[EricK]

Approximately what proportion of hands does an expert player declare perfectly? And how about on defense? By perfect play, I mean not only playing the right card (on a single dummy basis), but playing it for all the right reasons.

[Fluffy]

Appart from grands claiming and similar I doubt any hand is played perfect and for the right reason. the hand has to be really simple for a level 4 or lower to be able to be played perfect

[FrancesHinden]

I think a lot of hands are quite simple to play, and to some extent 'perfect' is subjective (e.g. opening lead). I think it's well over 50%.

[hotShot]

I don't know why they play the rights cards, but i can tell you from "serious" BBO play versus a DD-Solver analysis that they lose less than 1 trick in 4 boards to the DD-Solver.
And since this count includes unlucky leads and not fishing a single K offside, the error rate is even lower than that.

[Mbodell]

hotShot, on 2011-December-04, 11:15, said:

I don't know why they play the rights cards, but i can tell you from "serious" BBO play versus a DD-Solver analysis that they lose less than 1 trick in 4 boards to the DD-Solver.
And since this count includes unlucky leads and not fishing a single K offside, the error rate is even lower than that.

That is only a little bit helpful as there could be conflicting mistakes where declarer/defense doesn't take advantage of the mistake. In addition, often an inferior line will work fine on "normal" splits but a perfect line would pick up some/all of the bad splits - and if the board breaks normally your DD-solver will not ding the player who takes the working but poor line.

[Statto]

Less than 100% but close, maybe 99%. Consider the large number of routine hands, and that most of the examples of 'interesting' hands from the recent world championships revolved around the bidding. Also, the best % line may be due to a deduction from the bidding, something which is hard to judge even post-mortem.

[hotShot]

Mbodell, on 2011-December-04, 21:09, said:

That is only a little bit helpful as there could be conflicting mistakes where declarer/defense doesn't take advantage of the mistake.

This is included as I compared the DD-Solution prior and after every played card.

Mbodell, on 2011-December-04, 21:09, said:

In addition, often an inferior line will work fine on "normal" splits but a perfect line would pick up some/all of the bad splits - and if the board breaks normally your DD-solver will not ding the player who takes the working but poor line.

This is of cause true.

[han]

Statto, on 2011-December-04, 21:46, said:

Less than 100% but close, maybe 99%.

Either you are much better than I am, or you are way off.

[gnasher]

I agree with Han. I think even the very best players make tons of mistakes, most of which don't cost.

[gwnn]

Where was that article about three (very, very high level) experts playing a simple suit combination (I think 8 card fit missing K and T in some configuration) differently in the same contract? I think there was an American, an Italian and a French top declarer. At most only one of them played that particular hand "perfectly" (the combination had only one correct single dummy solution).

[EricK]

Statto, on 2011-December-04, 21:46, said:

Less than 100% but close, maybe 99%. Consider the large number of routine hands, and that most of the examples of 'interesting' hands from the recent world championships revolved around the bidding. Also, the best % line may be due to a deduction from the bidding, something which is hard to judge even post-mortem.

I think you massively overestimate how many hands are genuinely "routine" i.e. there is no subtlety to the play at all, even to take account of very extreme distributions.

As to your last point, when, at the end of a session, an expert goes over the hands they played (which I'm sure they do), isn't one of the things they look for any deductions they might have missed?

[FrancesHinden]

gwnn, on 2011-December-05, 06:26, said:

Where was that article about three (very, very high level) experts playing a simple suit combination (I think 8 card fit missing K and T in some configuration) differently in the same contract? I think there was an American, an Italian and a French top declarer. At most only one of them played that particular hand "perfectly" (the combination had only one correct single dummy solution).

That doesn't necessarily prove the point you want to make, because suit combinations don't necessarily have one correct single dummy solution.
I remember a slam I played in a very close KO match a year or so ago. I spent about 15 minutes trying to decide how to play the trump suit. Half of that was working out the correct single-dummy line. The other half was spent deciding not to the take the single dummy correct line because I decided that there was an inference from the choice of lead about the trump suit. I don't know if I was correct to take the inference or not, but we gained 17 imps. It's not obvious which of my opponent in my seat or I took the 'correct' line.

[aguahombre]

I will guess that the one which gained 17 IMPs was the correct line.

I watch a lot of Vugraph, and have no idea about all the reasoning going on in Declarer's brain when he chooses a different line than the commentators or we peons would have taken.

Some things happen at the table to which we are not privy; past history with the opps, nuances of a competitive auction we haven't considered, state of match, falsecards to disrupt defensive signals, extraneous stuff. It would never occur to me to try and figure out the answer to the OP question with all the variables involved.

I just watch, enjoy, and try to figure out what happened.

[Statto]

EricK, on 2011-December-05, 12:20, said:

I think you massively overestimate how many hands are genuinely "routine" i.e. there is no subtlety to the play at all, even to take account of very extreme distributions.

I would imagine top players could and would generally find the 'correct' safety plays. But there are other considerations such as the value of an overtrick vs the risk of going down. Those small gains can rack up over the course of a match (with careful risk analysis).

Quote

As to your last point, when, at the end of a session, an expert goes over the hands they played (which I'm sure they do), isn't one of the things they look for any deductions they might have missed?

Sure. The 'correct' play based on the bidding or the defenders' play is likely to be subjective which is why it's always good to discuss these things . What I was suggesting is that it is rare for an expert declarer to make what might be considered a mistake.

SuitPlay may tell you which way to play a suit in isolation, but an expert (which I'm not BTW) will take account of many other factors to decide what is best for the specific circumstances. And I think in 99% of cases their reasoning will hold up.

[xcurt]

At double dummy, this question was answered by Ginsberg along the way to building GIB, see: http://arxiv.org/abs/1106.0669. The most interesting thing about the relevant part of this paper is that along the way it makes the point that if you're going to try to induce misdefence, there's a huge edge to putting the problem to the defenders early.

At single dummy, well, we don't have a good single dummy solver yet...

[Cascade]

gwnn, on 2011-December-05, 06:26, said:

Where was that article about three (very, very high level) experts playing a simple suit combination (I think 8 card fit missing K and T in some configuration) differently in the same contract? I think there was an American, an Italian and a French top declarer. At most only one of them played that particular hand "perfectly" (the combination had only one correct single dummy solution).

That is not necessary. It is possible the declarer's had different information which led to different lines.

I recall a Grand Slam at two tables which eventually required a two-way finesse but there had been different opposition bidding at the two tables. The finesse was taken one way at one table and the other way at the other table. Looking at the hands later I decided that both declarers had played well but only one had got lucky (or unlucky).

[Phil]

gwnn, on 2011-December-05, 06:26, said:

Where was that article about three (very, very high level) experts playing a simple suit combination (I think 8 card fit missing K and T in some configuration) differently in the same contract? I think there was an American, an Italian and a French top declarer. At most only one of them played that particular hand "perfectly" (the combination had only one correct single dummy solution).

A92

QT876

for 4 tricks.....

[nige1]

gnasher, on 2011-December-05, 06:17, said:

I agree with Han. I think even the very best players make tons of mistakes, most of which don't cost.

Agree with gnasher. Typical examples

• Not choosing the best percentage line (for instance settling for a 95% line when there is a 99% line).
• Not taking into account what you know about the other players' philosophy and psychology.
• Not providing opponents with the maximum opportunity for error.
• Not choosing the best deceptive play when it is possible that deception may be necessary.
• Not correctly "randomising" your choice of plays when games-theory dictates a mixed-strategy.
• Taking too long to work all this out, so that the partnership has insufficient time later.
• Expending too much effort on unimportant hands, so not conserving adrenalin for later important decisions.

And I agree that, luckily, most mistakes don't cost and/or go unnoticed

[Cascade]

There are so many problems that no one knows the answer to. Its not possible for players to knowingly some of those hands correctly.

While we know that 3=2 breaks occur 68% a priori. Probabilities vary depending on information from the bidding and play. Given that additional information some of which is subjective no one knows what the real probability very often. Sometimes the information will only move the a priori probabilities an insignificant amount and therefore the best play will not vary from playing the hand without the information. Sometimes the information will vary the probabilities a critical amount and the best play will vary.

[gwnn]

Yes, Frances and Cascade have it right, but I somehow remember that in this case the play went identically in both rooms and there was no interference in the bidding. Yea yea, they were playing different systems (and their opponents had different temperaments) so the negative inferences available from the opposition silence were different. So my conclusion is far from logically sound. I will stop now.