[Jinksy]

 mikeh, on 2014-December-08, 18:11, said:

As is usually the case, the people who advocate one style are quite good at pointing out the merits of their style but reveal their ignorance when they claim to set out the pros and cons of the opposite style. I am not suggesting any intention to mislead...this is just human nature. We tend to see reasons why we do what we like to do, and to ignore reasons for doing something else.

I think a good charitable principle that gets overlooked (inc by me) too often is that almost every gain for System A over System B must have a contrastable loss. If we communicate more in one auction, it typically means that in another we've communicated less in another (eg we've said more in XYZ after 1x 1y / 1N than in the alternative, more in our style after 1x 1y / 1z).

Even the benefits themselves on any given auction are often irregular verbs: I find a better contract, you reveal more info to the defence; I find thin games, you bid dodgy games; I find a better denomation, you give them room to compete.

All else being equal, I pay attention to system arguments to the extent that their authors are scrupulous about acknowledging the trade-offs (and then explaining why they're worthwhile in expectation).

[mycroft]

 helene_t, on 2014-December-09, 02:14, said:

One disaster occurred when partner thought my 1nt rebid promised four clubs (which it would if I would never bypass, and which is a case for bypassing since otherwise you tell opps not to lead a club).

Odd, that's the second time I've heard this. When I did play not-bypass, we played that 1♣-1y; 1♠ *promised 4-4 at least in the blacks", so 1NT rebid would guarantee exactly 3 clubs if it had 4 spades.

I don't understand why one would play the other way, except I guess to guarantee 4 clubs for the pullback later. But I play XYZ, or CBS, or whatever, and don't have a 2♣ retreat anyway. I'd much rather "imply unbalanced" with my suit bids than "includes 4333s but not 4234s". Like Mike says, of course, I'm sure I have a blindspot for my own methods, though, and I'm sure there *is* a reason for it.

[jallerton]

 mikeh, on 2014-December-08, 18:11, said:

The other main theoretical problem seems to be playing 2N when opener lacks both 4 spades and 3 hearts. I accept that this is a real issue, but again there seems to be a tendency to overstate matters. I have played the bypass method for many years, including in world championships, and have not ever, to my recollection, lost a swing on this layout. Yes, obviously it can happen, but bear in mind that when 2N fails, the 5-2 heart fit that the up-the-line bidders claim they would get to almost certainly doesn't happen.

Say you hold an invitational hand with 4=5 majors, and partner rebids 1N, denying spades. Are you passing? No. You are relaying to find if partner has 3 hearts. He bids, say, 2♦ (because you use 2-way), and now you bid 2♥, invite with 5 hearts. Why won't he bid 2N? Why does he have to pass 2♥? He has no idea how strong your hearts are and to pass with, say, Qx and find you have Jxxxx isn't going to work out very well. Even if he does pass, why do we infer that the 5-2, on often dubious trump, will play better than 2N? Maybe it will, maybe it won't. I doubt that we could show much basis for saying opener should always pass 2♥ with a minimum and 2 hearts.

This is a downside of playing 2-way checkback rather than anything else.

I agree with the rest of your long post though. I recall a hand from the Bermuda Bowl a few years ago on which one of the USA pairs bid uncontested 1♣-1♥-1♠-3♣-Pass. 3♣ going off in a 3-4 fit with two balanced hands was a silly contract with 8 or 9 tricks easily available in a NT part score.

[Trinidad]

 Giangibar, on 2014-December-07, 16:30, said:

Yeah, I'm sorry. I wrote "checkback" but my Macbook automatically converted it into "checkbook"... I have now corrected my original post. Sometimes the autocorrect feature is cumbersome

Perhaps you should send your Macbook Mac back.



Rik

[johnu]

 akwoo, on 2014-December-08, 17:07, said:

The reason is that your system might give you 55%, but your card play gives you 65%.

A top player is not aiming just for better than average; he or she is aiming for much better than average.

Of course things change when you're playing teams (either BAM or IMPs) or in a very very strong field where no one is 65%.

Then design a system that gives you 65% In real life, without assigning numbers, just play a system that gives you the best overall results. If you start with a system that gets you to higher percentage contracts, superior card play will only increase your advantage. If you voluntarily throw away your advantage by playing the same system as the hoi polloi , you are decreasing the gap between you and the average/below average players. Things don't really change in a very strong field except you need superior bidding methods and card play compared to the average of the field to get consistent top results.

[helene_t]

I am a bit perplexed by all this talk about playing with the field. You don't know what the field does. Even if everyone plays the same general approach there will usually be judgement and style differences. You might carefully navigate to the inferior field contract, hoping to make a top by superior play, only to discover that you are the only declarer to receive the killing lead, or that an unlikely split makes your superior line lose. Or that half of the people who declared this board were better declarers than you. Or that most of them found the superior contract. Or that your solid play gives you 60% while you needed 90% to get in the prices.

Stephanie gave a good example of a situation where you want to play low variance, namely a qualifying round. But generally it is not clear if variance is good or.bad.

[awm]

I've played both styles pretty significantly (bypassing in my current partnership with Howard Liu, bidding up the line with Elianna and Sam). The reality is that this will usually not make much difference. If you play decent check-back methods, it should not matter on invitational or better hands. You have plenty of space to sort out your major fits and opener's potential unbalanced shape.

There are two main places where I've found this to matter, both most often on partial hands. Bidding up the line gets you to better part-score contracts, because you don't miss 4-4 spade fits, and because you can sometimes play 1♠ in a moysian fit. I have not found it to be the case that we often play 2m on a 5-3 after opener shows an unbalanced hand, or that 2m would often be better than 1NT in these cases. Of course, this is all much more significant at MP scoring!

The second place is that opponents seem to make worse leads against 1NT when we bypass spades. Some of this is a lack of awareness of the style, which feels a little like bad disclosure (and in fact some opponents have complained that we should alert, but it's not alertable by any official rules I'm aware of). However, I do think that opponents truly have less information here. Of course, this advantage is much less if we land in 3NT because of check-back or lack thereof (often they have a lot more information about responder's hand on these auctions, which compensates for less about opener).

I will comment that things are different in a weak notrump system, where it seems that bypassing spades to rebid 1NT carries a lot more information about strength (strong notrump) that might make it easier to bid to the right games.

[mikeh]

 awm, on 2014-December-10, 10:11, said:

I've played both styles pretty significantly (bypassing in my current partnership with Howard Liu, bidding up the line with Elianna and Sam). The reality is that this will usually not make much difference. If you play decent check-back methods, it should not matter on invitational or better hands. You have plenty of space to sort out your major fits and opener's potential unbalanced shape.

There are two main places where I've found this to matter, both most often on partial hands. Bidding up the line gets you to better part-score contracts, because you don't miss 4-4 spade fits, and because you can sometimes play 1♠ in a moysian fit. I have not found it to be the case that we often play 2m on a 5-3 after opener shows an unbalanced hand, or that 2m would often be better than 1NT in these cases. Of course, this is all much more significant at MP scoring!

The second place is that opponents seem to make worse leads against 1NT when we bypass spades. Some of this is a lack of awareness of the style, which feels a little like bad disclosure (and in fact some opponents have complained that we should alert, but it's not alertable by any official rules I'm aware of). However, I do think that opponents truly have less information here. Of course, this advantage is much less if we land in 3NT because of check-back or lack thereof (often they have a lot more information about responder's hand on these auctions, which compensates for less about opener).

I will comment that things are different in a weak notrump system, where it seems that bypassing spades to rebid 1NT carries a lot more information about strength (strong notrump) that might make it easier to bid to the right games.

My experience has been different from your, tho maybe because I rarely play anything but imps. There, avoiding the rebid dilemma after 1♣1♥1♠ on Qxx AQxx 10xx Jxx is fairly worthwhile: I can safely bid 2♣.

As for the lead, my practice has always been to announce that the 1N doesn't deny spades, unless I am playing serious bridge, in which case it is for the opps to ask, and they either already play that same style or at least are aware of the possibility. Not announcing against the typical lol (of either gender) in a swiss or early stage of a regional feels wrong to me, even if technically ok.

In addition, concealing declarer's hand is always more useful than concealing responder's. For one thing, opening lead is through dummy and into declarer, and that is an important difference. In addition, dummy is revealed early, while declarer's hand often can't be reconstructed for at least several tricks.

[aguahombre]

For the bypassers, when responder has inv+ values it seems neither 2-way checkback nor plain PLOB adequately compensates. 2S is needed to show the suit, and that removes it as an effective tool in sorting out both Opener's size (within the NT rebid range) and shape..at a convenient level. The importance of this wrinkle --compared against the plusses mentioned by Mikeh is questionable. Also, the importance of being able to bail out in 2 of Opener's minor is questionable. But, they are factors in our choice.

[Jinksy]

Is XYZ particularly prevalent in the US? It never would have occurred to me to mention it to UK opps, on the grounds that I've never encountered any pair who claim to play it.

I guess that might be related to weak NT being prevalent over here.

[mikeh]

 aguahombre, on 2014-December-10, 10:53, said:

For the bypassers, when responder has inv+ values it seems neither 2-way checkback nor plain PLOB adequately compensates. 2S is needed to show the suit, and that removes it as an effective tool in sorting out both Opener's size (within the NT rebid range) and shape..at a convenient level. The importance of this wrinkle --compared against the plusses mentioned by Mikeh is questionable. Also, the importance of being able to bail out in 2 of Opener's minor is questionable. But, they are factors in our choice.

I don't understand your concern about sorting out values. I suppose it depends on the scheme. Here's what I play:

2♣ puppets to 2♦, to play or about to make an invitational action. If responder bids 2♠ now, it shows 5+ hearts and 4+ spades, longer hearts always

2♦: artificial gf, opener to make cheapest descriptive action

2♥: if playing weak jumpshifts, then this is constructive but less than invitational. Opener will rarely move but is allowed to do so on exceptional hands. The constructive info is primarily to assist in bidding and defence if the opps balance. If not playing weak jumpshifts, then this is simply 'to play'.

2♠: invitational, precisely 4=4 majors

2N: puppet to 3♣, usually to play but can be various slam tries with 4 hearts and 5+ opener's minor, with all values in the two suits

3suit: slam try. If new suit, including opener's minor, then 5-5 or better, concentrated values, no side control. If 3♥, then rejects 3N as playable contract (allows 3N to be used for other purposes). With COG long heart suit, go through 2♦

3♠: autosplinter (exception to metarule that 3new suit is natural)

I don't recall ever having much trouble sorting things out with this style.

[aguahombre]

 Jinksy, on 2014-December-10, 11:03, said:

Is XYZ particularly prevalent in the US? It never would have occurred to me to mention it to UK opps, on the grounds that I've never encountered any pair who claim to play it.

I guess that might be related to weak NT being prevalent over here.

If we can gather from the North American BBO contributors..yes some form of XYZ seems to be used by many thoughtful partnerships --- and abused by a whole lot more.

[mycroft]

 johnu, on 2014-December-09, 20:14, said:

Then design a system that gives you 65%

Rolleyes is right. As a system wonk, in the end, bridge is a game of card play, and I admit that system can't get you 65% (at least not without sufficient card play to make the better contracts, at which point...)

Quote

If you start with a system that gets you to higher percentage contracts, superior card play will only increase your advantage. If you voluntarily throw away your advantage by playing the same system as hoi polloi, you are decreasing the gap between you and the average/below average players.

Not necessarily. Look at it this way:

If I play the field contract, the field-way up, my expectation is 60ish% (it's not, of course, but pretend I can play).
If I play the field contract, the wrong way up, some large percentage of the time it doesn't matter. When it does matter, it's random whether I'm behind or ahead. When I'm behind, I expect to lose 60ish% with my zero, or 40ish if my superior play gets me back to average. When I'm ahead, I expect to win <40% with my top. So my randoms are a net loss that the system has to gain back for me (and when we're talking weak NT v strong NT, that's a *lot* of randoms - most openers with 11-17 sort of balanced will end up played the other way up or with much different information available to opps)
If my system gets me to a superior contract, again, it either works or it doesn't. If it works, my card play doesn't matter, but again I'm gaining <40% (say I find the slam. If it requires superior card play to make, how many more MPs am I getting for it than for 4+2?) But if it doesn't work, my card play again doesn't matter, but I'm losing 60ish%! So the system wins have to actually pay off more than 60% of the time - not just "higher percentage contracts" - just to break even. Now add the extras you need to recover from the random system losses (and no matter how superior your system is, there will be random system losses), above, and you might need to build a 70% system to not lose against playing down the middle and letting card play get you to your expected 60%.

It's the tradeoff of a high-variance system, whether that variance is built into the system, or whether it's just different - even if you stack the deck in your favour, when you put your results in the hands of the Card Gods, sometimes the draws end up against you. And when they do, your card play "advantage" goes away - at best, it limits the losses.

[johnu]

 mycroft, on 2014-December-10, 13:04, said:

Rolleyes is right. As a system wonk, in the end, bridge is a game of card play, and I admit that system can't get you 65% (at least not without sufficient card play to make the better contracts, at which point...)

Not necessarily. Look at it this way:

If I play the field contract, the field-way up, my expectation is 60ish% (it's not, of course, but pretend I can play).
If I play the field contract, the wrong way up, some large percentage of the time it doesn't matter. When it does matter, it's random whether I'm behind or ahead. When I'm behind, I expect to lose 60ish% with my zero, or 40ish if my superior play gets me back to average. When I'm ahead, I expect to win <40% with my top. So my randoms are a net loss that the system has to gain back for me (and when we're talking weak NT v strong NT, that's a *lot* of randoms - most openers with 11-17 sort of balanced will end up played the other way up or with much different information available to opps)
If my system gets me to a superior contract, again, it either works or it doesn't. If it works, my card play doesn't matter, but again I'm gaining <40% (say I find the slam. If it requires superior card play to make, how many more MPs am I getting for it than for 4+2?) But if it doesn't work, my card play again doesn't matter, but I'm losing 60ish%! So the system wins have to actually pay off more than 60% of the time - not just "higher percentage contracts" - just to break even. Now add the extras you need to recover from the random system losses (and no matter how superior your system is, there will be random system losses), above, and you might need to build a 70% system to not lose against playing down the middle and letting card play get you to your expected 60%.

It's the tradeoff of a high-variance system, whether that variance is built into the system, or whether it's just different - even if you stack the deck in your favour, when you put your results in the hands of the Card Gods, sometimes the draws end up against you. And when they do, your card play "advantage" goes away - at best, it limits the losses.

??? Having a better bidding system also includes playing the contract from the most advantageous side, not giving the defense a blueprint for defending when it matters, etc. It makes zero sense to say that you got to a 60% contract but only average 50% or less in real life. I call that a 50% expectation, not a 60% expectation. Don't be distracted by variance. Sure, for any given hand a terrible contract can turn out great and a great contract can end up with a zero, but over 100 or 1000 hands, higher expectation systems will come out ahead. If you don't come out ahead in the long run, either you are very unlucky or the valuation of your system was incorrect.

That's not to say you can't win in the short run with a high variance, below average expectation. Everyone has seen a team of bad players beat a team of world class players in a short swiss match or even a short knockout match, or a pair of bad players give a couple of zeros to a pair of world beaters in match points.

[akwoo]

Let's try a slightly different example, johnu:

For simplification, let's suppose that, when you get to a better contract than the field, you get 100%, and when you get to a worse contract, you get 0%. (If you don't like these numbers, put in 90% and 20% for a good player and 85%/15% for an average one. We can tweak everything else sufficiently to make it work.)

Let's say that you are playing a system where

25% of the time you get to a better contract than the field
60% of the time you get to the same contract
15% of the time you get to a worse contract

That's a pretty good system, right?

Well, if you're playing just as well as the field, it's a good system - you score .25+.30 = 55% on average (or 53.5% for 85/50/15).

Let's suppose, however, that you play much better than the field, and when you end up at the same contract, you get 70%.

Then your result playing this system is .25+.42 = 67%. (At 90/70/20, it becomes .225 + .42 + .03 = 67.5%)

Playing this system instead of the field system has now reduced your expected score from 70% to 67%!

[mycroft]

What I'm trying to explain is that because your expectation is > 50% already, your system has to do better than better than that to beat "play down the middle". You can, in fact, lose in the long run with a high variance, above average expectation system - if you're sufficiently above average expectation already.

It's a standard matchpoint decision:
- What is the chance that X is right over where you are now, and
- What is the expected win if you're right, vs the expected loss if you're wrong.

Throw away all the "doesn't matters" hands.
Here, the expected win from getting to the right contract, played the right way, giving the opponents different information that helps rather than hinders you, is <50% (not quite 100% - 50+%). The expected loss from getting to the right contract (that happens to be wrong this time), or played the right way (which happens to be wrong this time), or in the process giving the opponents information that helps them - is >50% (not quite 0% - 50+%). Sure, you get more smaller wins than your larger losses (or you wouldn't be in the right spot), but is it enough to get to what you'd get by playing down the middle? Note that in addition to those, there will be hands where you're forced into the wrong contract by system (because more often than not your system gets you to the righter contract, but this is one of the other hands), which will also win sometimes (< 50% of the time, for + <50% of the MPs) and lose sometimes (> 50% of the time, for - >50% of the MPs). Your card play advantage will eat into the losses and add to the gains, as will others that field-protect for you, but not enough to even balance that, never mind get it to 60-when-you're-right, 40-when-you're-wrong.

For me, the 50% cardplayer (shhh), it's a totally different question. If my system is 60% on the field, then I get 60% over the year. It'll be a bit pajama, and the variance will be high, but it will be well to my advantage. In addition, because I'm presenting the good card players with problems that a) the rest aren't getting, and b) they're not as familiar with as either I am, or as they are with "normal", they're likely not going to be as good card players as the usually are. That's not worth much (certainly not as much as the "they only play this to confuse us with their unfamiliar systems" people believe), but it is worth something.

Or as I said to the highest masterpoint holder in our unit, when he asked "but [EHAA] totally randomizes the results. Why would you want to do that?": "Say I flip a coin twice for all the matchpoints on these two boards; my expectation is 50%. Tell me that's less than what I would expect against you playing straight."

Question, just for the amusement:
You know that 6♣ is mildly odds-on (say, it's "9 without the Q" and some chances, so 55, 56%). You also know that the field, because of their system, is never going to even sniff at it - they'll all be in 3NT, which should be close to 100%. Do you bid it?

Now I tell you that you are a seeded pair in a field of 75 tables in a city where the sectionals draw 45, including a few from out of town. Still bidding it? Or are you going to bid 3NT with the rest of the room, knowing you'll take all the tricks and be solidly A+?

Does it make a difference if I tell you that while top on most boards is 50, this board is only going to be played at 16 tables, and your score is going to be factored from a 15 top to 50?

Does it make a difference if I tell you that the auction's at 3♥ now, and you could still be off two fast tricks in a suit (but the slam's about 75% if you're not), and if you try to find out, you'll lose the chance to play 3NT (and will sit in 5♣ for a clear, if potentially shared, bottom)?

These are the boards that are wins for your "better system"! And I know this all too well, having played Precision with 14-16 NTs and lots-a-gadgets for many years. These were the boards that greyed my hair - "I know this contract is better, but is it better enough to take the zero if this time it's wrong?"

[Edit: Aaaaand, while I'm typing, akwoo says what I am saying in a third the verbiage with double the comprehension.]

[Cthulhu D]

Isn't this basically a justification for serious events to be scored Cross Imps? This analysis only applies at matchpoints, due to the weird equating of 'bidding a slam' to 'getting 1 extra overtrick'. I think basically every serious event around here is scored X-imps (or butlered imps or similar), so obviously I've never really considered the field system, it only matters on random club nights.

[aguahombre]

 Cthulhu D, on 2014-December-11, 05:40, said:

Isn't this basically a justification for serious events to be scored Cross Imps? This analysis only applies at matchpoints, due to the weird equating of 'bidding a slam' to 'getting 1 extra overtrick'. I think basically every serious event around here is scored X-imps (or butlered imps or similar), so obviously I've never really considered the field system, it only matters on random club nights.

Yes, the weird idea of an event where your placement depends on scoring better than other pairs, regardless of numerically how much better --- with every board being equal to every other board regardless of whether it is a part-score battle or a slam hand --- has been around for much too long. Good thing they aren't considered to be serious events.

[Cthulhu D]

 aguahombre, on 2014-December-11, 06:10, said:

Yes, the weird idea of an event where your placement depends on scoring better than other pairs, regardless of numerically how much better --- with every board being equal to every other board regardless of whether it is a part-score battle or a slam hand --- has been around for much too long. Good thing they aren't considered to be serious events.

Mhmm, sarcasm. I know it's popular in America, but it *is* a very weird feature of the scoring system that you are encouraged to play methods that you think are worse, because that results in a better expected score.

[helene_t]

(Must resist the temptation to rant about IMPs vs MPs - would be too much of a threaddrift)

Anyway, either method is obviously inferior to T-Walsh.