4333 Hands

[mtvesuvius]

What are the statistics for 4333 hands, and how much should I downgrade them over 1NT (13-15)?

 

Axxx AQx Qxx xxx

AJT Qxx Axx xxxx

xxx QJx ATx KTxx

xxx Kxx AJxx Kxxx

 

How would you treat these hands?

 

My partner thinks that even at IMPs, we should be downgrading these hands a good 2 points. I disagree, since I think we should be in any games that have a prayer. Something such as Kxx Jxx Kxxx AJx he suggests we should invite. What are your thoughts on this, and how much should we be downgrading here? Or has he just gone mad?

[awm]

I'd probably downgrade them by closer to one point.

 

Opposite 13-15 notrump (and assuming you don't upgrade very many 12s):

 

(1) I'd game force on any 12-count, even 4333.

(2) There exist 11-counts where I would game force, but not on a 4333 11-count.

(3) I'd invite on many 10-counts, but would pass 4333 tens unless holding amazing spots.

[kenrexford]

IMO, "cloud 3NT" contracts make a lot more than one would think, if you Declare really well and if you have body.

 

A "cloud" contract is one where you have general junk opposite general junk kind of spread out everywhere. These are a mess to declare, and a mess to defend. So, if you declare better than they defend, you make. A "cloud" with 13 opposite 11-12 seems nice, if the body is right.

 

"Body" means more than 10's. I would look as deep as 6's. A suit like A1086 has a lot of cloud potential, as a simple J32 from partner makes that 6 interesting.

 

How much body is a judgment call, I suppose. A feel or something. Hard to quantify. But, when I see your proposed hands, all of them mention only 10's and higher. Make the "x" cards 2-3-4, and I'm not liking the hands. Make then 986, 97, 87, stuff like that, and I like it.

[jjbrr]

bid 3N with all 11s. invite with many (most) 10s. dont worry about shape.

 

taking tricks is where the money's at.

[jdonn]

With this shape I would bid game always with 12, and usually with 11 but would invite with a bad 11 or a good 10. My invite range is very small.

 

On your sample I bid 3NT 3NT 2NT (good because of the tens and the jack being with a queen I would consider this an invite) and 3NT.

[kfay]

2 points? wtf?

[gnasher]

I'd invite on all four of your hands. I think I deduct most of a point for being 4333, That ought to rule out the third one, but the high-card structure makes it look nice enough to invite.

[Jlall]

Downgrading 2 points is ludicrous.

[mtvesuvius]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

[kayin801]

Agree with jdonn on all bidding choices.

 

You can subtract 2 points for the shape just so long as you add two points for the fact that you should expect to make (as you calculated) on 58-60% of the hands on perfect defense and perfect play.

 

Clearly if your partner is declaring you can't give them this upgrade :( (no, I have no idea who your partner is)

[Cascade]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

In the first simulation there will be some hands that clearly do not belong in 3NT.

[jdonn]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

You should run the simulation opposite hands on which partner would reject an invite to see how a direct 3NT compares to inviting. Of course it's not as simple as that (for example it could be right to bid 3NT directly even if it makes less than half the time opposite those hands, for reasons we have discussed before) but it's another question worth asking.

[xcurt]

You're right that 4333 is not a bad NT shape. This has been proven in earlier DD simulations.

 

Let's assume an invitational hand. What exactly is the range of an invitational hand isn't important, adjust it up or down a little depending on style and range for 1NT. Where I think you need to start adjusting is if you have suitish honor holdings and 4333. If you have suitish honor holdings, and, say, 4324 shape, you have two ways to win by inviting -- either you find a suit fit and now your hand is better, or you don't find a suit fit and partner rates to have some length in your shorter suits. If you can only invite in one strain, you don't get the vigorish that your hand might improve after an invite, so you should only invite if your expectation of playing 2NT facing a minimum or 3NT facing a maximum is greater than your expectation playing 1NT, plus a little bit for the digit that you might collect if LHO balances unwisely.

[helene_t]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

You may want to compare that 2% difference to the effect of reducing responder's strength with 1 HCP. Say if 1 HCP less means 4% less chance of making game, it means that the difference between 4333 and 4432 is roughly half a HCP. (It is a little more complicated than that since responder's HCPs are negatively correlated with opener's HCP, while the probabilities of being 4333 are positively correlated, but you get the picture).

 

Something else: People ten also to downgrade 4432's if the doubleton contain a jack or a queen. If you don't take that into account, you may reach the conclusion that 4333 are not much worse than 4432. But some of the 4432 you compare to would also have been downgraded by most players.

[dake50]

I like kenrexford evaluation. Two body cards to get back to evens hcp-count (after all 4333 has no source of tricks). Third body card invites.

[NickRW]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

I've done a lot of simulations of different specific combinations of distributions in both hands. I came up with 4333 being about a quarter of a point worse than 4432 and about half a point worse than 5332 (and about a point worse than something with a 6 carder) specifically for NT contracts. A 2 point downgrade is way way over the top.

 

4333 is worth about a point worse than 4432 for suit contracts with 5332 only very slightly better still.

 

Nick

[Cascade]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

 

Next, at my partner's request, I ran the same simulation, except constraining it to 2-4 in all suits. This only affected the result by 2%:

 

Making: 58%

Down: 42%

 

I think this really proves that these hands shouldn't be downgrading... Is this true?

I've done a lot of simulations of different specific combinations of distributions in both hands. I came up with 4333 being about a quarter of a point worse than 4432 and about half a point worse than 5332 (and about a point worse than something with a 6 carder) specifically for NT contracts. A 2 point downgrade is way way over the top.

 

4333 is worth about a point worse than 4432 for suit contracts with 5332 only very slightly better still.

 

Nick

Double dummy?

[TimG]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

Assuming you have a way to invite, I don't think you want to include 15 HCP for opener in your simulation -- whether you force or invite, you are always getting to game when opener has 15 and responder 11 or 12. 11 or 12 opposite 13 might make up a significant portion of the 40% of all hands that are down.

 

I also think if you do this by specific HCP pairs (opener 13, responder 11; opener 13, responder 12; opener 14, responder 11; opener 14, responder 12) you will be surprised by the different percentages you get. And, this should help you more with deciding which hands to invite with and which to force with.

 

To get some sort of idea of the value of 4333 hands, you ought to compare the results of two simulations: 1) 4333 11 counts opposite balanced 14 counts; and 2) 4432 11 counts opposite balanced 14 counts. Or similar. There will be a statistically significant trick difference.

[Mbodell]

OK, I ran a simulation on 2000 11-12 point 4333 hands opposite a 13-15 NT opener (2-4♠/♥, 2-5♣/♦) and came up with these results:

 

Making: 60%

Down: 40%

Assuming you have a way to invite, I don't think you want to include 15 HCP for opener in your simulation -- whether you force or invite, you are always getting to game when opener has 15 and responder 11 or 12. 11 or 12 opposite 13 might make up a significant portion of the 40% of all hands that are down.

While I agree more data to look at is valuable, when playing real single dummy defense telling about exact ranges may hurt much more than adding the small refinements in precision. If you invite and opener accepts you've told the defense, before the opening lead even, almost exactly what each of your HCP is. That may well cost you some good chunk of the wins.

 

I'd personally bid 3nt, 3nt, 2nt or 3nt depending on opponents skill, 3nt on the 4 hands at IMPs. Of course it may also depend how often partner opens a 12 point 13-15 nt. :)

[gwnn]

Stop downgrading! You're gonna get us all killed!

[NickRW]

I've done a lot of simulations of different specific combinations of distributions in both hands.  I came up with 4333 being about a quarter of a point worse than 4432 and about half a point worse than 5332 (and about a point worse than something with a 6 carder) specifically for NT contracts.  A 2 point downgrade is way way over the top.

 

4333 is worth about a point worse than 4432 for suit contracts with 5332 only very slightly better still.

 

Nick

Double dummy?

Yes, those were DD results. I've since used what I learnt at the table - I am not dissatisfied with the results - but counting in quarters or whatever isn't going to appeal to many.

 

Nick

[dellache]

1. Using DD simulations and SD checks, here are the results I came up with facing a 13-15HCP no trump :

- RED, always bid 3NT with 11+, never invite with less (never play 2NT)

- WHITE, always bid 3NT with 12+, usually invite with 11.

 

2. Downgrading 4333 by 2 points is way too much.

We have very good results downgrading them -1 (relatively to 4432 with no isolated quacks).

 

3. BTW, it seems that the fate of bad 3NT contacts that involve a 4333 depends more often than not of the strength of the oppos. So I will tend to relatively overbid when playing against not so good defense, or against tired experts.

--

FD

Codesigner of the Spider system.

[cherdanno]

Even a one point downgrade is too much.

[nigel_k]

Simulating two balanced hands of approximately equal strength and no 8 card major fit, I get:

 

Combined 24 HCP: less than 8 tricks 24.40%, 8 tricks 39.58%, more than 8 tricks 36.02%

Combined 25 HCP: less than 8 tricks 12.41%, 8 tricks 31.94%, more than 8 tricks 55.65%

 

Bidding 3NT with 24 combined HCP therefore has a gain of 0.5 IMPs per board when vul. To avoid underbidding you have to take into account the times you only lose 3 IMPs because 2NT doesn't make.

 

Non-vul you lose 0.3 IMPs per board bidding game with 24 HCP but gain 1.5 when holding 25 HCP. So this suggests bidding game any time you have a better than average 24. Plus, they always find the right lead double dummy but not in real life, especially after 1NT-3NT. So a good rule of thumb is probably that 24 HCP is enough for 3NT.

 

Regarding the 4333 shape, I wouldn't adjust it at all for NT bidding. For example I simulated KJ92 A32 32 KQ32 and KJ92 A32 432 KQ2 opposite a balanced 11 and they came out almost identical. Certainly an extra nine (maybe even an eight) would more than compensate for the difference between 4333 and 4432.

 

Obviously it's different at pairs because you want a higher chance of making and would like to stay with the field (to the extent you can when playing a 13-15 NT).

[WesleyC]

I did a simulation a while back and the results agree with Nigel_k. The setup was:

 

- North is 11-19

- South is semi-balanced with enough points to give the partnership a total between 20 and 30.

- 10,000 simulations each for North being 4333, 4432 and 5332.

 

Double dummy analysis suggested the average number of tricks when North's shape is 4333 is actually slightly HIGHER than when he holds 4432 or 5332. This unexpected results is likely caused by the limitations of double dummy analysis. "Cloud 3NT" contracts (as Kenrexford calls them) are aided by double dummy declarer play and occur more frequently when opener is 4333.

 

However, the underlying message is clear - 4333 hands should not be significantly devalued for declaring NT. I believe upgrading hands with Aces and interior texture, while downgrading hands with Queens and Jacks is a better way to split the marginal hands when considering a No-Trump invite.