mikestar

The level must be taken into account. 3 of of suit is a much better prospect than 2 of a major, and 4 of a minor is better yet (though rare) for extracting penalties. In To Bid Or Not To Bid Larry Cohen suggests "With four trumps consider making a penatly double." (Intending this advice for primarily the three level and above). Monocled One's analysis of the two situations is correct--you will leave in a takeout double at this level more often than you will make a penalty double. An extreme example: Neither side vul, the auction is (3S)-X-(P) to you and you have xxxx xxx xxx xxx. You place partner with 1-4-4-4 for total trick analysis. They have 8 trumps and we have 7, for fifteen total tricks. If they make, you are down 4. Leaving in is -530, taking out is -800. Even if they make an overtrick its -730 vs -1100. On the other hand, if partner has a strong double and they go down, the profit is obvious. This can go wrong if partner is seriously off shape (in which case he will be strong and they probably go down, but you might miss a game) but it looks like the percentage action.

Ben, Per Zar's web page you don't count both honors in partners suit and superfit, to avoid duplication.

Zar, Let me restate my question. Let's say that the auction for our side is 1S-3S (limit) and futher suppose that the Zar point range for 3S is 21-25 Zar points as shown on your website. After revaluation, if opener has 31 or more Zars, he bids game--the combined total is at least 52. But let's say he only has 30 Zars--should he bid and risk a 51-point game or pass and risk missing a 55-point game? With 29 should he bid and risk a 50-point game or pass and risk missing a 54-point game? What is your advice on this point? Of course I could try it both ways, but coverting my old sytem notes is a bit of work.

I agree provided that our agreements are in order and partner will recognize it as voidwood and know what to do if they bid 5S. If this is not the case, I bid 6H.

Zar, I am currently converting some of my old system notes to use Zar points, and I need a clarification. Currently I use a 26 point game target. Ideally, we bid all 26 point games and stay out of all 25 point games, but this level of accuracy is not possible. I set my point ranges on the assumption that it is equally inaccurate to bid a 25-point game or to miss a 26-point game, it is equally inaccurate to bid a 24 point game or to miss a 27 point game, etc. So although the target is 26, it is ussally correct to bid game when certain of 25, because you may miss a 27 point game if you don't. Does similar reasoning apply to Zar's 52 points? Is is right to bid game when sure of 51 to avoid the possible miss of a 53 point game? To bid game when sure of 50 to avoid the possible miss of a 54 point game? Or alternatively is 52 a minimum standard and you should be willing to risk missing 53-54 point games when you can't be sure of 52 and don't have room enough to find out? My guess would be the former, but I'd rather ask the expert. A careful reading of the example point counts on your web site convince me that my guess was in error and the second alternative is correct

Zar, When I stated that Zar does best for grands and worst for part scores (I should have said "least good"), I was not intending to compare it to other methods which undoubtledly show the same characteristic. I apologize for not being clearer. My real intention was to call attention to the bias of double dummy solvers. In point of fact Zar's at the table accuracy for part scores and slams may be more nearly the same than your figures indicate, and this will be true in varying degrees of other methods. I accept that your figures show Zar superior to 1-3-5 at all levels, with increasing difference in accuracy as the level goes up. My understanding is that we can be fairly confident that Zar is more accurate, but not as certain how nuch more accurate--the double dummy bias is highest at the grand slam level and this will tend to magnify small differences in accuracy. I hope that this is a clearer statement of my points.

It is simpler and just as correct to calculate the mean and standard devaition for each method in terms of its own points and then compare relative error (Standard deviation divided by the mean). Finding the theoretical best method doen't entirely answer the question of which to use, complexity is a factor as well. Binky may be the best of all but it is unusable at the table. Zar is no more intrinically complex than the others, but it is unfamiliar. Teaching a beginner to count 1-3-5 instead of 1-2-3 isn't such a big step and will improve his accuracy. By the way, while the double dummy solver approach is the best available, it does have some biases. When the points are equally distributed, the times that the contract makes double dummy when but goes down at the table are compensated by the times the contract goes down double dummy but makes at the table--perhaps even overcompensated, as there are two defenders to make a mistake, wrong guess, etc. vs. one declarer. But as the defenders hands get weaker and weaker, double dummy defense gets less and less useful. Let's say we're in a grand missing the Queen of trumps and have to guess a two way finesse. Double dummy declarer play guaratees we make every time, while double dummy defense doesn't affect the result at all. Maybe in part Zar seems to perform best on grands and worst on partials because of this bias. I believe this bias will affect all methods, so that the ordering of the relative errors will be relaible, but the magnitude of the differences in relative errors will not be. And part of the question is "how much better?"

On hand #2, Ben is right that 5D here is a clearcut slam try over 4S (4N was available as ununsual), but what if partner bids 5C? Might he not have a spade diamond two suiter with game values? Pehaps the best way to bid this hand is to double, intending to bid 5D over 4S, but to bid 6D over 5C. I think a penalty pass is unlikely here as partner will assume they have at least 10 trumps, which means we are certain to have a nine-card fit or two eight-card fits.

1) Double. We've got to bring clubs into the picture as well as spades. If partner is stong in both black suits, there may be a slam and he may be able to bid it. Even if slam is not in the cards , if partner has S xx C Kxxx (perfectly possible on the bidding) I want to be in 5C, not 4S. We can't count on them doubling us in 4S if it is wrong. 2) Ben makes a fine case for double, but I'd take a shot at 6D. 3) low heart -- works better than KH if partner has AH, works about the same if he has JH. I know players who would bid 3NT here praying that H Jxx will be a stopper with help form partner or that hearts won't be lead. I'd bid it myself in a shot.

The Laws do not currently define the term "natural", but they do contain this definition: Convention 1. A call that, by partnership agreement, conveys a meaning other than willingness to play in the denomination named (or in the last denomination named), or high-card strength or length (three cards or more) there. However, an agreement as to overall strength does not make a call a convention. 2. Defender's play that serves to convey a meaning by agreement rather than inference. Two suiter methods are by definiton conventional, though something like 2H showing both majors is also arguably natural as it suggests playing in hearts--it is conventional because it also suggests playing in spades. The converse of "conventional" is "non-conventional", not "natural".

Another form of this count that I used for teaching beginners. The count of each pattern is identical, but it attributes points to more factors. Long Suit Points Count 1 point for each card over 4 in a suit. Short Suit Points Count 1 for a singleton and 2 for a void. (Nothing for a doubleton). Short suits count double in an enemy suit if you have 3+ trumps. Short suits don't count in partner's suit or with only 2 trumps. Short suits count negative with 0-1 trumps. Pattern Points Add 1 point for a two-suiter, add 2 points for a three suiter. Fit Points Add 1 point for each "extra" trump, subtract 1 point for each "missing" trump. Subtract 1 point if your have fewer than 4 trumps.

I will answer this in detail If it is your opinion that 4333 should count -1 and 4432 should count 0, you may be right. Zar is quite correct when he asserts that we are really assigning points to hand patterns, not individual lengths/shortnesses. For example (counting 4333 as 0), if we determine that 5332 is 1 point better than 4333, does it matter whether we attribute that 1 point for the pattern as a whole to the five card suit or the doubleton? Not entirely correct, especially with singletons and voids. Also, it is my experience that 5-4-3-1 plays better than 5-4-2-2, though both have identical long suits--the stiff opens up enhanced possibilties of crossruffs, dummy reversal, or just plain ruffing losers if partner has the long trumps. Neither does any other method including Misho's justly beloved losers and cover cards. Honor location is an adjustment you learn to make when you start aquiring judgement. In point count terms I would rate your first hand maybe 2-3 points stronger than your second, depending on the bidding. Of course--the whole purpose of point count or any other evalution method is to let you get to reasonable contracts a fair percentage of the time while your aree aquiring good bidding judgement.

I couldn't agree more--I don't actually use this anymore, but found it valuable 10-15 years ago when I certainly didn't have the judgement not to count points! My purpose for bringing it up was for comparison with Zar--I wanted to pulish the method so others who might want to use it for the same purpose could know how I got my numbers.

Ron, I think the way it was done was obscene. Since BBO is not sanctioned by the WBF or any NCBO, the Laws are binding only as far as BBO, or the individual club or TD determines. But any deviations from or amendments to the Laws should be clearly posted in the Conditons of Contest. If a TD insists on a no psychics tournament, the right way to do it is to: 1. State the rule in the Conditions of Contest. 2. Make psychics illegal--adjusted score if damage occurs, procedural penalty for repeat offenses, etc. The self-alert requirement is nonsensical as you say.

In case anyone is interested, here are the details of my 1-3-5 based evaluation method for suit contracts. The normal count for distribution is 1 per doubleton, 3 per singleton, and 5 per void. If the shortage is in partner's bid suit (if the suit is likely to be long or strong), reduce the count in that suit to 1-2-3. In an enemy suit (bid and raised, or the bidder promises length and strength), increase the count of that suit to 1-4-7! If you are supporting partner with a doubleton trump, all shortages count 1-2-3 (including the trump shortage), with a singleton or void in trumps, they count 1-1-1. The total short suit count is modifed for the degree of fit. Each "extra" trump adds 1 point and each "missing" trump subtracts 1 point. Also subtract 1 more point with three or fewer trumps. Add 1 point for a double fit and an additional point for each card over 8 in the second fit. Points are counted for shortness and fit for simplicity--not all 5 of the points for a void come from the value of the void itself--they also reflect the long suits you have in your hand. The total HCP+distribution translates to tricks at 3 points per trick, plus the assumption that declarer with a good trump fit wins an extra trick. So 26 points = 9.667 tricks (26/3 = 8.667 then add 1) = major suit game. Similarly, the theoretical magic numbers are 29 for minor suit game, 32 for small slam, and 35 for grand slam.

This hand is (for my methods and probably for Zar's) comparable to the slam exceptions--the point count says bid it but you'll stay out at the table, if you have good methods. I rate South as a good game try and North as a borderline acceptance. After 1S-2S South will bid 3C if playing short-suit game tries or 3H if playing help-suit game tries, and in either case North stops at 3S.

Two average hands plus 2 Jacks and a bit = 9 tricks in a suit or 8 in NT. This is in accord with my experience playing Precision, opening 5-3-3-2 11-counts. In our partnership we amended Culbertson's mantra "an opening bid opposite an opening bid is a game" to "an opening bid opposite an opening bid is a game invitation."

I wouldn't charge board 15 against ZAR, at this level partnerships will have methods that can determine you are off two Aces even with a void. For comparision, I evaluted these hands with my own 1-3-5 based method and it agreed with Zar on all hands except #3, where it indicates 3H for E-W.

Actually, I rather like the idea of opening 1H then bidding diamonds until Hell freezes over or we buy the contract, which ever occurs first.

This is not a good position for a trap pass. Strengthen your example hand by an Ace and you still may have trouble setting 1S if partner is broke. A trap pass at the one level really must have five good trumps at least. A better use is for pass to show weakness (relative to having opend a big club). Typically, this would be a balanced or semibalanced hand without a stopper, not shaped right for a takeout double. You can still trap with excellent spades--if partner reopens with a double (showing some values) you leave it in and bid spades otherwise. If partner leaves it in with a weak hand you may or may not set them with the strong spades, but defending 1S undoubled should be OK, as it is their hand and they have a better fit elsewhere most of the time. All of the above presupposes that their 1S is natural here--many partnerships use variations of their defense directly over the big club in this position.

I agree with fly, the 5 diamonds are too much of a surprise to defend. Better would have been 2NT unusal for the red suits on the previous round, instead of 4H. This is constructive rather than preemptive here (we don't preempt over preempts) and emphasizes shape, while a double would emphasize HCP and might be only 4-4 in the reds.

The illegal 5 and 5 rule has been replaced by a legal [not defending it as a good thing] rule outlawing conventional responses and rebids to a weak two bid that might have less than five cards in the suit or a range of greater than 7 HCP (inclusive--say 4-10). You can still legally open 2S on Axxx but your side is allowed no conventions in the subsequent bidding. No such restrictions apply to three level preempts, nor ever have. In the ACBL, a ultra light prempt is not alerted, but may require a prealert. The standard for three level prempts according to ACBL's site is suits worse than Qxxxxx. So even in the overly restictive ACBL, this hand doesn't calkl for an alert, and no prealert is needed if you wouldn't have opened it with Jxxxxx or QJxxx.

Someone thrown out for playing something ACBL Limited Convention Chart legal? Good grief!

Free, Your point is well taken if you are considering what bidding system is best for computers--but for play by humans, given two systems of approxiamtely equal technical merit, the more easily remembered one is clearly superior--that is, your partnership will make fewer mistakes with it and win more often. And isn't winning the primary reason for the quest for the best system? As an example comparing two systems of the same type, Meckwell Precision is clearly superior to Goren-Wei precision if you are a computer or a Bridge god like Meckstroth or Rodwell, but two average players on BBO or in a club will have a much higher win percentage with Goren-Wei: Meckwell will confuse them so badly that it will cause them to make hundreds of mistakes. In math terms, If System A is only 50% as good as system B on pure technical merit but a given partnership makes 10 times as many mistakes playing system B, which system should it play?

This is also ACBL General Convention Chart legal, where the only restriction on 1NT defenses is thall all direct calls other than double and 2C must have a known suit.