antonylee

I dislike the 2NT response to 4SF -- that'll wrong-side 3NT with an awful frequency. Sure, if responder insists then opener should bid 3NT but if the Ds afre for example Ax opposite Qxx 3NT will be better from responder's side.

I think Yu is right, (in a pickup partnership) I should just bid 6♣ over 5♣ and hope partner does not have the only death hand (QJxx xx Q AKxxxx). I agree with bidding 7 at matchpoints; not that sure that I'd do so at IMPs (is the spade lead that clear? I don't know). Now for another challenge: get to slam if South deals.

Interesting, however with my regular partner 4♠ would be LTTC (♦ control, but -usually- no ♠ control). Well actually it wouldn't because we would be in a minorwood auction but that's what it would be were we in a major instead.

Can you bid to 6♣ (or more, if you like 50% grands) on these hands? [hv=pc=n&s=sa5hkqj86dkj3cqjt&n=st863hada7ca97654&d=n&v=0&b=1&a=1cp1hp1sp2dp3cp4cp4dp4hp5cppp]266|200[/hv] I gave the auction I had last night (playing regular 2/1 with a pickup partner). His 4♦ looked like showing shortness to me so I gave up after his skipping the spade cue. My solution below...

For sure, a relay system sounds best for robot constructive auctions. Unfortunately (afaict) they won't cover competitive auctions, so I am not sure you can completely bypass the problem of "poorly defined" bids (i.e., auctions where sampling is hard) and auctions where both partners must exhibit "judgement" (or at least cooperate somehow).

That is the whole point. I understand very well that simulations work better than rules in many games, including for card play in bridge, but I have doubt that simple simulations (or any "simple" set of rules, in fact) will be able to handle cases such as the one I gave (AJx xx KQxx KQxx after 1♠-2♣-2♦). Note that if I held KQx xx AJxx KQxx instead then the correct bid would probably be 3♦, not 2♠! The approach I imagine would be as follows. After the auction I gave, responder sees (i.e. a rule says) that by claiming captaincy (hopefully captaincy will be easy to define :-)) he will be able to either ask for keycards and the "trump" queen either in spades or in diamonds. Now, for each value of {number of keys, presence of the TQ}, responder generates a number of hands to place the final contract (by DD or SD simulations) and computes the average score, which is then weighted by the probability of opener holding each number of keys + TQ. Now hopefully the EV of asking for keys in spades will be higher than the EV of asking for keys in diamonds (as we will have strictly more information available), so we set spades as trumps. By the way a strong notion of captaincy is also required here to avoid overruling by opener for the final contract :-)

Here is another challenge for computer bidding. This is a hand I actually played in a side pairs in Philly (but let's assume we're playing imps for the sake of the question). I held AJx xx KQxx KQxx ans my partner opened 1♠. Playing "standard" 2/1, the auction continued 1♠-2♣-2♦. Now I (or a simulation) can already see that if there is a slam (is assuming partner has a heart control and enough keycards), the diamond slam will usually be better than the spade one, as I'll be able to pitch my hearts on the spades. So it seems normal to agree diamonds and enter a cuebidding/keycard auction using your favorite methods. But this is actually not optimal as I'd much rather agree spades for now in order to be able to ask for the ♠K and ♠Q, which may be critical to get to a grand, so our auction continued 1♠-2♣-2♦-2♠-2N-3♦... and now my partner will have not problem understanding that my 6♦ later in the auction, for example, is a signoff and not a grand slam try. (As it happened, partner had only two keys so I stopped in 5♠ making). Anyways, developing more and more complete system databases is not going to help your computer get to the right sequence on hands like this one... something "more" seems to be needed.

Not really... Does something such as Kungsgeten's BML (http://www.bridgebase.com/forums/topic/52574-bml-markup-language-for-full-disclosure/) suit you? (adding a specific syntax for specifying the bids) 1N; 15-17hcp bal 1N 2C; 8+hcp (4H or 4S) 2D; 5+H 2H; 5+S Say for example that the description of each bid is a list of items, which are "and"ed together, except that you can use parentheses and "or" as well. An item is either a single token ("bal", "unbal", etc.) or "quantity object" (quantity: number, number+, number-, number1-number2; object: hcp, controls, suit). There is a BML -> FD converter; I don't know if FD -> BML exists but that should be not too hard to write anyways and now you get to use bidedit.exe to input your system... Anyways, I'm going way too much in the details here. My whole point is that all that is only a minor step towards writing bidding sims.

<troll mode>Switch to a Unix system.</troll mode> More seriously, I don't think "without having to undergo the drudgery of learning a programming language" is a very interesting goal -- I can write a domain-specific language for specifying a bidding system in a couple of days if you want, and others have done that too (e.g. BML). Whether you specify you system in XML or in BML or in S-expressions or in C(!) hardly matters. The hard part is writing the bidding system itself. (define 1NT (and (>= hcp 15) (<= hcp 17) (balanced)))

Bluecalm: the first part of what you describe sounds like the approach Deal and Redeal use to generate rare hand types through "smartstacking".

I think there are a few interesting questions here. The first example can be seen as, given an auction that starts 1D-1S, and assuming that opener has 4 spades (let's forget about 3-crd raises :-)), with what hands should he bid 2S? 3S? 4S? We can make this question even simpler. Assume that opener opens 1N, and that responder has a balanced hand without a 4cM (or responder has any hand, but we only allow NT play -- allowing suit play only obscures the essential point). Essentially, there are 3 calls possible: pass, 2N and 3N (let's forget about slams too). An "invite strategy" is basically a partition of all possible hands for responder into 3 subsets, RESP={pass, 2N, 3N}, and a partition of all possible hands for opener into 2 subsets, OPEN={accept invite, reject invite}. How do we evaluate how good a strategy is? One possibility is to make the program play against itself, or against other programs. However this is not good because we would then try to exploit each program's peculiar weaknesses and bugs (say that my opponent, for some reason, misdefends whenever he doesn't have the C2; then I'll bid much more aggressively whenever I have it) -- this is pointed out, in the context of bidding, by Ginsberg in his paper linked above. An improvement would be to score the final contract against the par DD contract (again, only allowing NT contracts). An even better refinement would be to score against the par DD contract but with single dummy opening leads -- blasting to 3N compared to going through 2N is well-known to make it harder for the defense to find the right opening lead. Anyways, even with DD scoring, is it possible for simulations to give us a better hint at what the RESP and OPEN sets look like? Most of us use a description that resembles: RESP={pass: <8HCP, 2N: =8HCP, 3N: >8HCP} and OPEN={accept: <=17HCP, reject: >=17HCP} with some judgment in between... but better descriptions should exist.

Thomas Andrews' Deal (http://bridge.thomasoandrews.com/deal/) is quite good.

Then why is a 2♥ contract never reached according to the data you posted?

I'll jump on the opportunity to re-advertise my dealer program, Redeal (https://github.com/anntzer/redeal), which should be reasonably easy to use if 1/ you know Python, and 2/ you work on Linux/Mac or don't need the double-dummy solver. Mostly, my goal is to have an easy way to compare different bidding strategies (such as, holding ♠652♥K752♦53♣9862 opposite a partner who has shown a 22-24NT, should you pass, bid 3N or bid Stayman? and does it matter whether you play regular or Puppet?)

This makes it extremely easy for the opponents to balance into the auction (after all your point range is ~11-21 at that point, who would not balance?).

You hold K543 opposite dummy's J862 and your (imps) goal is to score one entry to dummy. In addition, you are only allowed to lose the lead once to lefty, whereas you are allowed to lose it twice to righty. Best line?

Standard in France... if you grow up with it you get confused later when the ACBL bridge bulletin has bidding contests where you hold AKxxx x xx AQxxx, the auction starts (1S)-2S-(P), and everybody tries to guess what is happening :-) (not that this wouldn't be a legitimate problem even with Ghestem in the picture of course).

1C-1D-1H would show a WNT with better hearts than spades (or equal) and 1S a WNT with better spades than hearts? That seems way too restrictive. In reality you can bid 1M with any hand where you don't mind your partner passing, and given that he'll only pass with 0-4 (with 5+ he can bid 1N or raise you or 1S over 1H), you can definitely bid 1M (intending to rebid 2M) with 17-19(20) and 5M.

At least in a Polish club context, 2M should be 21-23 (approx) and (17)18-20 hands go into 1M (together with WNTs and 15-17 5♣4M). Over 1M it's easy: weak hands pass/bid 1♠/1N, 2m=5-11, jump to 3oM=12-16 bal, asking opener to bid 3N. If opener bids 2M then weak hands should probably go to game directly (or pass), strong hands can take it slowly (more or less anything should work).

I would just open 2N.

I thought a Precision 2♥ would be harder to defend than a Precision 2♦ because you cannot expect responder to keep the auction alive for you.

I'd interested in an explanation of why it is not an asset.

Smartstacking is in, as well as the ability to do most scripting directly from the command line (if you like that and are not scared by long lambda forms). A quick comparison on the same "Gambling 3NT" example as the one given by Deal gives similar statistics :-)

<advertisement mode on>If you know Python, you can try https://github.com/anntzer/redeal (or Thomas Andrews' Deal, if you prefer Tcl).<advertisement mode off> Otherwise BBO's hand generator will give you the frequencies as well. As an aside, in my Polish club partnership, 4414 hands are, as a rule, opened 1♣ (1N possible if in NT range and singleton A/K, as usual); they are rare enough that it isn't really a problem.

Screens are used in France in all national finals, even of lower flight events (say the equivalent of the GNTC). After all, without them I would probably never have seen a French top pair give me a top on a gross misunderstanding in the national junior pairs :-) (they were apparently experimenting a WNT-based system, which is very uncommon in France.)