Join Bridge Winners
All comments by David Parsons
1 2 3 4 5 6 7 8 9 ... 18 19 20 21
You are ignoring the author of this comment. Click to temporarily show the comment.
And this is a “weakness” of the simulations? All I can say is “Wow…”
Jan. 19
You are ignoring the author of this comment. Click to temporarily show the comment.
Maybe there is a symantics problem here. In my view, “Matchpoints” is MPs, not IMPs. Being in 1NT +90 instead of 2 +110 costs the same as being in 1NT -200 rather than 2 +110.

Read: http://www.rpbridge.net/6b00.htm, where it clearly states that “at matchpoint scoring it makes no difference by how much your score differs from other scores.”
Jan. 18
You are ignoring the author of this comment. Click to temporarily show the comment.
You guys are having way too much fun. :-)
Jan. 18
You are ignoring the author of this comment. Click to temporarily show the comment.
You have a point – 6 is an good slam with these two hands.
Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
I'm missing your point. This is an MP problem, not an IMP problem, so unless you're thinking that vulnerability makes a difference to opponent's bidding (in which you would have a point), the comparison of 1NT to 2 is insensitive to vulnerability.
Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
Doug, I ran Bridge Analyzer with the same sample size and biases as you state above, and I can confirm your numbers as being replicated by Bridge Analyzer, within a tolerable margin. Bridge Analyzer does take it one step further and calculates the advantage at MPs by comparing the DD result of each individual hand and contract. To use two of your examples:

When heart suit is xxxxx, 1NT has an MP advantage of 68% to 32%.
When heart suit is AKxxx, 1NT has an MP advantage of 56% to 44%.

These all use the 7-9 HCP range in your biases, and not the 7-10 HCP range in the OP.
Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
Actually, if you fix opener at two hearts, the simulation says 2 is better than NT whenever responder is at least as shapely as 5-4-2-2. But if responder is 5-2-3-3, 1NT is marginally better than 2.
Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
I confess that I was the partner that pulled the 3NT contract to 4. And, when dummy came down and showed:

xx
Axxxx
Kxx
xxx

I couldn't have been happier that I pulled it; I was so certain that with that dummy the hand must play better in 4, because we are so exposed to a spade lead.

When I got home, I checked my intuition against the simulator, and to my surprise, even with that dummy, we should be playing in 3NT by a margin of 52% to 48%! Wow, was I surprised – I expected that 90% of the time 4 would be better with that dummy.

Now to understand the reason – With a spade lead, opponents can not hold on to their spades to make use of them, because they have to protect the other three suits from setting up the extra length!

I'm so grateful to the simulator for correcting my intuition so that I'll have better intuition in the future. And also thankful for an understanding partner, who I put through the ringer sometimes with my experiments and too-strongly-expressed intuition, that turns out to be wrong not infrequently.
Jan. 17
David Parsons edited this comment Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
You might be surprised to learn that the simulator says that under the conditions of the OP:

Qx Q6432 Jxxx Kx favors 2 over 1NT at MPs by 68% to 32%.
xx KT987 QJT9 Qx favors 2 over 1NT at MPs by 69% to 31%.

Not much difference between those hands, according to the simulator.
Jan. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
Hah…yeah, me too in robot games – the robots are so stupid. :-)
Jan. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
If you open 1 and rebid 2, we're aware that you could be 4-5 in diamonds/clubs. So, we assume the minors are of equal length and show a preference for diamonds only if we're longer in diamonds than in clubs. But, yes, if the clubs are strong enough compared to diamonds, we can rebid clubs and lie about the length.
Jan. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Yep, we would open that hand 1 and rebid 2. Provide a hand where you get into trouble doing that, Paul, to prove your point. Meanwhile, all those factors you suggest don't actually enter into the equation. If opener is forbidden from rebidding 1NT with a singleton heart, you should always rebid 2 according to the simulator. If you know that partner only has two hearts, then you should bid 2 only if you are distributional, but it's pretty close even if you're not distributional. If partner has three hearts, 2 is a no-brainer.

Reaching the right contract of 2 seems to be a plus for the system where 1NT promises at least two hearts.
Jan. 16
David Parsons edited this comment Jan. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Argh — I meant 12-14 HCPs. And my simulation says the same thing. Always bid 2.

I finally was able to change it…sorry.
Jan. 16
David Parsons edited this comment Jan. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Right you are — sorry! I changed it; thanks!
Jan. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Nicholas, I ran the same constraints you did on Bridge Analyzer but for 10,000 deals instead of 1,000. According to Bridge Analyzer, doubling scores 69% (not 62.4%) vulnerable vs bidding 3 spades non-vul and doubling scores 71% (not 73%) vs passing. You're right that the reopening doubler isn't guaranteed to have four hearts, but I thought I'd be generous with respect to testing the premise that Passing in the Balancing seat is never the right move (and therefore my conclusion that agrees with that premise). Certainly if the reopening doubler has less than four hearts, a bid is that much more compelling.

So, your results are in general agreement with Bridge Analyzer, though I suspect Bridge Analyzer is more accurate; thanks for confirming this.

I am curious why you think that Bridge Analyzer has any problem at all. What are you seeing that you think is suspect? I assure you that Bridge Analyzer has no problem with the random hand generator given a set hand. It sets the hand, deals the rest of the cards randomly, and then checks to see if the other criteria are met. If they are, it keeps the hand, else it discards it. After getting 1,000 hands (or 10,000 hands…whatever I want) that meet the criteria, it then runs a DD solver to determine the score at MPs between two contracts.

After testing things on DealmasterPro, do you agree now that “Pass” is the worst of all the options in the OP?
Jan. 15
You are ignoring the author of this comment. Click to temporarily show the comment.
The simulation biases do not give opponents 8 hearts. Stated again (this is a long thread, so it's easy to miss that I stated this before), the biases are:

“(1) Partner has 8-9 HCPs, exactly three spades, and 2-3 hearts.
(2) The reopening doubler has four hearts, less than three spades, at least three of each minor and at least 10 HCPs.”

With the 5=2=3=3 shape stated in the OP, this bias means that opponents can have eight or nine hearts. You are right that I am making the biases even more conservative to my point by eliminating the fact that opponents could have seven hearts…but if they do, that proves my point even further. Plus, I think that partner will be doubling 3H with his four hearts, if they have seven hearts.

My hand was also stated in this thread:
“Responder had 3=3=3=4 distribution with the ♦AQ and ♣K.”

The program I am using is Lorne Anderson's Bridge Analyzer program, which is available for purchase at:
http://myweb.ticali.co.uk/lorne.anderson/Bridge/

This program allows me to put in the biases above. That program also allows me to analyze matchpoint results (via the “DD IMP” option) and put in the vulnerability. I use 1,000 random deals that meet the criteria of the biases and the shape/HCP criteria of the OP.

It's always possible that I've put in the wrong parameters, so let me know what you can't reproduce and I'll try it again.
Jan. 15
You are ignoring the author of this comment. Click to temporarily show the comment.
Perhaps you're addressing another hand and not the original post (OP).

No, you can't have 10 or 11 HCPs and have the shape dictated by the OP in first position unless you're playing something other than standard.

No, you can't respond with 5HCPs and a singleton, as the constructive raise dictated by the OP. That's not a constructive raise.

And finally, no, it is highly unlikely that opponents will double you non-vulnerable if they understand matchpoint scoring and LOTT, given that you are non-vul (the worst vulnerability for a double) and that your partner made a constructive raise of your 1S opening.

P.S. And I was responding to your comment “With only 12 HCP, passing might be right,” which (thinking that the conditions were met in the OP except 12 HCPs rather than 14 HCPs) I did not agree with.
Jan. 14
David Parsons edited this comment Jan. 14
You are ignoring the author of this comment. Click to temporarily show the comment.
DD simulation shows that *if* partner has four spades and six clubs, then the best place to play at matchpoints is 3, not 2, by a margin of 68% to 32%. And, according to another of my polls, not a single player out of 70 believe that the reopening double guarantees four spades.
Jan. 14
You are ignoring the author of this comment. Click to temporarily show the comment.
There's some wisdom in what you say, Bruce, but only because partner made a constructive raise, which gives him some defensive strength. However, having run a number of simulations, I don't agree that you should pass with 12 HCPs at MPs…it's almost always wrong at this vulnerability and at matchpoint scoring. The general rule, supported by simulation, with 12-13 HCPs seems to be that if you have quick tricks outside of spades, you should double hearts, but if you don't you should bid 3S. The rationale is completely supported by LOTT, and is as follows:

(1) LOTT says that both 3 and 3 will not make.

(2) Your partner gave you a constructive raise, which means that 2 is almost certain to make.

(3) From (1) and (2), LOTT will tell you that either 3 will go down one or 3 will go down at least one.

(4) If you don't believe 3 will go down, then you must bid 3 at MPs because, even doubled, 3 will cost only 100 points while 3 making will cost 140 points.

(5) If you believe 3 will go down, then you must double 3 at MPs because, making 2 or 3 will score higher than 3 down one undoubled.

(6) Human judgment and hand evaluation about whether 3 or 3 should make (with your knowledge of partner's three-card support constructive raise hand) has to be better than flipping a coin. But, if you need to flip a coin, it's not going to give you a worse matchpoint score than passing.
Jan. 14
David Parsons edited this comment Jan. 14
You are ignoring the author of this comment. Click to temporarily show the comment.
You expect that there's a slight chance that 1-1NT is passed out when opponents have nine hearts between them (eight is not possible)?

Not in my club, it's not. But if it is, NV, I'd love the top score at matchpoints…thank you very much. And if you think that I'm worried about the matchpoint score if the bidding is 1-1-1-1NT, think again.

The reason that I wouldn't open 1 is the one bidding sequence you didn't mention… 1 (1) X (P) 1NT. That's the one I don't like, with a singleton heart. But it's not terrible.
Jan. 14
David Parsons edited this comment Jan. 14
1 2 3 4 5 6 7 8 9 ... 18 19 20 21
.

Bottom Home Top