Join Bridge Winners
All comments by Doug Bennion
1 2 3 4 ... 8 9 10 11 12 13 14 15 16 ... 32 33 34 35
You are ignoring the author of this comment. Click to temporarily show the comment.
That hand's ‘sim value’ is somewhere around 17.5 balanced. Opposite 8-9 balanced, 16 balanced makes 3NT ~ 46%, 17 balanced ~ 62%, 18 balanced ~ 78%, THAT hand ~ 72%.
Oct. 5, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Dale: Clubs doesn’t change much. When responder is 7-8, matchpointing. 2 compared to 2 wins/ties/loses 23/23/54%. Opener has 3/4/5 clubs 58/32/10%

If opener rebids 2 with 5-3-3-2, then 27/27/46%, and opener has 2/3/4/5 clubs 30/42/22/6%

Leonard: Opener has 16-17, responder 5-7. Opener has 5 spades and 4-5 diamonds (if he responds 2 with 3 diamonds, then I think he has a balanced 5332 1NT opener). When responder has 5 diamonds, then 2 compared to 2 win/tie/lose is 36/13/51%. When responder has 4 diamonds, 42/23/36% (but 2 loses avg 0.6 IMPs)
Sept. 29, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
This is coarse, I know. When responder has 9 HCP, and should be inviting, and opener with 15 HCP should be accepting or at least thinking about it:

5 makes 55%, and 4 makes 45%. 4 matchpoints better 43/23/34%, but IMPs worse by an average of about 0.9 IMPs.
Sept. 28, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
I did some sims. It's not close, you should pass. I gave opener 5 spades and 3-5 diamonds (if 3 he was 5332) and not 4 hearts and 12-15 HCP. Responder has 2 spades and 5 diamonds.

When responder has 5-6 HCP, matchpointing, 2 wins/ties/loses 9/21/70%

When responder has 7-8 HCP, 23/20/57%

When responder has 9 HCP, 37/19/44%

It becomes much more interesting when responder has just 4 diamonds .. with 7-8 HCP, 36/29/35% a virtual tie

If opener could have some 6x4x hand, and responder has 7-8 HCP, 31/20/49%
Sept. 28, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
You can do it, but you always lose some randomness (I think). Decades ago when I was building a bridge-playing program, and when computers were much much slower, I decided instead of ‘deal then screen’ I would try to ‘construct’ required hands. I would randomly deal the proper distributions, then ‘add’ whatever HCP the hand ‘needed’ I used to think of the process as ‘reasonably random’, or ‘random enough’ for my purposes.

To do that ‘reasonably’, I used a table I built that was something like ‘expected honors for these lengths of suits for an X HCP hand’. ‘So the spade ace has a y% chance of being added to this 4-card spade suit in this 16 HCP hand’. Something like that, there is probably a lot more to it, I haven't looked at that code for literally decades.

I wasn't too concerned that it wasn't 100% cleanly random. If I could get it to 98% (or whatever) accuracy, that's all I thought I needed. My choices were in one minute I can choose between (1) 5 literally random deals or (2) 1000 ‘random enough’ deals, For my purposes I thought I could get better results with (2).

So I just dragged out that old stuff, see how it behaves on a fresh new fast computer. I asked DPPro to deal AQxx KQxxx xx xx in one hand, and some 4-5-2-2 (no other constraints) in another. In 40 seconds DPPro managed to find 1! such deal in 25 million tries. In 12 seconds my stuff banged out 1000 deals. They ‘appear’ to be random, they pass a quick eyeball look, although it's possible some not-minor flaw lurks. But they would be more than OK for something like bidding practice, for example, or dealing a pool of sample hands for double-dummy play.
Sept. 25, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Extremely unlikely that sequence is in GIB's ‘catalogue’ of bids, so it is simming.

Simming that specific north holding against a 19-20 HCP 4-0-5-4 hand (as GIB described it), takes a loooong time (I did it), so GIB might not have ‘had time’ to analyse more than a small handful of deals.

In the few dozen deals I produced, making 5 was going to be slightly more likely than making 6. Not infrequently you were off two quick black tricks.
Sept. 23, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
How would this fly?

Pass 1 with hearts.

X is negative X of hearts.

Then transfers: 1 is 5+ spades, 1 is however you play 2, 1NT is natural because it has some situational value, 2 is diamonds … etc
Sept. 22, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Frances: With those definitions I agree with you. When my pool of 1 overcalls is (1) 5-6 hearts and no other 5-card suits, plus (2) blacks 4-5 long — it's very close to 50/50

When my pool of 1 overcalls is (1) 5-6 clubs and no other 5-card suits, plus (2) reds 4-5 long — it's about 43/57 clubs/reds
Sept. 22, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
And when opposite the 1NT opener, responder sees 5-6 spades in his hand, overcaller will have the 5+ heart holding 89% of the time.
Sept. 21, 2017
Doug Bennion edited this comment Sept. 21, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Here are some frequencies, I think. For the 1 case, I assumed 1 included all balanced hands (including any 5M332), or was unbalanced with other suits not longer. Tweaking that definition won't make much difference. For the 1NT case, it includes all balanced hands.

Overcaller bids 1, showing either 5-6 hearts, or 4 spades and 4 clubs.

Before he looks at his hand, responder opposite 1 can expect his RHO to have 5+ hearts 77% of the time, and 4s-4c 23%. When he looks at his hand and himself sees 5+ hearts, he can expect his RHO to have 5+ hearts 36% of the time.

Opposite 1NT by opener, 74% of the time responder's RHO will have 5+ hearts. If responder sees 5+ hearts in his own hand, RHO will have 5+ hearts 25% of the time.
Sept. 21, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Diamonds. Play transfers there. X = diamonds. 3 = unlimited heart support, 3 just competing.
Sept. 16, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
I coded (badly) a little utility that kind of did some of the things you're trying to do. I used to include it with a software package that I used to sell. Some information here:

http://www.bridgebuff.com/matchpointoverview.html

I personally used it for a few weeks then got lazy and stopped. It was starting to produce some interesting data (maybe 250 or so deal inputs), I should have kept it up, might still.

It analyzed any aspect of your system that you chose to analyze, and ‘scored’ it relative to your matchpoint average. Say our average overall score was 58%. I might have found that our average when we opened an 11-14 1NT was 55%, when we opened 1 was 62%, when we chose not to pre-empt with a very weak hand, how we scored with RHO declarer compared to LHO declarer (compared to partner I need to sharpen my leads), how we scored when I declared compared to when partner declared etc etc.

Some have pointed out that when you measure this thing here, you must be on the lookout for that thing over there. For example. your notrump range/style and your other-suit openers, especially minors, are closely linked. I think I recall our 1NT openers were performing a little below average, but our 1M openers were awesome. That might have something to do the fact we opened 5M332 hands 1 or 1NT, so our 1M openers were ‘unbalanced’ and sharpened.

Eventually I tired of the zzzzzzzz manual input routine and gave it up.
Sept. 13, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
You are ignoring the author of this comment. Click to temporarily show the comment.
We alert the ‘strong notrump’ rebid, and also the 2M raise as approx ‘either 15-17 with 4 trumps, or unbalanced with 15+ dummy values’. And so also do all other weak notrumpers we encounter.
Sept. 12, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
We play 11-14 with transfers. More than once we've been kind of ‘gently scolded’ for not playing ‘obviously superior’ 2-way Stayman. So eventually I just had to find out how eccentric we were.

I checked the convention cards for a major international event (sorry do not recall which) for weak-notrumpers. I forget the precise totals, but expert pairs who played transfers outnumbered those who played 2-way by about 25:1 or 25:0. So eccentric we are not.
Sept. 11, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
That's a super-sweet 4333 for notrump purposes, with no weak suits, and nice spots. I have that specific holding making 3NT opposite 15-17 5(332) at 76.7%, whereas 4 makes 65.0%. Repeating for a random 4(333), 3NT makes 56.2%, a 20% drop because random isn't as sweet, and 4 at 42.7%.

A negative feature that is often (usually) overlooked for notrump evaluation is an unstopped doubleton. If I move the club Q to hearts, so the hand becomes T843 AQT8 KT9 85, that hand makes 3NT 56.0% (yes coincidentally close to random 4333), whereas 4 making shoots up to 83.4%. If instead I move the heart K to clubs, so the hand becomes T843 AT8 T9 KQ85, 3NT makes just 50.6% and 4 makes 83.3%.

4333 isn't your worst enemy, ‘xx’ is. These were all 5000 deal samples.
Sept. 10, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Reminds me of the claim by a mathematician that there are no uninteresting numbers. Someone took him up on it with ‘well what’s so interesting about 29'? Mathematician replied with ‘easy — that’s the first uninteresting number'.
Sept. 10, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
If we played a strong notrump, we'd include 11-14 5M332 in 1, which would include all balanced hands below our 1NT range. However we happen to play a weak notrump (11-14), so we include 15-17 5M332 in 1.

You win some and lose some playing an unbalanced 1M. The major loss is sometimes by opening 1 or 1NT you miss a 5-3 M fit, even a 5-4 fit. We play a scrambly kind of Stayman which mitigates this issue (see my profile!). However it is a huge gain knowing that 1M opener always has either (1) extra M length or (2) a second ‘genuine’ 4+ suit. This knowledge permeates throughout your system. For example you can play 1NT as unconditionally forcing, with an interesting follow-up system. We play transfer opener rebids but I suspect other methods (Gazzilli?) might benefit as well without 5M332 clogging things up.

We make an exception for 18-19 5M332 hands which we open 1M.

Oh your 1 opener is always unbalanced as well, so in competition you can raise comfortably with 4-support, even 3.
Sept. 5, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
Yes 13 might be stretching it, but I get all kinds of squarish quacky 12-pointers that I don't want to force to game opposite some of our 1M openers. We never open 1M with 5332 (unless 18-19) and 1M can be frisky. 1NT is always forcing.
Sept. 4, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
The 2 rebid has a wide range. We use 2 like a 2NT rebid, only stronger. 2 might be 12-13, 2NT 10-11.
Sept. 3, 2017
1 2 3 4 ... 8 9 10 11 12 13 14 15 16 ... 32 33 34 35
.

Bottom Home Top