tysen2k

As was said earlier, it's probably best for you to review what has already been done. Here and here are the best places to start. No point in reinventing the wheel. Also about the accuracy of DD data compared to real world declarers. Peter Cheung did an extensive study of 383,000 okbridge hands (25 million plays) and found that on average there is only 0.1 tricks difference. A DD declarer has the advantage in slam contracts, but the DD defenders have the advantage at partscores. Around game, DD is very accurate. Tysen

4♠ is a stretch after a balance, but I still think it's right. You shouldn't have posted partner's hand though because that could potentially bias people's responses. Looking at both hands, why haven't the opponents pressed on with their 11-card fit? If we were looking only at the east hand, after this bidding I think you would be surprised to find a singleton in partner's hand. Tysen

If a transfer is doubled, you can also play that pass could contain some hands with support, but are hands that don't gain anything by having the opening lead come up to them. So in these cases you can sometimes make partner declarer and thus putting the doubler on lead himself. Tysen

It may not be obvious, but my real interest is not in finding the most accurate evaluation method, but trying to come up with a methodology to measure how that strength changes as the bidding progresses. Hands are dynamic and everyone knows that they change in value, but by how much? You can come up with rules and say add a point or two for honors in partner's suit, add or subtract a point for shortness there, etc. But where do those values come from? Usually the author just guesses at a reasonable number and sees if it makes sense with a few examples. How do you know you should adjust by 1 point instead of 2 or 3? My research is mostly in trying to figure out a quantitative way of measuring the change in valuation during the bidding. You pick up: ♠Kxx ♥Qxxx ♦x ♣Txxxx Not a superb hand, but what if your partner opens 1♥? How much is your hand worth now? Your RHO then overcalls 1♠. How much does this change things? And let's not forget that measuring offensive strength and defensive strength are two different things. You can't use one point system to measure both. If you decide that you want to give priority to offensive strength measurement, then you'll open light distributional hands. But you might not be able to double the opponents if they bid over you. You have to decide which you think is more valuable. No evaluation method will substitute for judgement. I don't really use complicated methods at the table. But what looking into this kind of research has done is give me a feel for what the correct magnitude of change really is. I still use my own judgement all the time, but this gives me a good starting point. Tysen

My vote is for 3♦, which the way I play with my partner shows limit+ with exactly 3-card support. After partner rebids 3N, I'll cue with 4♠. Although partner may have wasted diamond values, my hand is a lot better than a minimum and I need to say something. Partner should have at least 17-18 HCP, maybe more. Tysen

Don't open the 4432 hands with 2♥. A lot of the time the right final contract is going to be 1NT. Oops. Keep your definition the way it was and pass as responder on those 3352 hands. Tysen

Zar is describing what he considers to be an accurate measure of the strength of a hand, that is its trick taking potential. Note that a system for measuring the strength of your hand is totally independant from the requirements you have to open. You can use the ABC system to decide how strong your hand is and the XYZ system to decide if you want to open. I won't go into a long rant about why Zar points aren't the most accurate measure of strength, since I've said plenty on this forum already. The short story is that yes, Zar points are more accurate than HCP, but there are other methods that are more accurate and simpler to use. Tysen

Was 1N forcing? If it's not then I don't think they will bid the game voluntarily and I'll settle for 3♥. If it was forcing then West could be hiding some values or support and I'll jump to 4♥. Someone has longer spades than they've shown. Neither of these bids should show any points, just shape. Someone also has to have the missing points. I bet it's partner. Tysen

I think 3♠ could be made on a rare 3-card support. What would you do with a balanced minimum with no heart stopper? I think the double by itself is practically game forcing. While I like my spades, my values in hearts are likely wasted, so I don't think I have much more than a standard opener. Partner probably doesn't have more than 4 spades, unless he is very strong. I'm going to bid 3♠ and expect my partner to bid on almost every time. Pass and 3NT are also both good and could easily be right. Tysen

When I was teaching a group of brand new students, I put a bunch of bridge "rules" up on the board. The first rule was: 1. Every bridge rule has its exception - including this one Tysen :P

There is no expert consensus, but let's say for this problem the partnership agreement is that doubling and then bidding shows 17+. I think modern bidding theory is pushing this higher and making a simple overcall with a higher maximum. Tysen

[hv=d=w&v=e&s=sxxhjxxdaqxxxcxxx]133|100|Scoring: IMP (1♣) - Dbl - (P) - 1♦ (1♥) - 1♠ - (P) - ?[/hv] Very simple bidding but it left me unsure of what the right response is. Any opinions?

Yes they were questionable, but in studying the system and his results, it looked like the reason his results were so strange was because his assumed continuing effeciency was too high. The opening bid definitions that his computer eventually produced said to pass on ~80% of all hands. What was happening was that the computer assumed that the follow-ups would be so effecient that it didn't want to waste any precious bidding space on opening since his partner would be able to open better than he could. I've talked with Matt about this and he agreed with my conclusions. In duplicating his work with a lower effeciency I'm producing much more "normal" results. Ginsberg did abandon the project but that was because of a lack of time, not interest. This was before GIB was released and so once that became a product he spent his time maintaining that. He did try to start to tackle the problem again about a year ago, got some programmers to volunteer to help him, and then everyone ended up flaking on him. Tysen

I've seen a 1♥ bid used as a negative response, but the beauty of the simulations that I ran is that you don't even have to know what the continuations are to be able to judge the merits of the opening bids. For those of you who are wondering how that's even possible, I use a method described by Matt Ginsberg here. The basics are that you don't assume what the continuations are going to be but rather assume that you will be able to define the continuations later at some given level of efficiency (I assumed 18% which seemed to work out quite well). Tysen

Within a precision or strong diamond context, a natural 2♣ bid was best when it promised 6+ cards and no 4cM. The optimal point range was also quite wide, something like 8-18 as well, although points didn't make that much of a difference and narrowing the range did not hurt much. The optimal range for a natural 1NT opening is up for discussion, mainly because of the double-dummy assumptions used in the simulations. In real life at the table, after 1NT - P or 1NT - 3NT a declarer will usually do much better than double-dummy results. So the simulations felt that 1NT was doomed a lot more often than it really was and it sought shelter in a suit contract. Therefore the simulations rated a weak NT worse than it would fare in real life. I made my own adjustments to the DD results, giving bonus fractional tricks to NT contracts, based on Peter Cheung's research on how DD results compare to SD declarers. Even after these adjustments, the sims still thought strong NT was better. The sims also felt that a wide point range was acceptable, and if the minimum points are high enough, you should be able to handle the wider range. Something like 17-21 or even 17+ or 18+. Playing that at the table should raise some eyebrows of the opponents. B) Tysen

When I did my simulations, the definition of each bid measured not only the effectiveness when it was bid, but it's effect on other definitions as well. Although I did not do an exhaustive search of all possible definitions, I tested several possibilities within a natural system. The strongest thing suggested by the simulations was that unbalanced hands need to be opened as much as possible, even when weak. Balanced hands can afford to pass at first, even if moderately strong (thus my earlier proposal of a 0-16 balanced pass). The simulations also greatly valued a 5-card spade suit. It felt that whenever you had one, that information was so valuable to give to partner that you should open 1♠ whenever possible. I don't remember the exact numbers, but within a GCC context I think 8-18, 5+ cards was the best range if unbalanced, 11-18 balanced. For a 1♥ bid it prefered a more narrow point range with only 4+ cards. Something like 8-13 unbalanced, 11-14 balanced. I also found it interesting that the simulations suggested that a strong 1♦ was superior to a strong 1♣. The extra bidding space was more valuable for the limited hands than it was for the strong ones. Tysen

For me "to play" means that I expect partner to pass 97% of the time. I would never describe 1m-2N (NF) as "to play" but rather as "non-forcing" or "invitational" Tysen

Has anyone run more hands? Is the dealer still at a disadvantage? How about trying different convention cards? This seemed like a real puzzle to me and I'm interested to see if it gets sorted out.

Wayne, Thanks for starting the simulations. What convention card are you using for Jack? Maybe it should be done twice, once with a conservative approach and once with an aggressive one. Actually with only 200 deals, we're not quite at the 95% significance level, but it's getting close. I think we need at least 1000 deals. Tysen

This is a question I posed to rgb several months ago and I got a lot of "that's a cool idea, I'll look into that" but it never went anywhere. I wanted to try to actually quantify how much is the "dealer's advantage." If anyone on this forum is good at running this kind of simulation, I'm sure everyone would be interested in seeing the results. Have GIB or Jack run a team match of a few thousand hands with the computer playing in every seat. At one table North is always dealer and East is always dealer at the other table. This has the effect of one team always having the advantage of being dealer on every hand. How much does this team win by?

I added the fact that 2♥ shows 5+. Our 3♥ bid could be made on xxx but not 4-3-3-3 shape. Does anyone want to draw any inferences from the fact that partner went to 4N over 3♥ instead of going slower?

IMP's Both vulnerable ♠AQxx ♥JTxx ♦xx ♣Qxx 1[cl] [space](art. strong) [space] [space] [space] 2[di] (9-11 balanced) 2[he] (nat, 5+) [space] [space] [space] 3[he] (3+ support) 4NT (RCKB) [space] [space] [space] 5[cl] (1 or 4) 5[di] (Do you have the HQ?) [space] [space] [space] ?

I'm interested in people's preempting styles. Matchpoints, none vul ♠xx ♥AQJTxx ♦Txxx ♣x (1♠) - ?

Okay, so you're at 4♣ with neither partner having shown shape information. Good thing you were able to stumble into a club fit. So if responder had 6 hearts and no other 4-card suit I assume he would bid 5♥ over the 4♠ relay?

6 KT976 QJT 6543 T94 A AK98 AKT84 Pavell, please indulge us with two examples. Please show us how these 2 hands bid when South opens 1♣ and West overcalls 3♠. Now do the same for these 2 hands (which I assume is still a 1♣ opening): 6 KT976 QJT 6543 T94 A K842 AT852