Join Bridge Winners
All comments by Doug Bennion
You are ignoring the author of this comment. Click to temporarily show the comment.
Thanks Benoit, looks interesting.
Dec. 19, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
We play 1M is always unbalanced (natch 1), unless 18+ balanced. All 5332 hands we open 1NT (11-14) or 1 (15-17). So after 1M 1NT, opener always can transfer into a secondary suit or self-transfer, or with the 18+ balanced hand, jump to 3NT.

The self-transfers after 1NT help to break opener’s holdings into approx 11-14, 15-18-, 18+ groupings. If we open 1, and have 15 to bad 18 then:

With 55xx, we jump rebid, non-forcing. With 5-x-x-4 we jump rebid. So with AKQxx-QJTxx-Kx-x we rebid 3

With 6+ spades, we generally first self-transfer, then rebid a secondary suit at the 3-level, or 2NT with some 6322 hand. However with a good suit and 6322, we can choose to jump to 3. So with AKQxxx-QJTx-Kx-xx we self-transfer with 2, then over the (normal) 2 acceptance, we rebid 3, showing a 64xx goodish hand. The ‘pause’ that the transfers give you, lets you do that kind of thing.

With 5 spades, we generally first transfer to the second suit, then rebid 2NT or the 3-carder with some 5431. AKQxx- QJTx-Kxx-x we rebid 2 (hearts) then over the (normal) 2M response, rebid 3. With some 5422 (that we didn't consider ‘balanced’ and open 1), we might rebid 2NT

The system seems to work nicely. Secondary clubs is the weak link. The 2NT rebid is fine-fine for the most part .. I'm also not so sure about opening 1 with AKJxx-Kxx-AJT-xx. Oh I doubt it matters, but we play transfers over 1
Dec. 18, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
2NT is any unbalanced 18+ GF. It's a puppet to 3C usually. (If responder had a limit raise, he bids 3H. With a bad hand and hearts he bids 4H. With a long good suit he bids it.) After 3C, opener shows his secondary suit with 3D, 3S or 3NT (clubs). 3H sets trump and is slammish. 4H is heartsier.

I also like playing unbalanced major openings, so after a forcing 1NT I play transfers. 2C = diamonds, 2D = 6+ hearts, 2H = x5x4, 2S = ‘weak’ reverse.

Dec. 18, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
I like to use 2NT as good 18+ here: 1 1NT; 2NT. So 2 is a ‘weak’ reverse 16-18 or so.
Dec. 17, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
I always include any 5332 playing 11-14. However as somebody previously mentioned, you can get killed when you miss your 5-4 skinny and shapely game. The 5-3 misses aren't critical, partials often playing as well in notrump, and we find the games with 3C Puppet. On balance I think we win more than we lose because the 1NT opener pre-empts the oppponents out of competing with some of their own 5-3 and 4-4 fits, not to mention the ungainly fits they sometimes bid into.

More recently we stuff all 5332 15-17 hands into 1C (and play transfers over it), so our 1M openers are unbalanced or 18+. That seems to work fine as well, although opening 1C with the likes of AKxxx-AKx-Qxx-xx makes me a little queasy.
Nov. 5, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Well like Buddy and Lynn, I'm a member of the ‘played 12-14 for 30+ years’ club, the past few years stretched to 11-14 to include all those gorgeous balanced 11's. I quite like the ‘Pass forces XX’ methods (I've played many), because it does give your opps more options. In my experience, you want them to have beaucoup d'options because most of the time, they will have the balance of power, and their strong hand sits over yours. I agree those methods are better suited to matchpoints and non-expert opponents.
Oct. 23, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
My memory cells are shot, so I like to keep things simple. Real simple. So opener rebids 2D with any minimum hand, say 11-14, and ‘naturally’ otherwise. I like to play an ‘unbalanced majors’ system (5332 included in 1C or 1NT), so we always have a side 4-bagger, or 6+ in the opener. It's simplistic, but it helps sort out those frequent deals where one or both of you have goodish hands in the 14-16 range, and you have trouble sorting out if your combined holding is slammish.
Oct. 23, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Most defenders double when they are at the upper end of your notrump range, or higher. Particularly if your notrump range is wide, say 11-14, then it is usually their hand when they double, so I like a runout system that gives them a lot of chances to remove themselves from the thrashing that they were about to bestow. Using Pass as a transfer to XX, keeps their options open.

I don't know what the system is called, but Pass forces XX (to play, or DONT-type 2-suited runouts), a direct XX to show a weak 1-suiter, and direct transfer bids to show forward-going hands. This gives both opps max bidding space to rescue us.
Oct. 23, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Google suggests the chances of making a two-pointer are very close to 50%. That begs the question why a two-point attempt isn't even considered if a team has scored very late in the game and need one point to tie … why they always kick for the tie, so they can go on to play a 50% overtime. They're simply stalling their coin flip.
Aug. 6, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
I still suspect it is due to the, umm, combinatory power of honours. For example in the first 0-kings run, you have more honours in play, and the 7-point holding AQJ is worth approx 2.5 tricks, depending on K location. In the 4-ace-plus-1-king run, the 7-point holding AKx is worth just 2 tricks.

Also when you have more honours in play, as you have in your first run, spot cards gain value. For example the 4-point KJT combination is worth approx 1.5 tricks, more than the 1+ tricks for the 4-point ATx.
July 19, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
A couple of points. Several years back, when in a simming mode, I built a point count system from scratch and called it Little Jack Points. LJP at (AKQJ) = (6.5, 4.5, 2.5, 1.0) which is close enough to schmoints, but devaluing jacks slightly more. LJP also noted that most honour combinations were worth more than the sum of the individual honours … KQx was worth more than Kxx and Qxx which seems pretty obvious. I seem to recall you might add 0.5 points to most combinations.

A couple of questions. How many sims did you run? I was surprised at how many you needed before you squeezed all the randomness out of the study. Try running the ‘4 Kings’ versus ‘0 Aces’ scenarios again, but for 10X the number you previously used.

Your analysis ignores honour combinations. You conclude that jacks are not as bad as thought, but the jacks in your study are benefiting (way more than aces, say) from being part of a combination, and the values for the jacks are increased by that circumstance. Values for (more so) queens and (less so) kings will also be goosed somewhat.
July 18, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Hmmm you really need a poll with multiple choices, or multiple polls. I think you want to know what we would do when opener has 4 spades, and when he hasn't. But unless I'm missing something, we can't answer both questions :-)
July 14, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
No problems re: our written defence, have ever arisen. If in our pre-alert we forget to mention we have a written defence, we're often asked for one. We've had directors suggest we keep one handy. It's certainly a time-saver. The opps are free to adapt a different defence.

The ACBL defensive database includes two options for a Multi defence. Is it a problem that the ACBL does not recommend which option is best?
July 14, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Playing these transfers in a Mid-Chart pairs game is both instructive and amusing. You are required to pre-alert, so by the end of the day you are getting really tired of “we have a pre-alert .. we play transfers to our 1C opener .. we have a ‘defence’ if you want one”. The range of responses runs from ‘we could care less’ lol, to ‘wow that’s interesting, why do you do that'? They often come up with the suggested ‘defence’ themselves.

If the auction goes 1C (P) 1D(h) ??, we simply suggest they Double with diamonds, and bid 1H (our suit) as a takeout double for spades and diamonds. I'm not sure we are required to provide this, but we do out of courtesy. Our 1S response simply says ‘no majors’, so since it does not promise ‘four or more in a known suit’, that bid might actually require the suggested defence (double for majors), dunno.

Of course there is no ‘suggested defence’ in the ACBL database, and in fact, I believe all of those suggested defences are for opening bids, not for responses.
July 13, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
As others have suggested, just ask your local club. I've been playing TW for years in Toronto-area clubs, with permission and with no complaints whatsoever from opps. There is so much unnatural GCC-compliant stuff played over 1C (in one club here playing Montreal Relay is almost a requirement) that simple transfers are considered to be no big deal.
July 10, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
My own simulations (years ago), suggested that borderline games (24 or 25 combined HCP) were slightly more likely to make than 4M with a 5-3 M fit. However by not transferring you will miss some 5-4 even 5-5 fits, which almost always play better in the M fit. I think another factor might be how flexible shape-wise are your 1NT openers. If you open with many 5422 shapes (or with singletons), you'd be more likely to miss the really nice M fits by not using transfers.
July 1, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Lose whatever it is you decide XX should be, instead start a transfer ladder with it. So after (1) 1 (X); then

XX = transfer into clubs, here a cuebid raise
1NT is normal (leave out of ladder for positional reasons)
2 =
2 =
2 = constructive raise
2 = garbage raise
June 30, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
The human declarers are more successful because they are benefiting from imperfect human leads. Another table shows what happens when you remove the opening lead … DD play is more successful from that point.
May 7, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
Mr Bethe is right … DF will take more tricks than declarer. However the studies are also right, showing a small advantage to human declarer results. How so?

A human declarer cannot outplay a DD analyser. DF effectively peeks, and that makes it stronger than any human. Note however this is not the same as saying a human declarer cannot outplay a computer declarer; any decent player can. Typically the computer declarer deals a sample of suitable deals (which conform to the auction), DD-analyses those positions, and selects the optimum play based on that sample.

Several previous studies (including mine, long ago) suggest that dealer advantage due to imperfect leads, is in the range of 0.50 to 0.75 tricks. I’ll call it 0.60 tricks. These studies of the playing data are showing a smaller advantage in the range of, say, 0.25 tricks.

The smaller advantage actually makes a good deal of sense. The dealer advantage is worth 0.60 tricks, but that is reduced in the studies to 0.25 tricks, because subsequent to the initial lead, the human declarer underplays the DD analyser, apparently by approximately 0.35 tricks (human must underplay the analyser which sees all information).

Put a little differently, the studies compare human declarer play with human leads, to DD declarer play with DD leads. Humans ‘win’ that comparison by 0.25 tricks. However we know from previous studies that human leads gain declarer 0.60 tricks, so DD declarer play is stronger by the difference, or 0.35 tricks.

So rather than illustrate the superiority of human declaring, I think the studies quantify the superiority of DD declaring, in the range of ~ 0.35 tricks.

If your objective was to more aptly compare human declarer play to DD play, you might consider a 51-card DD solution (DF optionally could provide this). You would front-end some kind of ‘rules-based’ leads for play number one. For that one play, DF would not peek. The ‘rules’ would mimic (crudely, and unless you wanted to build bidding engines, devoid of any clues resulting from an auction) the kind of thought a human leader puts into his choice … top of honor sequence, singleton if looking for a ruff, 4th best of unbid suit, blah blah. With an error-prone lead, but perfect subsequent play, that analysis would better emulate results in the real world, I think.
May 6, 2013
You are ignoring the author of this comment. Click to temporarily show the comment.
You can also use those Transfer Double ladders after any (X) Y (Z) auction, provided you are willing to give up your normal meaning for DBL … here, usually penalty, sometimes Snapdragon. You gain a transfer into the 4th suit, plus an additional raise-type. For example (1) 1 (1); then

DBL = transfer into 2 cuebid, say a limit raise
1NT = normal, for positional considerations
2 = transfer to
2 = constructive raise
2 = normal minimum raise

You don't need limit yourself to Advancer position. The Transfer Doubles work well at Opener's first rebid, in Support Double position. You retain all the benefits of Support Doubles, and gain a transfer. For example 1 (P) 1 (1); then

DBL = transfer to
1NT = normal
2 = rebid!
2 = stronger raise (maybe ‘stronger’ for you here, is 4 trumps, in keeping with Support Double theory)
2 = weaker raise (3 trumps if you like Support Doubles)

Note the Transfer Doubles work nicely beyond the normal range for Support Doubles … after 1 (1) 1 (3) then

DBL = transfer to (transfer reverse here)
3 = stronger raise
3 = weaker raise
May 2, 2013
.

Bottom Home Top