Tune my judgment

[WellSpyder]

Oh wow, I'm terrible. I actually downgraded the first hand but not the second one...

I think it's interesting that views differ not just on whether to downgrade, but also on which hand is closer to a downgrade. Despite the majority view, I'm probably closer to your view that the first one is closer to a downgrade than the second. The second just looks like an entirely normal 1N to me. The first is trickier, since it is largely aces and spaces, with limited trick-taking potential in NT, but of course you may not end up in NT just because that is what you open. Personally, I see no reason not to open 1N, but my regular partner might well see it your way - last week he passed a very similar hand with an ace less, rather than open the equivalent of an 11-13 NT, whereas I suggested to him that two aces and a king was about as good an 11 count as he could hope for...

[han]

Try 10-11, 12s probably have enough raw grunt to make a very high percentage of the time, and you may well still be in game if you downgrade.

 

I tried it, same results. Maybe it is time to accept what everybody has been saying all along, aces are good, queens and jacks are bad.

[Antrax]

It wasn't so obvious to me, so thanks, han.

[han]

I don't think that this can be obvious. Of course everybody can decude that an ace is better than a queen, but you can't easily deduce that an ace is worth a little more than 4 and a jack or queen is worth a bit less than 1 or 2. However, it is very well known, and it is often mentioned on these forums. Of course, sometimes conventional wisdom is wrong, but more often it is not.

 

If you want to read more (even a lot more) on this topic I recommend the website of Thomas Andrews, who spent a lot of energy on this topic. It won't make you a better bridge player though. If you just remember to appreciate your aces and not to disregard your tens, then you'll do better than most.

[gwnn]

Han, it sounds to me like your results confirm that 4321 is a pretty good estimate for NT purposes. The fact that people have been trying to meddle with Work point count in both directions (giving slightly less/slightly more to queens and jacks, the opposite for aces and maybe kings) would also point in this direction.

[Cyberyeti]

I tried it, same results. Maybe it is time to accept what everybody has been saying all along, aces are good, queens and jacks are bad.

Well can you please publish the results (they will not have been identical, the numbers will have both reduced), I wanted to use them for something else, ie that opposite 10-11 (the situation that you'd face in 3rd seat as 12s would have been opened, as in fact would some of the 11s, some of which will invite if you downgrade, so this will overestimate the true trick yield) you don't want to be in 3N in either case.

 

I'd be astounded if 3N was with the odds opposite balanced 10s on either hand, the difficulty is identifying 11s that wouldn't open and wouldn't invite opposite a downgrade to get a true picture, but my natural inclination is moving to suggest that you downgrade both in 3rd.

[Antrax]

I don't think that this can be obvious. Of course everybody can decude that an ace is better than a queen, but you can't easily deduce that an ace is worth a little more than 4 and a jack or queen is worth a bit less than 1 or 2. However, it is very well known, and it is often mentioned on these forums. Of course, sometimes conventional wisdom is wrong, but more often it is not.

Indeed. But other pieces of conventional wisdom dictate that a 4-3-3-3 pattern is bad, and that aces are better for suit contracts. The issue is to understand and be able to balance the contradictory notions, I believe.

[gwnn]

Indeed. But other pieces of conventional wisdom dictate that a 4-3-3-3 pattern is bad, and that aces are better for suit contracts. The issue is to understand and be able to balance the contradictory notions, I believe.

I think 6421 (plus something for shortness) for suit contracts and 4321 for NT (plus a little something for a 5- or 6-card suit and a bit for tens) is fine and probably as far as you want to go into numerical hand evaluation (note that most experts don't even bother to use 6421, it is more subconscious). The important point to make is that aces are not overvalued in 4321 for notrump, it is just that they are less undervalued than for suit contracts.

[han]

Indeed. But other pieces of conventional wisdom dictate that a 4-3-3-3 pattern is bad, and that aces are better for suit contracts. The issue is to understand and be able to balance the contradictory notions, I believe.

 

I don't see any contradictory notions. Aces are good. They are good for notrump, and they are even better for shapely hands. That's no contradiction.

 

4333 is bad, although not nearly as bad as the religious nutcases who subtract a full point make it sound. Maybe subtracting one fifth of a point is more accurate. Note that we were comparing two different 4333 hands, so the 4333ness isn't an issues since they both suffer from it.

 

As gwnn said, the 4321 count really is quite good, certainly for balanced hands. Don't believe those experts that tell you they don't count points. They do. For unbalanced hands kings and especially aces are worth more, particularly when you have a good fit.

 

For slam purposes you want to stop counting points at some point during the auction and try to visualise possible hands for partner to see how likely possible slams will play. You'll sometimes want to do that for game hands as well, depending on the kind of hand you hold and what you know about partner's hand.

[han]

Well can you please publish the results (they will not have been identical, the numbers will have both reduced).

 

1765 vs 1720.

 

If you find those numbers interesting I suggest downloading Thomas Andrews' Deal. It's free and it is good.

[nigel_k]

Han, it sounds to me like your results confirm that 4321 is a pretty good estimate for NT purposes. The fact that people have been trying to meddle with Work point count in both directions (giving slightly less/slightly more to queens and jacks, the opposite for aces and maybe kings) would also point in this direction.

4321 is fine when you play in NT and both hands are balanced. But neither of those can be known when you are considering the opening bid. For example, it makes sense to use 4321 when raising partner's 1NT opening and use ZAR after an opening one of a major has been raised. But people seem to rely on these things too much on opening bid questions where to me it makes more sense to do something in between until you find out more.

[Zelandakh]

I think 6421 (plus something for shortness) for suit contracts

It is worth pointing out that 6/4/2/1 is functionally identical to 3/2/1/0.5 and 4.5/3/1.5/0.75 as well as to MLTC with -0.25 losers adjusted for a jack. The second of these (Ace = 4.5, etc) is closest to Milton and probably the best option for intermediates looking to add adjustements to their normal point count methods. I have posted before that I think it is better still to decide whether/how much to deduct for quacks by the honour combination that they appear in (eg xx/KJTxx is better than KT/Jxxxx).

[Mbodell]

I tried it, same results. Maybe it is time to accept what everybody has been saying all along, aces are good, queens and jacks are bad.

 

I ran two sets of 10,000 board experiments for 10-12 and 10-11 range. The first there were no restrictions on W shape. For 10-12 I got:

 

Hand A makes 3nt 73.8% of the time and averages 9.03 tricks.

 

Hand B makes 3nt 63.1% of the time and averages 8.79 tricks.

 

For a 10-11 range I got:

 

Hand A makes 3nt 67.3% of the time and averages 8.85 tricks.

 

Hand B makes 3nt 57.5% of the time and averages 8.65 tricks.

 

The second time through I restricted W to only those hands that would pass a 1nt opening. This made only very negligible difference. For 10-12 I got:

 

Hand A makes 3nt 73.7% of the time and averages 9.04 tricks.

 

Hand B makes 3nt 62.2% of the time and averages 8.77 tricks.

 

For a 10-11 range I got:

 

Hand A makes 3nt 66.9% of the time and averages 8.85 tricks.

 

Hand B makes 3nt 57.5% of the time and averages 8.64 tricks.

 

So this is consistent with Han's 200 hand test and conclusion. I guess the only caveat is that these are all double dummy results, and double dummy always guesses locations of finesses, so maybe undervalues middle honors a little? Probably not close to enough to make up for the 10% make difference or the .2 to .25 trick difference.

 

Edited: The first time I ran it I promoted the ♠Q to the A in the B hand, then B was substantially stronger. :)

Edited May 7, 2012 by Mbodell

[sfi]

4333 is bad, although not nearly as bad as the religious nutcases who subtract a full point make it sound. Maybe subtracting one fifth of a point is more accurate. Note that we were comparing two different 4333 hands, so the 4333ness isn't an issues since they both suffer from it.

 

I recently read somewhere that the 4333 pattern is actually a small plus for NT, according to simulations. The reason given is that it doesn't have the structural weakness of a doubleton, which means it's less likely that the opponents can run a suit.

 

Of course I can't remember the source, but it was recent. Hopefully someone else has seen it and knows where it's from.

[Cyberyeti]

So this is consistent with Han's 200 hand test and conclusion. I guess the only caveat is that these are all double dummy results, and double dummy always guesses locations of finesses, so maybe undervalues middle honors a little? Probably not close to enough to make up for the 10% make difference or the .2 to .25 trick difference.

The interesting thing to come out of this as far as I was concerned was that the averages, particularly for 10-11 were significantly below 9 tricks, but the % making was well above 50%.

[han]

Cyberyety, that is to be expected when you think about it. Say for simplicity that the number of tricks is always 8 or 9 tricks. Then you will make 3NT half the time when the average number of tricks is 8.5.

 

Of course that is a simplification, but it gives you a good idea for discrete datasets like this.

[han]

I recently read somewhere that the 4333 pattern is actually a small plus for NT, according to simulations. The reason given is that it doesn't have the structural weakness of a doubleton, which means it's less likely that the opponents can run a suit.

 

Of course I can't remember the source, but it was recent. Hopefully someone else has seen it and knows where it's from.

 

I read this on the site of Thomas Andrews, apparently there was a rpg discussion about it. From what I remember Thomas Andrews discarded it as a double dummy anomaly.

[sfi]

I read this on the site of Thomas Andrews, apparently there was a rpg discussion about it. From what I remember Thomas Andrews discarded it as a double dummy anomaly.

 

Thanks for the reference. It certainly seems that any advantage was theoretical rather than of practical use.

[Cyberyeti]

Cyberyety, that is to be expected when you think about it. Say for simplicity that the number of tricks is always 8 or 9 tricks. Then you will make 3NT half the time when the average number of tricks is 8.5.

 

Of course that is a simplification, but it gives you a good idea for discrete datasets like this.

This is true, hadn't thought about it in that way. It means you can get some bizarre results potentially (on hands where opps could cash a lot of tricks, say a 3N with a stop of x opposite Kx) where the average number of tricks is well below 8, but game is still 50:50.

 

This says to me that average number of tricks is not what you really want to look at in judging contracts.

[Antrax]

Thanks again, I have to say that the simulation results are both surprising and enlightening.

[daveharty]

I finally downgraded a hand out of our 1NT range: ♠KJ8 ♥K97 ♦Q864 ♣KQJ. First time ever I think.

[diana_eva]

I finally downgraded a hand out of our 1NT range: ♠KJ8 ♥K97 ♦Q864 ♣KQJ. First time ever I think.

 

Did this have a happy ending?

[daveharty]

Did this have a happy ending?

Sort of, although I can't claim my hand evaluation was the reason; we had an accident and just got lucky:

 

[hv=pc=n&w=saq64h8dak3ct8542&e=skj8hk97dq864ckqj&d=n&v=n&b=5&a=p1dp1sp1np2dppp]266|200[/hv]

 

My partner thought we were playing XYZ. +130 was a pretty good score because 3NT fails on a (relatively) normal heart lead.

[Statto]

I finally downgraded a hand out of our 1NT range: ♠KJ8 ♥K97 ♦Q864 ♣KQJ. First time ever I think.

The ♣KQJ tight are def a negative, but ♠8 with KJ, ♥97 with K, ♦8 with Qxx are reasons to think twice.

 

Interesting thread and thanks Han for the research; there isn't that much between the 2 hands in the OP (½ HCP it seems).

[SteveMoe]

In spite of the nice intermediates, the 2nd hand counts to 8 losers and has no source of tricks if partner wants to play NT. The first counts to 6.5 losers, adjusting for the excess of top honors over Quacks. I would always bid the first as full value. 7 controls (21 control equivalent HCP) suggeest this hand is stronger than a mere 15. (Deck has 40 HCP. Max Hand has 37 HCP. Deck has 12 Control Points [A=2, K=1]. Ratio HCP to Cntl Points = ~3 to 3.3)

 

Bergen offers a good HCP adjustment: A: 4.5 K: 3, Q 1.5 J 0.75 10: 0.25