis. I believe you are on the right track, and that we will be seeing more analyses like yours as our computer power gets stronger. I have some suggestions about how to improve your analysis and come up with more definitive results:

1) You don't specify the parameters on responder's hand other than saying it is a balanced 10-count. Presumably you are looking for hands which will drive to game over a 1NT opening, but will stop at 1NT (either by responding 1NT or by passing a 1NT rebid) over a 1 of a minor opening. In order to avoid complications from responder looking for other strains, I would suggest the following parameters:

a) 10 HCP
b) No 5-card major
c) No singletons or voids
d) At most 1 doubleton

2) The results would be more meaningful if converted to expected IMPs. There would be 2 sets of results – one vulnerable, one non-vulnerable. For each trial, the result would be IMP'd compared to, say, 3NT at the other table.

3) You fail to take the “declarer's advantage” into account. This is a very real thing. My studies over many hands have shown that empirically the declaring side averages about .25 tricks better than the double-dummy result (mostly because of the opening lead, of course). I think this must be included in your analysis in order to make the results realistic.

For example, on a given trial suppose the double-dummy result is 8 tricks. Incorporating the declarer's advantage, the expected realistic result would be 8 tricks 3/4 of the time and 9 tricks 1/4 of the time. So, if non-vulnerable, the expected gain from stopping at 1NT would be: +5, +5, +5, -6 = 9 IMPs over 4 trials, or 2.25 IMPs. If vulnerable, it would be: +6, +6, +6, -10 = 8 IMPs over 4 trials, or 2.0 IMPs.

Taking this approach, you could get an average IMP gain (or loss) from stopping at 1NT over your 250 trials for each of the 10 example hands. I'm sure everybody would be quite interested in the results.
Sept. 15, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.
In this sort of situation North could have been thinking about a lot of things. The use of 4 as RKC here is unusual, so North might have been double-checking the meaning. North might have been verifying that 1430 applies. This isn't just a question of counting keycards. Given that, North's BIT by itself doesn't signify anything or suggest any particular action.

On the other hand, South didn't have a marginal decision. South had a very clear pass of 4 given the meaning of the bids. As in most unauthorized information situations, the cards speak. Here the cards say that South had, perhaps subconsciously, taken advantage of the unauthorized information. In addition, the North hand says that South read the BIT accurately. Therefore, it is clear to roll the contract back to 4 if the slam made.

The question of how to deal with South's justification for the 6 call can be tricky. Here it is clear that the statement was likely self-serving. However, suppose instead South had said that she thought the 4 call showed 1 keycard, not being used to playing 1430. Then things could be more difficult.
Sept. 12, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

The main cost is the inability to make a slam try in a minor at the 4-level when a minor might be in the picture. I believe that outweighs any gains from your suggestion.
Sept. 10, 2011
You are ignoring the author of this comment. Click to temporarily show the comment.

My current philosophy is:

Below 3NT, no slam tries unless a 9-card major-suit fit has been established.

Above 3NT, no choice of games Q-bids.

So yes, my philosophy has changed.
Sept. 10, 2011

Bottom Home Top