tysen2k

No need to get defensive. I'm just trying to help everyone out and make sure we all use good practices when evaluating anything. I can't help it, it's part of my job. :)

My research for hand evaluation has come to similar conclusions. However, I found that if partner shows a 5+ suit and you have 0-2 support, a better evaluation method is to switch to length points: 1 point for each card over 4. And yes, you can count for either length or shortness (or both) from the beginning, but the 5-3-1 method is surprisingly simple and accurate. Each hand pattern has its unique value, but 5-3-1 comes closer than any other method that doesn't involve fractions. Tysen

Problem is if you don't look at every single hand (even the uninteresting ones) you don't capture the whole problem. Zar's main weakness is not in the missing of games or slams. It's weakness is when only a boring partscore can be made but Zar tells you to bid game. By only looking at the hands where game can be made, you are ignoring this aspect. Tysen

I was just running a comparison too. On #3 I'd count 4♥ as a minus for Zar. I wouldn't want to bid 4♥ looking just at the EW cards. On #23 game is close under my method of calculation. A total of 25.5 points. Worth it to stretch to game. Evolved Binky predicts 9.82 tricks.

Your calculations are right. The average Zar distribution is 11.77. However, two average hands together take on average 8.43 tricks in their best contract. "Two jacks and a bit" probably will increase this by about half a trick, bringing you close to 9 tricks, but not 10. Tysen

I currently use the alternate way and I find it superior. It's more than just encouraging/discouraging though. If you and your partner have some good and solid agreements about forcing passes, you can handle a lot more situations. It typically goes like this: In a forcing pass situation: bidding = one dimensional hand that wants to play double = mostly take out, suggesting 1-2 trumps. You expect partner to remove the double, but are prepared for him to leave it in. pass = sort of transfer to double. Partner should double unless he would remove a penalty double. You can have either a penalty hand or a hand with 2 places to play (show this by removing partner's double next round). Flame's original example probably couldn't have 2 places to play, but it comes up in precision auctions often: 1♣ - (1♥)- 2♦* - (3♥) P - (P) - Dbl - (P) 3♠ * Natural, GF Opener has spades and another suit (maybe diamonds)

They all should show both minors, each one showing more extreme shape than the last. Dbl = 44, 1N = 54, 2N = 55 seems reasonable.

I ran the same test with the hands you suggested. It suggests these values: Length [space]Points 0 [space] [space] [space] [space]2.1 1 [space] [space] [space] [space]1.3 2 [space] [space] [space] [space]0.6 3 [space] [space] [space] [space]0.0 4 [space] [space] [space] -0.2 5 [space] [space] [space] -0.1 6 [space] [space] [space] [space]0.1 7 [space] [space] [space] [space]0.7 8 [space] [space] [space] [space]1.6 9 [space] [space] [space] [space]1.7 10 [space] [space] [space] 1.5 0.72*Longest + 0.24*Next – 0.25*Shortest (Zar) 1.19*Longest + 0.40*Next – 0.41*Shortest But I think these values are worthless. Picking all hands that give a NT contract within 20 points of the best suit contract seems sort of arbitrary. Sure, you get more balanced hands than a random sample, but there are still plenty of unbalanced hands in the mix that do okay in NT. For example you might get these hands that make both 3♥ and 2NT: Q9432 987 94 K98 K AKJT654 QJT J2 Is this what I want to be counting? Do I care how many NT tricks I can take? Not unless I’m looking at slam, and in that case I’m not going to be using “points” to tell me if I should be there. There is actually a lot of noise in the numbers above and that furthers the notion that they shouldn’t really be counted. I’ve looked at other evaluations when NT is the best contract. In those cases distribution matters very little. Simply adding 1 point for having a 5+ suit is about the best you’re going to do. Tysen

I agree with the above posts. This isn't an always trap pass situation. It's just too rare of a situation that you'd have the hand that wants to defend 1♠x for you to use your cheapest bid to show this. Plus you're going to have trouble defining acceptable ranges for your balanced hands if you can't pass without a trump stack. Pass should be a minimum NT (15-17 or 16-18), bid 1NT with more than a minimum (18-20 or 19-21).

I’ve been doing a little research into what the “correct” values are for initial distribution. Interestingly, if you want to assign values to various suit lengths, the optimal values turn out to be: Length Points 0 4.8 1 2.5 2 0.9 3 0.0 4 -0.2 5 -0.1 6 -0.1 7 0.0 8 0.0 9 0.2 10 0.3 If you wanted to do it similar to the Zar or Bergen way of assigning coefficients to the relative lengths, these turn out to be the best values to use: 0.79*Longest + 0.62*Next – 0.87*Shortest These are relative to “normal” points so the “ideal Zar” coefficients would have to be multiplied by 5/3 or be: 1.31*Longest + 1.03*Next – 1.46*Shortest Tysen

I prefer the methods suggested in Rigal's "Precision in the 90s": 1C-1M: 2M = trump ask 3M = 19+ singleton somewhere 4M = dead minimum

First impression: I love light openings Those 2M openings are horrible and the 2m openings are almost as bad In response to strong 1♦, it's better to show suits first, not controls I'm just not sure what advantages this system would have over something like MOSCITO.

Okay, let me first post some of my original data for comparison: All Hands ERROR SCORE HCP 1.23 -0.49 HCP+321 1.07 0.00 HCP+531 1.05 0.07 Zar 1.05 0.08 BUMRAP+321 1.03 0.14 BUMRAP+531 1.02 0.21 Binky 0.99 0.32 This compares different methods of evaluation looking at ~2.8 million hands. ERROR is the average number of tricks that you are off between your evaluator's prediction and actual tricks. This error can never reach zero even if you knew your partner's hand perfectly, since you still don't know the location of the opponents' cards. SCORE is an estimation of the number of IMP's you would gain (or lose) per hand using this evaluation against a team using an unimaginative HCP+321 Now here are the results looking only at the hands where the best contract is not NT and where we can take 9+ tricks: No NT, 9+ tricks ERROR SCORE HCP 0.87 -0.31 HCP+321 0.79 0.00 HCP+531 0.76 0.10 Zar 0.76 0.10 BUMRAP+321 0.77 0.08 BUMRAP+531 0.74 0.19 Binky 0.71 0.30 No real surprises here. Everything becomes more accurate since we are only dealing with the subset of stronger hands. Zar gets better in comparison, but not much. Using raw HCP+531 gives practically equal results to Zar and BUMRAP+531 still does better. Binky (although not usable at the table) is still the best but it slips a tiny bit since it's optimized over all hands and not just the good ones. Tysen

1) Yes that's right and I agree with your reasoning as well. However, all the evaluators have to deal with NT hands at the frequency they come up. My current "best and simple" evaluator counts 5/3/1 points for distribution always. It has to cope with the same problems with NT and does so better. You can't just pretend the NT hands don't exist b/c they do exist. One of the know problems with Zar is that it goes overboard with a misfit. This happens. 2) Yes and No. When I reported the average error in tricks, that was based on a random sample of all hands. However, when I reported the average IMP's won/lost based on the evaluator, that was based on the side that "won the bidding" which is usually the side that can take more tricks, which is usually 9+. I can try this all again eliminating the NT hands and only looking at 9+ tricks hands as well. That shouldn't be too hard. Tysen

I might be wrong, but I think that any 1♣ or 1♦ bid does not need an approved defense as long as it has 10+ HCP since these are bids that are allowed in the GCC. Only bids that are allowed by the mid-chart but not allowed under GCC need the approved defenses. So you'd run into trouble with 1♥ showing spades, but 1♦ showing hearts or spades should be okay.

A direct bid of 4♥ is more than "I think we can take 10 tricks." You might have a hand where you think you can only take 8 or 9, but it's a pre-sacrifice so that it's tougher for the opps to find their fit. But an accurate evaluation method will predict how many tricks you are going to make. For example, if partner opens 1H, you expect to take about the same number of tricks with both of these hands: x Qxxxx xxxxx xx vs. Kxx xxx Kxx KJxx Zar is pretty close on this one, making them about the same number of points (20 vs. 21 if my math is correct) but I'm going to bid 4H with the first and 3H with the second even though Zar tells me 2H might be the limit with the first. Tysen

Let's not forget that the 5422 case has two doubletons. That might account for some of the closeness between the doubleton/singleton cases. Tysen

Let's see what the data shows: When partner has 5+ spades 5422 is worth 0.11 tricks more than 4522 5431 is worth 0.15 more than 4531 5440 is worth 0.34 more than 4540 So yes, the difference is more pronounced between the singleton and void cases when you have 5 support, but the overall value of the 10th trump is much less than the 9th. This seems low, but it's real data based on millions of hands. Now remember that these are tricks or points that we're adding to partner's valuation. Partner has shape that he's counting on too and so we can't add too many points or we'll start double-counting too much. I also looked into the case where you have a 5440 hand and a choice between a 5-3 and a 4-4 fit. The 4-4 fit on average is better by 0.25 tricks. Tysen

I was wondering this myself and decided to look into it. When opener has 5+ spades, 4333 is worth 0.26 tricks more than 3433. In comparison: 4432 is worth 0.37 more than 3442 4531 is worth 0.58 more than 3541 4630 is worth 0.63 more than 3640 I'm very surprised that there is so little difference between the singleton and the void case. So if you're going to use Zar, it looks like the 9th trump is worth 1 point with 4333, 2 with a doubleton, and 3 with a singleton or a void. Tysen

I second this. Actually, our 3rd seat 1N gets extended to 8-14. We assume we never have a game since partner can't have many points and is not extreme shape. All 2-bids by partner are sign-offs, even 2♣. Absolutely a great preempt when combined with a weak-opening system.

I've used this a few times at teams with no real disasters.

I disagree. The 2♣ opening is one of the weak points of precision and 5+,4M is much worse. Promising 6+ helps a great deal. But I don't like the nebulous diamond either. This is what I currently use for GCC matchpoint events. It's a modified version of matchpoints precision: 1♣ = 16+ balanced, (14)15+ unbalanced 1♦ = 10-15, 0+ but promises at least one 4cM 1♥/♠ = 11-15 balanced, 8-14 unbalanced, 5+ 1NT = 11-15, balanced, no 4cM 2♣/2♦ = 8-14, 6+, no 4cM 2♥/2♠ = weak 2NT = 8-14, (54)+ in the minors, no 4cM The 2NT opening is agressive, but we love it. This is great for matchpoints as you get weak openings and your 1N & 2m openings never have a 4cM so you don't miss 4-4 fits. The 1♦ opening is great and you can use a lot of negative inferences. 1♦-1♥-2♣ shows 5+ clubs and 4 spades. 1♦-1♠-1NT shows balanced with 4 hearts, etc. Tysen

Yes this is a FP situation for the reasons inquiry said. Partner bid 5♥ to make. I like the style that some pairs have adopted of switching the traditional meanings of pass and double in FP situations. So I'll double which suggests 1 (maybe 2) trumps and not extreme shape. Partner has a better idea of the total trumps than I do. He's smart and knows the vulnerability too. I'm not expecting him to leave the double in, but if he does, I'm betting it will be right. Tysen

And see my articles on revaluation (although long and detailed). I think I mentioned them earlier in this thread: [improving Hand Evaluation Part 1] http://tinyurl.com/25huc [improving Hand Evaluation Part 2] http://tinyurl.com/383e6 Zar, in reference to how many hands I've tested it on it's over 2.8 million. Let me just show some hard data so we're not talking about "IMHO." Tricks Tricks Tricks Error Error Real ZAR 5/3/1 ZAR 5/3/1 4-3-3-3 0.000 0.000 0.000 0.000 0.000 4-4-3-2 0.296 0.400 0.333 0.235 0.030 5-3-3-2 0.339 0.600 0.333 1.061 0.000 5-4-2-2 0.595 0.800 0.667 0.447 0.055 6-3-2-2 0.660 1.000 0.667 0.655 0.000 4-4-4-1 0.810 0.600 1.000 0.133 0.109 5-4-3-1 0.864 1.000 1.000 0.241 0.241 6-3-3-1 0.918 1.200 1.000 0.276 0.023 7-2-2-2 0.999 1.200 1.000 0.021 0.000 6-4-2-1 1.154 1.400 1.333 0.286 0.152 5-5-2-1 1.183 1.200 1.333 0.001 0.073 7-3-2-1 1.208 1.600 1.333 0.290 0.030 5-4-4-0 1.519 1.200 1.667 0.127 0.027 6-4-3-0 1.624 1.600 1.667 0.001 0.002 5-5-3-0 1.643 1.400 1.667 0.054 0.001 7-3-3-0 1.697 1.800 1.667 0.003 0.000 6-5-1-1 1.703 1.600 2.000 0.008 0.063 7-4-1-1 1.712 1.800 2.000 0.003 0.032 7-4-2-0 1.923 2.000 2.000 0.002 0.002 6-5-2-0 1.964 1.800 2.000 0.018 0.001 Totals 3.862 0.842 We're looking at how many more tricks you take with certain hand patterns over a 4333 hand. The first column shows what the actual average difference is over the 2.8 million hands. The second column is what ZAR predicts the difference will be. Third column is what 5/3/1 predicts. Fourth and fifth are the squared errors weighted by frequency. The totals on the bottom show that ZAR has over 4.5 times as much error as 5/3/1. Note that I'm using 3 points per trick for 5/3/1 even though it should be slightly less since 26/29/33/37 aren't perfectly 3 point steps. If I change it to a little more than 3, then the 5/3/1 count gets even better. Tysen

As I've said in previous posts and Mikestar has echoed, the evaluation scheme A = 4.5 K = 3 Q = 1.5 J = 0.75 T = 0.25 Plus distribution of 5/3/1 for void/single/double is more accurate than ZAR. This ratio of values for high cards (4.5/3/1.5/0.75 or 6/4/2/1) is not new. The Dallas Aces were using 3/2/1/0.5 in the 70's. But the ZAR distribution scheme is simply less accruate than 5/3/1. Yes using 4.5 for aces involves fractions, but it allows me to use my existing bidding methods. I don't have to worry if my strong club requires 30 or 31 ZAR or whatever. To me, if I have to change my bidding methods, it just isn't worth it. I have done extensive studies on many evaluation systems (see my previous posts). Zar, while I admire your research I do question a lot of your methods for determining how accurate a system is. One of your reasons that ZAR is more accurate is that ZAR has more "separation power" and a higher standard deviation than standard (meaning that there are more possible values and more spread out values). This does not give accuracy. I can create an evaluation scheme that doesn't give any values to high cards, counts the ♣2 as 1 point, the ♦2 as 2 points, ♥2 as 4, ♠2 as 8, ♣3 as 16, etc. I could make this scheme have 100 times more separation power than ZAR. Should I claim I've found a new system that's 100 times more accurate than ZAR? You also seem to sometimes go backwards in your assigning of accuracy. You find hands that make game or slam and then see what % of Goren, ZAR, etc. would bid that high. That's going the wrong way. You should first take hands and see where the points say you should be, then assume you bid there and see how well you do. Maybe the points say game, but the hand doesn't make it. Look and see how far you were off and how much that would have cost you. You're finding hands that already make game and then seeing if you would have bid it. Zar, don't get me wrong. I admire people trying to break new ground in research like this. Just make sure you're measuring the right thing. Let me know if you want more information about my studies and findings. Tysen