[thepossum]

Hi all

I know a few on this forum have advised that sometimes my bidding gives too much away. Sometimes I think it is swings and roundabouts in that by having informative auctions you often get to a better contract but you do sometimes risk giving too much information to defenders. I dont know what the statistics are but given that this is a beginner forum and when I learned I learned to describe my hand to give partner the chance of best finding a fit.

However paraphrasing a quote I believe Zia Mahmood used I'm wondering about the philosophy of "always bidding No Trumps and punishing the opposition", or at least bidding No Trumps more than I do on certain opening hands. And then on rebids not giving too much away - ie after stayman, transfers and other bids, depending on circumstance trying to jump straight to the required NT or suit game. Unless of course further exploration is called for.

The following hand is an example from a recent robot tournament. I know partner and opp were GIB but after analysing the leads the traditional lead from longest suit is the only one that brings 3nt down 1 trick. Any other lead allows overtricks.

So of course, after my auction and others who went down 1 trick, we nearly always showed diamonds on this hand either via 1D opener or as a jump bid after stayman. A regular stayman 2D response would have been ok. Those of us who showed our diamonds unfortunately received the best club lead. Everyone else who avoided mentioning diamonds received a diamond lead making a difference of 3 tricks

I dont think I will mention diamonds like that if we are planning to end up in NT

regards P

---

[smerriman]

3♦ is not a possible response to Stayman, despite the fact that GIB's description of it matches your hand. It describes a lot of impossible bids by trying to piece together different meanings.

If my human partner responded 3♦, I would assume they didn't know how Stayman worked. (This isn't an attack, and you are perfectly within your rights to make such a bid with GIB; I'm saying you shouldn't make a bid just because of the description; you should make a bid based on how the system works / how you were taught Stayman).

Ignoring that option, I would open 1♦ not because I wanted to show my diamonds, but because I would value the hand as worth 18 points, and thus is too strong to open 1NT, and the way to show a balanced 18 points is to open 1♦ and rebid 2NT.

The fact this causes GIB's random simulations to pick a different lead is random. Next time those that opened 1NT will get the bad lead; or more likely, those that open 1NT won't be in game and I'll be making 3NT.

[thepossum]

Ok thx for reassuring me that I should not have followed GIBs description. I thought that must be an option other than regular Stayman. I never bid it before. Wish I'd never seen it

Edit. PS I'm curious about when to adjust a no Trump bid up or down. Sometimes I do. I knew this was a strong hand but it had 15-17 hcp. Sometimes I have added an extra point and opened 1nt on 14 strong points or 2nt on 19 strong points. I've also been reading about different attitudes to balanced hands and whether a 2353 hand is balanced enough for NT opener. Also if you have a bad 5 card major in balanced hand if NT is ok. I know that maybe more an Acol weak NT thing

[thepossum]

Deleted duplicate post

[apollo1201]

You have also to know that GIB doesn’t always lead like humans and often picks a passive / short suit based on a simulation. Most humans with the W hand and their presumed 5 clubs would lead the suit and doom your contract.

[thepossum]

Yes, I know GIB often leads passively based on some book. I was raising a few points about information with bots/humans and also philosophy of leads. Yes, W had 5 clubs

[Tramticket]

As already pointed out, 3♦ is really not a response to Stayman. Imagine North has a Yarborough:

---

The auction shown is perfectly compatible with North's holding and would be a possible outcome with any human partner.

Your other question concerned the valuation of the hand for NT purposes. There are some good books out there on hand valuation (Brian Senior's book has been around for years and seems to have passed the test of time).

On this hand, there is one significant negative feature: the ♠J2 holding is really not pulling much weight and I would downgrade for that. But there are some very significant positive features: A 5332 shape tends to play significantly better than 4333 or 4432, The five cards suit has two top honours and good intermediates - this is a very positive feature. In clubs, the ten and nine supporting the ace don't need much from partner to give two tricks (say JX gives a 75% chance of two tricks!). The hand also has three aces. Overall, my judgement is that the positive features are significantly greater than the negative features and I would judge this to be a good 17 count, which you might upgrade out of the 1NT range (depending partly on partnership style).

[apollo1201]

Tramticket, on 2019-January-21, 08:35, said:

As already pointed out, 3♦ is really not a response to Stayman. Imagine North has a Yarborough:

---

The auction shown is perfectly compatible with North's holding purposes.

With this garbage you just transfer to S, don’t you? Maybe w/ C void could you just try a Stayman for fun. Or if you play gabrage Stayman, but GIB doesn’t. Anyway it’s also the first time I see this (ridiculous) 3D answer.

I’d also have upgraded, and I’d also go down in 3NT.

[pescetom]

apollo1201, on 2019-January-21, 12:49, said:

With this garbage you just transfer to S, don’t you? Maybe w/ C void could you just try a Stayman for fun. Or if you play gabrage Stayman, but GIB doesn’t. Anyway it’s also the first time I see this (ridiculous) 3D answer.

It depends upon the gadgets you have agreed. If you play crawling Stayman then the ♣ void is no problem, otherwise just transfer to ♠. I agree that 3♦ makes no sense.

[thepossum]

Ok, ignoring the invalid 3D response and the obligation to respond 2D I am curious about different auctions and uses of Stayman and Transfers with different types of hands.

Let's assume I play the 2/1 Card and rarely would open a strong NT with a 5 card major. Let's also assume I have decided to open 1NT rather than the alternative 1D

Responder on this hand has two possible majors and clearly enough to consider game so 2C is the best option followed by 2D, 2S. .

Suppose responder has no points with the same hand. I know a transfer is an option with a weak hand, but isn't pass also an option. Or would 2S always be best, even if I only have two spades in my NT hand.

I've always been interested in strength required for use of regular Stayman and Transfers with a strong NT. I haven't played alternative Stayman such as garbage. I think with weak NT it is probably better to always try to escape the NT with a weak responding hand

P

PS I am starting to use Smolen and bidding NT slightly more often with 5 card majors(with puppet). Depends on partner but that's a different issue

Also regarding robot leads and information. My understanding was that there was a combination of statistical analysis and rules for opening leads, not purely statistical.

[apollo1201]

thepossum, on 2019-January-21, 15:47, said:

Responder on this hand has two possible majors and clearly enough to consider game so 2C is the best option followed by 2D, 2S. .

Suppose responder has no points with the same hand. I know a transfer is an option with a weak hand, but isn't pass also an option. Or would 2S always be best, even if I only have two spades in my NT hand.

I've always been interested in strength required for use of regular Stayman and Transfers with a strong NT. I haven't played alternative Stayman such as garbage. I think with weak NT it is probably better to always try to escape the NT with a weak responding hand

As a note, possum, what follows is more or less standard after *string* NT, but alternate agreements exist.

Your 1st option is normal’after 1NT opening, showing invite with 5 or rarely 6 S. Often 4H on the side but could be an unbalanced hand eg 51(43) not strong enough to bid 3m and unsuitable for 2NT (more balanced usually) or 3M (6 cards). After this start, the prime 17 points and’ Jx in partner will probably propel opener to game (even after a non-standard 2NT inquiry for singleton).

And when responder is weak, yes, it is better to play 2M, even in a 5-2 fit, even requiring 1 more trick to make, than stay in 1NT.

Stayman usually requires invitational strength, unless trying sth special (weakfish three-suites short in C, as I mentionned). Transfer requires 0. No upper limit in both cases. Stayman is also priority if you can use it vs transfer (54 with 8+ points as in your example).

[thepossum]

Just by way of follow up in my comparison tables

Out of the approximately 160+ tables all but 9 were in 2 or 3 NT

Out of those in NT, approximately 125 received the favourable (but questionnable diamond lead) and approximately 25 received the unfavourable (best traditional NT lead. Almost all those who went down 1 trick all played the hand perfectly and scored around 11%. Not all those of the 125 played correct declarer play and received scores between 21-98%, not necessarily through additional skill. Correct declarer play following the diamond lead, which resulted in +2 received an 82% score (56 of the 125) and many made no overtricks and received a 44% score (47 out of 25). Note those who bid and made 2NT were a very small number and arrived there despite the diamond lead

So I feel, notwithstanding the issues raised on other forums regarding number of tables that there is a serious issue regarding the level of defence in tournaments. This was a forum daylong so many reading this forum will have seen the hand in discussion. So I do challenge the issue about accepting this as just a poor random outcome. Firstly it seems to be a poor lead under most expert advice on leads against no trumps. It also seems from my analysis of hands that those who received the clubs lead were more likely to have shown diamonds in the auction. This is fair enough. My issue is that I do not really accept that a passive lead from 3 ordinary diamonds should be accepted as ok when you are sitting there with QJxxx of clubs in that auction

EDIT. PS or would the auction (via stayman and other bids) and statistical analysis likely show that we had struggled to 3NT or may not make, in which case, from my reading by some experts passive is ok

PPS I also know over time statistical error tends to average out (maybe over a very long time) and we concentrate on the bad breaks rather than the good ones. However I am seriously interested in current theory about leads against no trumps, whether the advances in statistical analysis of bridge have changed older theories, or whether expert advice is backed up by statistical results. Also the philosophy of robot bridge. There are obviously two conflicting goals - one to produce the best robot bridge player in the world (which GIB used to be) and the other having robots that can play well with humans (ie behave in expected ways, more human, AI so to speak). I know this is more in scope for other forums so may raise it there. But if anyone has any comments on lead theory and current thinking I would be very interested on this forum.

regards P

[smerriman]

The number of times the favourable/unfavourable leads aren't really important - because of the way GIB is programmed to make the same lead in the same situation. That is, if you had 1 million people bidding one way, and 100 bidding another, the million would all receive the same lead - this doesn't meant it was 10000 times more likely; it may have been a 50:50 decision for GIB.

Regarding the lead being poor, you'll need to read https://www.amazon.c...d/dp/1554947596 which demonstrates the reasoning.

[thepossum]

Thanks everyone for your helpful comments

regards The P

[thepossum]

smerriman, on 2019-January-21, 18:53, said:

The number of times the favourable/unfavourable leads aren't really important - because of the way GIB is programmed to make the same lead in the same situation. That is, if you had 1 million people bidding one way, and 100 bidding another, the million would all receive the same lead - this doesn't meant it was 10000 times more likely; it may have been a 50:50 decision for GIB.

Regarding the lead being poor, you'll need to read https://www.amazon.c...d/dp/1554947596 which demonstrates the reasoning.

Thanks Stephen

regards Duncan

[blackshoe]

Marty Bergen has a nice little pamphlet out titled "Hand Evaluation". It's worth the ten bucks, IMO.

[thepossum]

I don't know if anyone is interested but in line with what everyone would expect I just ran a simulation (2000 accepted hands, 1877044 dealt) through bdeal(Piotr Beling) using just Wests hand and the following auction (1NT-2C-2D-2S-3NT) with a script and a few assumptions on points and distribution based on GIB descriptions - S semi balanced 15-17 (without 4 card major) and N with 4+ hearts and 5+spades and 7-11 points (I know thats a bit stronger than N was)

Not surprisingly the following are the results:

Any club lead had expected number of tricks approx 4 and probability of defeat 42-43%. Q and J were best leads, marginally better than 4th best
Next best were any diamond passive lead with expected approx 3.6 tricks and probability of defeat 28-29%
Third best was spades, approx 17-21% chance of defeat and approximately 3.4 tricks won. Note, don't lead the Ace
Worst lead was hearts with approx 19% chance of defeat and approximately 3.1 tricks won

Those are just rough figures summarised from the accurate simulation results and I have not tested what happens if South bids 3 diamonds on third round of auction which is what brought down some players

regards P

EDIT oops. forgot Wests hand. It was S A94 H KT D T32 C QJ652

[smerriman]

That sounds about right. The problem is, you need to simulate a *lot* of hands before the results reflect the long term average. If you only simulate 20 hands - and I don't think basic GIB comes anywhere near that - occasionally, the results won't be accurate. GIB then makes the wrong play.

Also, worth noting - even if a diamond lead was correct double dummy, I'd still prefer playing with a human who led the club. The "correct" lead assumes partner will know what to return - which he doesn't. If a human lead a club, I'd know what to return.