overcalling

[jtfanclub]

Bidding needs to be based on offense-to-defense ratio (ODR) and not just offensive value. While it's true that a hand with three-card support and lousy values may be equivalent offensively to a hand with two-card support and an extra ace, the second hand is much better defensively than the first.

I agree with what you said, and I think it applies here. However....

 

Let's take the following FOUR hands.

 

1. Jxx xx Axxx xxxx

 

2. Jx xx AQJxx xxxx (remember, no 2344s- gotta have some shape).

 

3. Jxx xx AQJx xxxx

 

4. Jxx Qxx AJx xxxx

 

Do you agree that most people will bid 2♠ with hand 4?

 

It's certainly true that the first two hands will be worth close to the same number of tricks in a spade contract (the extra diamond king in the second hand compensates for the missing trump). The third hand is worth about one trick more than either of the first two. But the issue is, the first hand is much worse defensively than the second and third. If the auction continues with a competitive bid by opponents and partner has to decide whether to bid 3♠, he will be right to bid on either the first hand (3♠ likely down one, but their three level contract probably making) or the third hand (3♠ probably makes, their three level contract probably down one). He will be wrong to bid on the second hand (3♠ and their contract both probably down).

 

But the ODR for hand 4 is, I believe (I don't know the math) actually lower than the ODR of hand 2. The combination of very defensive shape (no side suit, no ruffing value) and impurity (honors in the opponent's suit) is extremely defensive. So while he would probably be wrong to continue on with hand 2, he would also be wrong to continue on with hand 4.

 

There is an additional issue when partner has "game try" values. On the first hand, it's generally right to reject a game try, and on the third hand it's generally right to accept. Easy enough. But the second hand is a pain -- if partner is bidding simply on power than you can probably make 3NT (but 4♠ can easily be hopeless). If partner is bidding on shape then 3♠ could easily be the limit (partner's shape not worth as much opposite only two trumps).

 

But you run into the same problem with hand 4. If partner's bidding on shape, you're screwed- the queen of hearts is probably worthless, and you have no ruffing values or long suits. If he's bidding on power, you've got nice points and fillers for him, 3NT should be great.

 

2♠ doesn't show a high ODR like 3♠ does. In my opinion, if you compare the 2 card raise with the 3 card high ODR raise, you're absolutely right. But if you compare the 2 card with with the 3 card low ODR raise, they come out about the same, with the same benefits and flaws.

[helene_t]

So you having 4♣ with opener does not make the odds higher that partner has spade support. Is that right? Then why did Mike Lawrence say club length makes the 4-card overcall more attractive? Is it just the unlikelyhood of a club overruff?

Thanks to Adam we now know that the chance of a fit is negatively correlated with your club length but that the correlation is extremely weak. It's possible that Mike Lawrence is a victim of the urban legend about the positive correlation. But I can see four cases for overcalling with club length:

- P may have ruffing value

- LHO may be more obstructed by the overcall because he doesn't have primary club fit.

- You don't have an alternative such as a t/o dbl

- RHO is likely to be balanced so even if LHO has a trap pass there's a good chance that RHO won't reopen with a double.

 

Two cases against:

- P has ♣ shortness so if the the board belongs to us he'll reopen

- LHO may ruff my club winners. Of course this issue depends on the texture of my clubs.

[barmar]

Cohen's rule is 'eight never, nine ever'. I was going to grab more quotes to explain where I was coming from, but I never got the chance.

Just make sure they're relevant. :) "8 ever, 9 never" refers to whether to finesse for a queen versus play for the drop, and has little to do with bidding decisions.

[jtfanclub]

Cohen's rule is 'eight never, nine ever'.  I was going to grab more quotes to explain where I was coming from, but I never got the chance.

Just make sure they're relevant. :) "8 ever, 9 never" refers to whether to finesse for a queen versus play for the drop, and has little to do with bidding decisions.

"8 never, 9 ever" is Cohen's rule for competing to the three level. It's a play on words from the finessing rule. I think it should be 5-3 never, 6-2 occassionally, but that doesn't sound as good. :D

 

http://www.larryco.com/BOOKS/following.htm

[barmar]

Oops, sorry.

 

Although I thought he recommended competing to 3♥ over 2♠ if both sides have 8-card fits. The LOTT says that if they could make 8 tricks, you're only going down 1, which is better (assuming you're non-vul or they don't double, and the latter tends to be rare in these auctions). "8 never" applies to whether to bid 3 over 3, not 3 over 2, although there are frequent exceptions due to LOTT adjustments.

[SoTired]

1. South has 4-2-3-4. Partner's number of spades:

Expected (average) number of spades is 3.0595

 

2. South has 4-4-3-2. Partner's number of spades:

Expected (average) number of spades is 3.0982

This is a difference of 4 parts per 1000. That is probably within the margin of error and, even if accurate, it is close enough to zero difference to not affect our bidding approach.

 

So our holding length in the enemy suit does not change our chance of having a fit with partner.

 

So.... overcalling a 4-card suit at the 1-level if:

1) hand not suitable for 1N overcall or t/o dbl

2) Suit is particularly strong

3) Hand is full opening bid

Partner still treats it as 5-card suit.

 

Is that a final consensus?

[skjaeran]

Count me in on the consensus.

[ralph23]

I'm in on the consensus although (just like a dang lawyer, isn't it ..) I quibble a little bit with number 3. I'm not going to be a fanatic on insisting on full opening hand. I do think it should be a "good" hand but "full opening" is just a bit too lofty for my relatively low (as my partner will eagerly attest :lol: ) standards.

 

Partner is entitled to expect 5+ and overcaller must remember this in all subsequent decisions.

[awm]

I agree with SoTired on the general principles.

 

However, the difference in spade length is four cards per hundred and not per thousand, which means basically one hand in 25 has fewer spades for partner when you've got club length. The number of trials is over 500,000 for each hand pattern so a 4% difference should in fact be statistically significant.

[awm]

This effect (partner has fewer cards in support if you have length in RHO's suit) is actually much more significant when the opening bid shows real length. Suppose this time that RHO opened 1♥ instead of 1♣, playing five card majors. Now partner's expected spade length:

 

If I am 4-2-3-4 is 3.33.

 

If I am 4-4-3-2 is 3.00.

[skjaeran]

I'm in on the consensus although (just like a dang lawyer, isn't it ..) I quibble a little bit with number 3. I'm not going to be a fanatic on insisting on full opening hand. I do think it should be a "good" hand but "full opening" is just a bit too lofty for my relatively low (as my partner will eagerly attest :) ) standards.

 

Partner is entitled to expect 5+ and overcaller must remember this in all subsequent decisions.

That depends on what you mean by full opening.

I open "all" 11-counts, and I won't have anything less than that I think.

[jtfanclub]

Although I thought he recommended competing to 3♥ over 2♠ if both sides have 8-card fits.   "8 never" applies to whether to bid 3 over 3, not 3 over 2, although there are frequent exceptions due to LOTT adjustments.

This is why I need to find my book. Must have lent it to somebody.

 

At any rate, in spades, it's safe. If the auction went 1♠ 2♥ P, I'm not raising with 2 card support.

 

The LOTT says that if they could make 8 tricks, you're only going down 1, which is better (assuming you're non-vul or they don't double, and the latter tends to be rare in these auctions).

 

Really? I X a lot on these auctions...in fact, the pass by the opp is a danger sign.

 

1♥ 1♠ PASS 2♠

3♥ 3♠

 

1♥ 1♠ PASS 2♠

dbl pass 3♥ pass

pass 3♠

 

1♣ 1♥ PASS 2♥

2♠ 3♥

 

1♣ 1♥ PASS 2♥

pass pass dbl pass

2♠ 3♥

 

Anybody going to be shocked when the next call is a double?

 

 

P.S. I'm in agreement with the consensus on the 4 card overcall.

[joshs]

Bidding needs to be based on offense-to-defense ratio (ODR) and not just offensive value. While it's true that a hand with three-card support and lousy values may be equivalent offensively to a hand with two-card support and an extra ace, the second hand is much better defensively than the first.

 

Take the following three hands:

 

Jxx xx Axxx xxxx

 

Jx xxx AKxx xxxx

 

Jxx xx AKxx xxxx

 

It's certainly true that the first two hands will be worth close to the same number of tricks in a spade contract (the extra diamond king in the second hand compensates for the missing trump). The third hand is worth about one trick more than either of the first two. But the issue is, the first hand is much worse defensively than the second and third. If the auction continues with a competitive bid by opponents and partner has to decide whether to bid 3♠, he will be right to bid on either the first hand (3♠ likely down one, but their three level contract probably making) or the third hand (3♠ probably makes, their three level contract probably down one). He will be wrong to bid on the second hand (3♠ and their contract both probably down).

 

There is an additional issue when partner has "game try" values. On the first hand, it's generally right to reject a game try, and on the third hand it's generally right to accept. Easy enough. But the second hand is a pain -- if partner is bidding simply on power than you can probably make 3NT (but 4♠ can easily be hopeless). If partner is bidding on shape then 3♠ could easily be the limit (partner's shape not worth as much opposite only two trumps).

While I am the first person to preach the ODR gospel there are at least two other issues in determing ODR here:

 

Having intermediate honors in partner's suit and primary honors outside has a greater ODR than the other way around. Just as I commented that when you overcall on a 4 card suit your side tricks should be fast (Aces and Kings) the same applies to raises when you are short a trump, you should have good trumps and prime side cards. You should not raise on a short trump holding when holding quacks on the side. There are too many ways that can be wrong.

 

Second, Anyone who thinks that with KQxxxx x xxx Axx that its clear they should compete to 3 over 3 after the auction (1H)-1S-(P)-2S-(3H) has either missed a bridge lesson or their partner's 2S bid is much stronger than mine is.

 

From a Law analysis point of view, you think your side has 9 trumps and the opps have 8 or 9, probably 9.

 

First lets suppose its 9 and 9 and estimate tricks. Partner did not cue bid, so probably will only supply 2 cover cards, possibly a 3'rd, and occasionally only 1. So your side can typically take about 8 tricks, and the opps can take about 10. Your best available score here is defending 3H. You don't want to push them to 4H. If there are only 17 total tricks then bidding 3 over 3 might work out ok at mps (when you can take 8 tricks) but still doesn't gain all that much at imps, unless the opps make a later mistake.

 

Now even supposing that partner has the right cards and shape for you to be able to make 3S, such as Axx xxxx xx Kxx (an example of the purity adjustment to the law). Whenever spades are 3-1 in the opps hand they will make 4H (they have 6H tricks and at least 4 diamond tricks). Bidding 3S is total brinkmanship.

 

I think to bid 3/3 when partner hasn't cuebid you need extra offense and at least some extra defense (e.g. high cards) beyond what the simple overcall showed.

 

Now the 4/4 issue is totally different (and you might well get this one wrong). But how often have you seen the auction

1H-1S-P-2S

4H

 

Anyway, thats my two bits.

[barmar]

The LOTT says that if they could make 8 tricks, you're only going down 1, which is better (assuming you're non-vul or they don't double, and the latter tends to be rare in these auctions).

 

Really? I X a lot on these auctions...in fact, the pass by the opp is a danger sign.

 

1♥ 1♠ PASS 2♠

3♥ 3♠

 

1♥ 1♠ PASS 2♠

dbl pass 3♥ pass

pass 3♠

 

1♣ 1♥ PASS 2♥

2♠ 3♥

 

1♣ 1♥ PASS 2♥

pass pass dbl pass

2♠ 3♥

 

Anybody going to be shocked when the next call is a double?

 

 

P.S. I'm in agreement with the consensus on the 4 card overcall.

I wasn't talking about auctions where it's clear that the 3-bid is a stretch. I'm talking about:

 

1♥ 1♠ 2♥ 2♠

3♥

 

1♣ 1♥ X/1♠ 2♥

2♠ 3♥

 

Your first two examples are 3 over 3, not 3 over 2. In your third example, opener has reversed opposite a partner who hasn't shown anything, showing a powerhouse; also, the opponents haven't found a fit (the LOTT mainly applies when both sides have a fit and points are about evenly divided).

 

I'm not sure I understand what's going on in your last example. Third hand didn't make a negative double on the first round, then he reopens -- he must have ♦ and ♥ (why is opener bidding ♠?), so of course he's going to double us.

 

At IMPs it's dangerous doubling these 3-level contracts, since you risk doubling them into game. Also, it's difficult to tell that they only have an 8-card fit.

[jtfanclub]

I wasn't talking about auctions where it's clear that the 3-bid is a stretch. I'm talking about:

 

1♥ 1♠ 2♥ 2♠

3♥

 

1♣ 1♥ X/1♠ 2♥

2♠ 3♥

But these are constructive auctions. Both opponents have bid and shown their hands. Raising with two card support on these is pointless.

 

There's a huge difference between these and:

 

1♥ 1♠ P 2♠

1♣ 1♥ P 2♥

---- 1♠ 2♣ 2♠

 

also, the opponents haven't found a fit (the LOTT mainly applies when both sides have a fit and points are about evenly divided).

 

But this is what I mean by taking Cohen to its logical conclusion. Preventing the opponents from knowing if either side has a fit is a winner, provided you can do so without going to the 3 level. Make them make the last guess.

 

On your sample auctions, one side of their partnership already knows whether they have a fit before your raise. There is no guessing involved.

[TimG]

This effect (partner has fewer cards in support if you have length in RHO's suit) is actually much more significant when the opening bid shows real length. Suppose this time that RHO opened 1♥ instead of 1♣, playing five card majors. Now partner's expected spade length:

 

If I am 4-2-3-4 is 3.33.

 

If I am 4-4-3-2 is 3.00.

My simulation does not produce the same result.

 

I have given East a shape that would open 1♥ in a standard 5-card majors system (5+ hearts, hearts longer than spades, hearts at least as long as diamonds, hearts at least as long as clubs).

 

I have given south 4 spades, either 2 or 4 hearts, at most 4 diamonds and at most 4 clubs.

 

When south has 4 spades and 2 hearts, north's expected spade length is 3.312 (about 618,000 matches);

when south has 4 spades and 4 hearts, north's expected spade length is 3.357 (about 382,000 matches).

 

I'm going to tinker a little to make sure I haven't made any errors. But, if anyone else would like to do their own simulation, that would be appreciated.

 

Tim

[TimG]

1. South has 4-2-3-4. Partner's number of spades:

 

Expected (average) number of spades is 3.0595

 

2. South has 4-4-3-2. Partner's number of spades:

 

Expected (average) number of spades is 3.0982

I got

 

1) 3.0554

2) 3.1009

 

so, we're in agreement here. The tinkering hasn't revealed any problems with my simulation involving RHO's heart opening.

[hanp]

This is very interesting Tim, I assumed Adam's big difference (0.33 is a very large difference) was correct. Perhaps Adam could try to redo his simulation? I might try to do the same simulation as well.

[awm]

Re-running this, my results look more similar to Tim's. I'm not sure what exactly I did in my second simulation two years ago, except that I wasn't as careful as in the first sim.

 

One explanation of what's going on here is the following. Suppose we knew that RHO has exactly five hearts. Then if we are 4♠-2♥, the hearts are on average 5-3-3-2 around the table. Thus partner has 10 open spaces of 28 and in expectation should hold around 3.21 spades. On the other hand, if we are 4♠-4♥ then the hearts are on average 5-4-2-2 around the table. Partner has 11 open spaces of 30 and in expectation should hold 3.3 spades. Note that these are just a little less than Tim's actual numbers (with the difference being accounted by hands where RHO has 6+♥).

 

This means if RHO's number of hearts is fixed, then we expect partner to have more spades when we have more hearts. So basically there are two effects:

 

(1) We have more hearts --> RHO has less hearts --> RHO has more spades --> pd less spades

(2) We have more hearts --> partner has less hearts --> pd more spades

 

The question is which effect is "stronger." If we had no additional information at all about RHO's hand then they would exactly cancel out and partner would have on average 3 spades either way. So the issue is, given the extra information we have about RHO's hand, which effect becomes stronger.

 

When RHO opens 1♥, playing five-card majors, his expected number of hearts is actually pretty close to five. So increasing our number of hearts doesn't have all that much impact on RHO's number of hearts. This weakens effect (1) substantially, allowing effect (2) to become stronger and giving the effect Tim observed.

 

When RHO opens 1♣, playing five-card majors, his expected number of clubs is actually a lot higher than three. When we increase our number of clubs, it pushes RHO's number down more substantially. In addition, the hands with few clubs inevitably have some spades (i.e. if RHO has only three clubs, he must have at least three spades). It turns out that effect (1) is even stronger than it would be if we knew nothing about RHO's hand, and we see the effect of my first observation (which Tim agrees with).

[hanp]

I also did the simulations, although in my simulation I didn't take HCP into account. First I gave us 4432 or 4234 distribution and had RHO open 1H. Average number of spades in partner's hand:

 

4432: 3.357

4234: 3.312

 

Each was for 1,000,000 hands.

 

Using some more extreme shapes:

 

4711: 3.411 (note: this was much slower so I only ran 100,000 deals, still a large number)

4621: 3.402

4522: 3.382

4432: 3.357

4234: 3.312

4144: 3.288

4045: 3.264

 

So we see that in this case it is indeed true that the more hearts we have, the more spades partner will have on average. I still won't overcall 1S with 4711 shape though.

[hanp]

Rerunning the simulation when RHO has diamonds:

 

4423: 3.110

4243: 3.112