[joshs]

OK I did the expectations calculation, under the following assumptions:
A. in 1'st seat we open all 10+ hands and no weaker hands
B. In Second Seat they open all 12+ hands, and half the 11's
Here I am basically ignoring prempts. I don't think including them is going to change the conclusion much, but feel free.

From a table of HCP probabilities, I computed
E(HCP for A Given A passed)=6.47
E(HCP for B Given B passed)=7.35
Average for A+B=13.82
3'rd and 4'th seat have the same expectation=(40-13.82)/2=13.09

So 3'rd hand sides expectation=6.47+13.09=19.56 (I actually think this will go up slightly if you include premepts, assuming the prempting styles are similar since you are including some common hands, plus some hands that A opened 1 which B would pre-empt with)

OK So I guessed 19.7. Seems like a rediculously bad guess to me.

[awm]

I don't understand the relevance of our expected total high card strength.

Surely we should evaluate light openings on the basis of how many imps/mps we win by opening light, versus how much we lose on the sound opening bids by playing a wider opening range? Just because I have "on average" 13 hcp doesn't mean I shouldn't open with 8 and a suit -- in fact when I have 8 hcp it's probably a great time to open because opponents usually have game?

[benlessard]

Quote

People keep making claims about opening light in 3rd seat is important. Ok, just why is it more important (more effective) than opening light in say 1st seat?

There is no comparaison between the effectiveness of LOB of a 3rd seat opening and in 1st seat. Im not a strong believer in lob but light in 3rd is at least 2 or 3 time more effective then light first seat.

1- from a preemptive point of view you are preempting 1 opps and no partner. in 1 seat you are preempting 2 opps and 1 partner. Whatever psych or garbage preempt or crappy opening can backfire in 1st seat in 3rd they rarely do. Of course 1st bid are affecting 2 opps but the risk make it no comparaison between the effectivness on 3rd seat light.

2- from a constructive point of view already having pass is a big advantage for constructive bidding.
P---(p)----1H-----(p)
???

whatever 1st seat bid and whatever the continuation hes in much better position then if not a passed hand, this extra precision can allow for some slack in the requirement for the third set openings.

Quote

You think you will lose by less than 3 imps/bd on the strong club hands?

Ill drink half a bottle of Johnny Walker and i still wont lose 3 imps per boards against nobody.

Remove my 1C bid from my bidding box and i dont think ill lose 1 imps per board.

[pclayton]

I think Josh (Sher's) reasoning makes a lot of sense. Boost the minimum requirement for 1♣ and lower the requirement for a positive (or semi-positive). 1♣ - 1♦ sequences should not be the norm, and you shouldn't have a one or two point spectrum for a positive response.

I think this has only a small effect on your one bids. Like Han, I think opening lite in 3rd seat is still a big winner.

However, because you'll have a class of 15-16 / 15-17 hands in 3rd and 4th seat that you don't have after a 1/2 seat opener, I would recommend adding Gazzilli, or making the opening 2 bids beefier or something.

Josh - you might want to tighten up the IMP predictions however.

[joshs]

awm, on Feb 27 2008, 03:03 PM, said:

I don't understand the relevance of our expected total high card strength.

Surely we should evaluate light openings on the basis of how many imps/mps we win by opening light, versus how much we lose on the sound opening bids by playing a wider opening range? Just because I have "on average" 13 hcp doesn't mean I shouldn't open with 8 and a suit -- in fact when I have 8 hcp it's probably a great time to open because opponents usually have game?

Some Definitions:

"Par" - The best possible double dummy score you can recieve on the hand if the opps pass through out.

Note Par>=0

"Absolute Par" - A contract where neither side can improve their (double dummy) score by bidding higher.

Note: Absolute Par is usually unique, but on rare occasion both sides can make say 1N or 3N double dummy, played from a particular side, and here its not very well defined, so you would assign the side in 1'st seat the unique ability to get to said contract, to make it unique. There still are rare cases where Passout is Absolute Par.

Owner of the Hand: The side that went postive (double dummy) in the Absolute Par Contract

Claim: Given a Fixed Distribution of the HCP (A against B=40-A) and ranging over all possible hands with said distribution of the points that if A>B, then P(A Owns the Hand)> P(B Owns the Hand)

Three more definitions:
Constructive - making a bid with the aim of reaching my sides par contract

Destructive - makign a bid with the aim of preventing the opponents from reaching their par contract.

Lead Oriented - making a bid with the aim of improving the effectivesness of your sides opening leads

(Note this can be a bid that improves your partner's opening leads, or it can be a bid by you where the subsequent auction helps you figure out what to lead yourself)

Let me combine Descructive with Lead Oriented and call these Defensive Goals.

Ok that was preliminary, now my claim is:
The More likely it is for your side to own the hand, the more important Constuctive Goals are to Defensive Goals. And Visa versa.

Hence has your expected HCP decreases, there is some increase in the relative importance of Defensive bidding relative to constructive bidding.

I just think:
a. After PP, the probability that its your hand has only decreased a little bit from first seat, hence the importance of defensive bidding has only increased a little bit.
b. Hence the more important consideration is effectiveness, not importance. Is defensive bidding more effective in this situation (because partner is a passed hand) than it is in first seat (where the opps have exchanged no information, but partner is unlimited)?

I think most of the claims for the value of 3rd seat light opening bids is that they are much more effective than in 1'st seat since "partner can't hang you".

I have been questioning this claim. I feel partner can't hang you only by giving up on constructive bidding, which really is not much less important here than it was in 1/2 seat. If you want to be able to retain some constructive bidding, partner will have to hang you sometimes. (Seriously, after P-P-1S-3H(Intermediate) and you hold Qxx x xxxx AQTxx I think you need to force to game) So the question is, what are the "costs and benefits" assocaited with opening 1-2 points "lighter than expected" on certain hand types in different seats. Also what are the costs and benefits of more outragous psychs.

[awm]

Here's another way to look at things.

There's a pretty overwhelming consensus that if you play sound openings in 1st chair, you should open light in 3rd. Let's assume that this consensus is correct.

Now suppose you're opening light in 1st chair. So if your partner passes in 1st, they would pass playing sound openings too. However, there are some hands where partner passes playing sound openings but bids playing light openings (say 10-11 points). Consider what happens when we open light opposite such a hand. If this generated mostly bad results then we'd be strictly better opening light when partner can't have this hand. At the same time, when we have a "sound opening bid" the knowledge that partner can't have 10-11 lets us improve our follow-up methods. So we'd get even more benefit opening light in a light-opening system.

Thus it seems like any argument that we should not open light in 3rd chair in a light opening system should stem from reasoning that opening light does best when partner has 10-11 points and not so well when partner passed on a weaker hand.

Curiously, my experience has been mostly the opposite of this. Despite playing drury, it seems like a lot of times when I open on 10 high and partner has 10-11, we end up getting too high. Partner often has a non-fitting invite, or has to make his three-card raise in competition. The best results I've gotten by opening lousy hands in 3rd chair come when it helps partner on lead against the opponents' contract, or when partner's raise pushes their partscore up a level, or when the opponents have trouble finding their game or slam because they had to bid over my 1♠ call and/or partner's raise. All of these work out a lot better when partner has something like 5-6 points rather than 10-11.

[cherdano]

joshs, on Feb 27 2008, 02:00 PM, said:

OK I did the expectations calculation, under the following assumptions:
A. in 1'st seat we open all 10+ hands and no weaker hands
B. In Second Seat they open all 12+ hands, and half the 11's
Here I am basically ignoring prempts. I don't think including them is going to change the conclusion much, but feel free.

From a table of HCP probabilities, I computed
E(HCP for A Given A passed)=6.47
E(HCP for B Given B passed)=7.35
Average for A+B=13.82
3'rd and 4'th seat have the same expectation=(40-13.82)/2=13.09

So 3'rd hand sides expectation=6.47+13.09=19.56 (I actually think this will go up slightly if you include premepts, assuming the prempting styles are similar since you are including some common hands, plus some hands that A opened 1 which B would pre-empt with)

OK So I guessed 19.7. Seems like a rediculously bad guess to me.

Sorry I thought of a different question, which seems the more relevant one to me. Say you have 7 hcp in 1st seat. Then obviously your side's hcp expectation value is 18 hcp. Now if you have 7hcp in 3rd seat and it went p-p and partner opens light, then obviously your expected number of hcp is a lot lower.
That's why there is a lot more reason to open destructively in 3rd seat, and what your a-priori expected number of hcps is seems rather irrelevant to me.

[jdonn]

cherdano, on Feb 27 2008, 08:48 PM, said:

joshs, on Feb 27 2008, 02:00 PM, said:

OK I did the expectations calculation, under the following assumptions:
A. in 1'st seat we open all 10+ hands and no weaker hands
B. In Second Seat they open all 12+ hands, and half the 11's
Here I am basically ignoring prempts. I don't think including them is going to change the conclusion much, but feel free.

From a table of HCP probabilities, I computed
E(HCP for A Given A passed)=6.47
E(HCP for B Given B passed)=7.35
Average for A+B=13.82
3'rd and 4'th seat have the same expectation=(40-13.82)/2=13.09

So 3'rd hand sides expectation=6.47+13.09=19.56 (I actually think this will go up slightly if you include premepts, assuming the prempting styles are similar since you are including some common hands, plus some hands that A opened 1 which B would pre-empt with)

OK So I guessed 19.7. Seems like a rediculously bad guess to me.

Sorry I thought of a different question, which seems the more relevant one to me. Say you have 7 hcp in 1st seat. Then obviously your side's hcp expectation value is 18 hcp. Now if you have 7hcp in 3rd seat and it went p-p and partner opens light, then obviously your expected number of hcp is a lot lower.
That's why there is a lot more reason to open destructively in 3rd seat, and what your a-priori expected number of hcps is seems rather irrelevant to me.

typ

[hrothgar]

awm, on Feb 28 2008, 04:37 AM, said:

Here's another way to look at things.

There's a pretty overwhelming consensus that if you play sound openings in 1st chair, you should open light in 3rd. Let's assume that this consensus is correct.

Now suppose you're opening light in 1st chair. So if your partner passes in 1st, they would pass playing sound openings too. However, there are some hands where partner passes playing sound openings but bids playing light openings (say 10-11 points). Consider what happens when we open light opposite such a hand. If this generated mostly bad results then we'd be strictly better opening light when partner can't have this hand. At the same time, when we have a "sound opening bid" the knowledge that partner can't have 10-11 lets us improve our follow-up methods. So we'd get even more benefit opening light in a light-opening system.

Thus it seems like any argument that we should not open light in 3rd chair in a light opening system should stem from reasoning that opening light does best when partner has 10-11 points and not so well when partner passed on a weaker hand.

I don't think that the logical inferences from sound opening systems carry over to this case.

The reason that sound need to open light in 3rd/4th opposite Al Roth is that Roth might have passed a 14 count. If you bid the same you're going to miss an awful lot of games, not to mention all sorts of nice attractive part scores.

The reason that you open sound in 3rd/4th opposite a light opening system has to do with constructive allocation of bidding space.

Lets assume that you're playing an agressive strong club system in first and second and you're opening unabalcned 10 counts and most juicy 8-9 counts with a constructive opening. Moreover, your preemptive structure is going to be covering most unbalanced 6-9 counts.

Your passes are probably going to show either

a balanced 7-11 count
a three suited 7 - 11 counts
absolute crap

If you open very light in 3rd and fourth as well you're probaby going to go overboard on way too many hands...

[jtfanclub]

Why not go the other direction?

Take the 1 club opener show 15+, and have there be no positive responses.

1 diamond, 1 heart, 1NT, and 2 clubs are all transfers to the next suit, 1 spade is a transfer to NT. Opener accepts with 15-20, jump accepts or bids NT with 21+, and shows his own suit with a misfit.

Nothing shocking or new, except to do it on third seat only.

[han]

hrothgar, on Feb 28 2008, 08:36 AM, said:

The reason that you open sound in 3rd/4th opposite a light opening system has to do with constructive allocation of bidding space.

Great, does this have anything to do with winning?

[pclayton]

joshs, on Feb 26 2008, 03:37 PM, said:

The light 3'rd seat openers are not nearly as effective at that goal...

hrothgar, on earlier today, said:

If you open very light in 3rd and fourth as well you're probaby going to go overboard on way too many hands...

Other than system symmetry, explain why this is true. My experience is that it's not, if partner makes the appropriate allowances.

If pard doesn't have at most an 8 or a balanced 10, then opening light in 3rd seat looks even better, especially NV. Holding an 8 count, we simply *know* the opponents have game and we should jam them, or get our lead director in.

I suppose what Josh might be implying is that if we are playing our 1♣ opening in 3/4 as, say 18+, we need to keep the bidding open for pard with say a 2344 6 count after 1♠ JIC pard has the 17. This runs counter to traditional precision philosophy that says you pass 1 major with up to an 8 count and tolerance.

However, I haven't seen any real good strong club pairs bid this way in quite some time. Look at Meckwell. When was the last time they passed with a 7 count opposite 1 major?

[mycroft]

One interesting thing I see from Josh's simulation is that the expected value of partner's hand, having passed in first, is almost 7 HCP. Now, that will go down a bit given an additional criterion - third hand has a 1C opener. But that still means that the number of positives opposite (okay, a 16+, 17+bal, or the like; I play Precision rather than MOSCITO) is not the "very small" that was originally said. Sure, it's significantly less than half, but it isn't "we're always going to have p-1C-1D auctions", either.

If your goal is to get to the games, then perhaps stiffening third hand, stiffening strongly 1C is the right thing to do; it opens up a lot of sequences for game tries and other invites. If, however, you are of the "make 'em guess" school, you're going to be using those same sequences to show different kinds of "drop dead" bids, at the cost of some subtle game probes. At that point, it is likely best to play the same light openings and feel comfortable that "Pass opposite non 1C hand = no game, we're starting the partscore battle with the opener", setting the system so that as much as possible, responder's calls are NF, and as many of them are "to play" as possible, as well.

I think, that provided the system gets you to safe "pre-balance" places, that's a workable goal. So is "play sound, use the space to improve game bidding".

Michael.

[Echognome]

A bunch of comments:

1. If allowed in the competition in which you are playing, I'd much rather play a natural system in 3rd/4th opposite those conditions. You get back a natural, potentially lead-directing 1♣ response. If you're 1♦ is nebulous, it will no longer be nebulous. I'm indifferent on the 2♣ opening because it's also somewhat preemptive, however constructive auctions by a PH seem limited.

2. If not allowed a different system in 3rd/4th, then I would definitely beef up the 1♣ openings. (We don't but we already play 16+ and pass all bad balanced 12 counts.) I don't really understand the concern about the wide-ranging 1-bids. What do you think the range of 1-bids in a natural system where people make light openings in 3rd? Something like 8-22 I would guess. It's also unclear to me when I am playing a different system in 3rd/4th, but I guess just messing with the hcp ranges is ok.

3. I would make preempts slightly wider-ranging. (Beware the Gnome 3rd seat preempts.)

4. I would make my NT range at least 14-16 minimum. Weak NTs should either bid a lead-directing suit (4cM ok) or be content to pass.

5.

[han]

My thoughts exactly Matt.

[joshs]

cherdano, on Feb 27 2008, 08:48 PM, said:

joshs, on Feb 27 2008, 02:00 PM, said:

OK I did the expectations calculation, under the following assumptions:
A. in 1'st seat we open all 10+ hands and no weaker hands
B. In Second Seat they open all 12+ hands, and half the 11's
Here I am basically ignoring prempts. I don't think including them is going to change the conclusion much, but feel free.

From a table of HCP probabilities, I computed
E(HCP for A Given A passed)=6.47
E(HCP for B Given B passed)=7.35
Average for A+B=13.82
3'rd and 4'th seat have the same expectation=(40-13.82)/2=13.09

So 3'rd hand sides expectation=6.47+13.09=19.56 (I actually think this will go up slightly if you include premepts, assuming the prempting styles are similar since you are including some common hands, plus some hands that A opened 1 which B would pre-empt with)

OK So I guessed 19.7. Seems like a rediculously bad guess to me.

Sorry I thought of a different question, which seems the more relevant one to me. Say you have 7 hcp in 1st seat. Then obviously your side's hcp expectation value is 18 hcp. Now if you have 7hcp in 3rd seat and it went p-p and partner opens light, then obviously your expected number of hcp is a lot lower.
That's why there is a lot more reason to open destructively in 3rd seat, and what your a-priori expected number of hcps is seems rather irrelevant to me.

Yeah that is of course if I have a 7 count in 3rd seat the opps expectation is much higher than if I have a 7 count in 1st seat (26 vs 22). I thought we were talking about
opening 1 or 2 points lighter and am not talking about outright psychs. The question here was what should the nominal opening bid range be for a 3rd 4'th suit opener in this light opening bid system and I was arguing for 12-16ish, and the debate was if it should include the 10-11 counts that normally get opened opposite sound openers.

Now...

If I held a 3 count with 5 spades in first seat favorable I probably will get my best results by opening the hand 1S. Of course, if partner has to cater to this my results on other, more common hands will be worse. But the results with that hand will be better.

Similarly, if I have a 7 count with 5 spades in 3'rd seat favorable I probably will also get my best results by opening the hand 1S, at least if partner never hangs me. But again, if partner has to cater to this, my results on other more common hands will be worse. (You hold a 13 count here about 3 times as often as a 7 count)

As usual its just a cost benefit analysis....

I know one pair here in LA who are very wild in 3rd seat and consistantly does terrible over there 3rd seat openers because they do not very often get to the correct contract when it is there hand. I think a lot of the success people have with wild 3rd seat actions is that their partner bases there decisions on their table feel of 4th hand (and perhaps via the old black magic, on partner)

[joshs]

Help with a calculation:

Suppose I have a 7 count in 3rd seat under conditions described. Does anyone have a good way of calculating each sides expected HCP?

Here is my loose calculation:

Before looking at my hand the expectations were
A:6.47
B:7.35
C:13.09
D:13.09

Now I see my 7 count and now I am 6.09 HCP short of my expected values. I think that these 6.09 points should be distributed proportionally to my aprior estimates for A B and D,
Hence my new estimates:
A: 6.47 +6.09*6.47/26.91=7.93
B: 7.35 +6.09*7.35/26.91=9.01
C: 7
D: 13.09+6.09*13.09/26.91=16.05

(I lost 0.01 do to round off)

In any case this give our side an expectation of about 14.93.
I am not sure this proportionality distribution method is exactly right, does anyone
have an easy way to do this calculation exactly?

[han]

joshs, on Feb 28 2008, 03:17 PM, said:

I think a lot of the success people have with wild 3rd seat actions is that their partner bases there decisions on their table feel of 4th hand (and perhaps via the old black magic, on partner)

Yes, those two things (unfortunately also the second), but also:

-That third seat opens light on the right hands.

-That first seater does not hang partner for these openings.

[awm]

It does seem like one could do a bridge browser study on this. It helps that it's a style thing more than a system thing, and most people on BBO are playing a fairly standard system (often in pickup partnerships).

The relevant points are something like:

(1) If you open in 3rd chair after two passes on particular hand types, what is your expected IMP score? How awful do the openings have to be before you start losing IMPs (when compared to passing)? Obviously in Josh's example (the pair in LA that opens super-light) they do well on the light openings and much worse when they have a sound opening due to their unusual agreements, but in a pickup or short-term partnership (as most BBO pairs are) we're unlikely to see this kind of effect.

(2) If you open in 3rd chair after two passes on a particular hand type, how does your score correlate with passed hand partner's high card holding?

For example, suppose the auction goes Pass-Pass to me and I hold a 5(332) 8-count. How do I score when I open 1♠ versus pass (maybe also consider 2♠)? If we condition on partner holding 9-11 points for the initial pass in addition to my holding a 5(332) 8-count, how do I score for 1♠ versus pass? How about if we condition on partner holding 0-8?

[Echognome]

awm, on Feb 28 2008, 03:53 PM, said:

For example, suppose the auction goes Pass-Pass to me and I hold a 5(332) 8-count. How do I score when I open 1♠ versus pass (maybe also consider 2♠)? If we condition on partner holding 9-11 points for the initial pass in addition to my holding a 5(332) 8-count, how do I score for 1♠ versus pass? How about if we condition on partner holding 0-8?

The difficulty is that I would be inclined to open 1♠ on AKJTx xx xxx xxx, but not on Qxxxx Kx Qxx Jxx. I think suit quality is a major factor in deciding whether to open light in 3rd.