Join Bridge Winners
All comments by Doug Bennion
You are ignoring the author of this comment. Click to temporarily show the comment.
Opener who opens 1NT will be worried about missing game opposite responder with 8 HCP, who passes. He should be. That hand opposite a balanced 8 (and no spade fit) makes about 60% of the time, and a little more than 60% in a 4-4 spade fit. Aces are undervalued in the 4321 count, and that hand has about double the aces of a ‘normal’ 17-pointer.
March 19
You are ignoring the author of this comment. Click to temporarily show the comment.
It's a valuation issue. Using reasonable constraints, that East hand performs better than a typical 18-HCP balanced hand with 4 hearts, with which East would bid game.
March 16
You are ignoring the author of this comment. Click to temporarily show the comment.
I think this bot ‘error’ stems from the fact your 2 bid was thought to be showing 5+ clubs. At the point East discarded his heart, declarer was ‘supposed’ to hold 3-4 clubs and 1-0 hearts, so if necessary West's 3+ hearts could take care of dummy's 3rd heart, and East's discard selection couldn't matter. East will ‘never’ play declarer for just 4 clubs (unless revealed by previous play).
March 9
You are ignoring the author of this comment. Click to temporarily show the comment.
DealMaster Pro, using Bo Haglund's DDS, solved this in about one minute, on a year-old desktop.

When I was coding that stuff decades ago, the DD engine I crafted was lousy (= slow), so I had to resort to what I thought of as ‘spot smoothing’. I would rebuild long suits within a simmed deal so that the spots were equivalently ranked … AKT642 might become AKT654. This came at a teeny cost in accuracy, but really sped things up. If I was dealing 100 hands, a teeny flaw in a hand or two would rarely impact card selection, just get lost in the noise.

I would also walk through each simmed deal, trying to minimize choices the engine might pursue that were unlikely to be optimal. For example if the walk-through ‘saw’ that a finesse would lose (in the simmed deal, not the actual hand being played), the engine would de-prioritize taking the finesse, again (usually) speeding things up.

I'm quite sure the new engines don't have to resort to tricks like that.
Feb. 23
You are ignoring the author of this comment. Click to temporarily show the comment.
Your LHO opens Multi 2 and you are using a kind-of-standard X = 13-15 balanced, or so. Partner does double, and RHO passes. I gather it is normal to treat the double as a 4-suit takeout. You cannot play in 2NT because it is also normal to play Lebensohl to distinguish different strengths of hands. If you have a 4-card major or better, you will bid it, and failing that, you bid your longer minor.

The problem with that approach is you will land in an uncomfortable number of poor fits. When partner’s double is for 4 suits, he’ll have on average 3.25 cards for your call. If partner had doubled a standard weak two bid, he’d have on average more like 3.8 cards (giving him an average of ~1.5 cards in the doubled suit).

I simmed the position. Opener has 6 hearts or 6 spades, 5-9 HCP, and no second suit longer than 4 cards. Overcaller has a balanced 13-15 HCP (4333, 4432, 5m332). I excluded 5M332 and 5422 but doubt inclusion would make much difference.

Advancer will bid hearts if he has 4+ of them, and spades not longer. He’ll bid spades if 4+, with hearts shorter. With no 4-card major, he bids clubs when longer or equal to diamonds, otherwise diamonds.

He’ll land in these ‘fits’:

6-fit about 9%
7-fit about 28%
8-fit about 35%
9+ fit about 28%

That’s fairly atrocious, no? Some of those 6-fits are at the 3-level … advancer finds himself in 3 holding 3-3-3-4 opposite overcaller’s 4-4-3-2 say.

A little more grist. A robust 42% of the deals ‘belong’ in notrump, if you define ‘belong’ as having neither an 8+ major fit or 9+ minor fit. About one-third of those would be in a range where 2NT is comfortable.

Were overcaller doubling a standard weak 2-bid, his call a 3-suit takeout, he lands in these fits. (note he’ll be doubling 2M less often than 2 because the former is more shape-specific).

6-fit about 1%
7-fit about 11%
8-fit about 48%
9-fit about 40%

That is a gob-smacking difference. I wouldn’t want to play those 4-suit takeouts, they just don’t perform very well. Maybe I’m overlooking something.
Feb. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Thanks Andy

If I speed-read the article properly, playing just the ‘weak two’ version shows a small positive result for the actual 2 call, offset by a large negative result for using 2M as a Tartan Two.
Feb. 16
You are ignoring the author of this comment. Click to temporarily show the comment.
Steve: 5xyy hands with x = 0-2 and y = 3-4, are about as common as 42(43) hands, so now you've approximately tripled the original 4%. Thanks for the penalty double explanation. One would have to know how common those arise, to judge whether their value is greater than M transfers. Maybe someone has an idea …
Feb. 13
You are ignoring the author of this comment. Click to temporarily show the comment.
That's about how I read it. I don't see methods being built to penalize weak 2 openers, say, so not sure why the ability to penalize Multi carries such weight. I don't use it however (ACBL), and they do …
Feb. 13
You are ignoring the author of this comment. Click to temporarily show the comment.
I defined ‘double’ pretty narrowly for hearts as 4xyy, with x = 0-1 and y = 3-5. That produces a frequency of ~ 4%. If you include the likes of 4-2-(43), that approximately doubles the frequency, so that might explain your sense of it. I included 4-2-(43) in the ‘balanced’ category. Doubles of (42)(43) will be for the ‘wrong’ major about one hand in four.
Feb. 13
You are ignoring the author of this comment. Click to temporarily show the comment.
If you remove an ace or king from the hand, should weak notrumpers consider 1?

A consideration favoring 1NT (for any range) is that it pre-empts overcaller with a decent club suit from showing his hand (2 almost always conventional over 1NT}. It likewise prevents overcaller making a takeout double and subsequently finding their nice club fit (although with this opener they might not enjoy finding a major fit).
Feb. 9
You are ignoring the author of this comment. Click to temporarily show the comment.
A good chunk of those hands with small doubletons in the Fantasis analysis would be 5332 (I'm guessing), with the 5-suit often being a source of tricks.
Feb. 9
You are ignoring the author of this comment. Click to temporarily show the comment.
You didn't finish you eejit. How does that hand compare in a 4-4 spade fit? Perhaps the ‘xx’ isn't as major a defect.

Right. In a skinny game context, that hand playing in a 4-4 4S contract opposite 8 HCP, makes 4S 50%. A random 17-pointer in the same context makes 54%.

So if that hand finds a 4-4 M fit, it's slightly better than average. If there is no fit, it's run-of-the-mill.
Feb. 8
You are ignoring the author of this comment. Click to temporarily show the comment.
For what it is worth, DD sims think that hand is a precisely average 17-pointer. That hand opposite a balanced 9 HCP makes 3NT 68% of the time. A random 17-HCP 4432 hand makes game 68%. A random 18-HCP makes game 82%.

The honor combinations are nice, but ‘xx’ brings the performance back to average.

(2K sims)
Feb. 8
You are ignoring the author of this comment. Click to temporarily show the comment.
Not sure this example is a prime example of bot ‘foolishness’. At the point the 7 was played, Bot E had simmed N deals, and none of the layouts favored the 5 over the 7, so E played randomly. I’d be guilty of the same play, never projecting that my 7 might grow in stature to that extent.

Not helping matters is the fact the bot would never sim that specific declarer hand which is a jack short of what he ‘should’ have.
Jan. 27
You are ignoring the author of this comment. Click to temporarily show the comment.
I have little personal experience playing with the BBO bots, but a significant source of error seems to be taking too literally the ‘announced’ point ranges. The 4 bid showed a maximum of 10 HCP, declarer had already shown up with 11, so west ‘had’ to have the spade jack. On the 2nd lead, west still has to have that card, and east does not ‘remember’ that west failed to play it on a previous trick.
Jan. 4
You are ignoring the author of this comment. Click to temporarily show the comment.
In some sequences, where X (or XX) is a legal call, you can show both ‘good’ and ‘competitive’ raises via transfer bids that begin with X. In your example (1) X (P) 1; (2) ??, transfers begin with X = diamonds, then 2 = hearts, 2 = your good raise, 2 is competing. If responder had ventured (1) instead of (P), now X = diamonds, 2 = heart cue = great raise, 2 = good raise, 2 = compete.

In effect you swap whatever current use you put to X, for an additional raise type, PLUS you get to bid again, which is always good.
Jan. 1
You are ignoring the author of this comment. Click to temporarily show the comment.
Right. When overcaller has a weak notrump, the hand belongs as much to his side as the opening side, but opener got licks in first for his side. Overcaller takes on a little more risk, but you find all kinds of fits you wouldn't otherwise. What's more, you'll have 11-14, say, about 4X as often as 15-17. Often both sides can make 1NT, but you got there first.
Dec. 29, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
As a matchpoint tactic, try playing a 1NT overcall as weak. You need escape routines and some kind of scramble Stayman, but the overcall is a real nuisance to the opps. Once in awhile you get nailed, but hey that's one bottom. You'll also have to somehow manage your balanced 15-17 hands.

Weak notrump overcalls also ‘sharpen’ your takeout doubles. You still X with traditional shapes, but you have something to bid with 13 HCP and 4-2-4-3 when RHO opens 1.
Dec. 29, 2018
Doug Bennion edited this comment Dec. 29, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
2 is clubs or balanced, and 5 is a bucketload. So my answer is 1.5.
Dec. 29, 2018
You are ignoring the author of this comment. Click to temporarily show the comment.
I think your theory is a good one. Back in the day I used to program and market a bridge program, and one of the problem areas was valuing a hand especially in competitive auctions, when ‘shape’ can be at least as important as HCP. Here for example the 4 bid ‘showed’ a rebiddable heart suit (no kidding) but doesn't state what the minimum length might be. If it was 6+, there is more room for diamond cards, with 4 less likely to make.

From responder's point of view, what does ‘rebiddable’ suit mean? Does AJ9xxx qualify? KQxxxx? Heck if opener has one of those rebiddable suits, plus 21 HCP with room for diamond cards, I might try 5 myself.

Also 4 allegedly showed 21 HCP which is a clear fumble.

I just checked what my old program did with that hand. Opener also rebid 4, showing 7+ and 18 ‘points’ (total, not HCP). Partner passed 4, the proxy hands for opener being dealt not so narrowly. I see if opener had gone on to 5, that would have shown 8+ hearts, and 6 would have shown 9+ hearts.
Dec. 28, 2018
.

Bottom Home Top