How far do you trust Bird & Anthias?

[MrAce]

Fwiw, the notes by coaches of Turkish national team (Alan Levy - Karen Mc Callum - Erick Kokish - Krzysztof Martens) doubleton leads were forbidden by most of them, let alone stiff vs NT contracts. Other "no no" leads vs NT were, if i remember correctly by Alan Levy, AJxx, Axxx.

 

I think people are making huge deal of shortness leads by GIB. If you check the GIB carding, it leads smallest from xxxx and smallest from xxx. It almost never leads small from Hxx(x) unless it is the winning lead or unless it has no other option. And pd GIB always knows when it led x from xxx and when from Hxxx. A human pair can not work this out always without coughs or some other side method. People forget easily the contracts GIB gives away on the lead and always remember the leads that works out very well, after a double dummy defense. Don't forget, a human pair, if agreed to lead their shortest major frequently, will have to disclose it to the declarer (i dunno the recent laws but i guess they will have to disclose it if their pd knows this is the tendency of the pdship) Knowing that a pair is leading always their shortest major, can give declarer a huge advantage. A human pair will lead the highest spot from xxx or mud at least. Never smallest. This is unplayable without losing a lot on some other hands imho.

 

I like passive leads, but leading a ♥ would not even occur to me, unless i am seeking a spot on the daily bulletin. (which is quite understandable, if you are leading a ♥ on this hand, you are unlikely to win anything, so why not try to make it to bulletin http://www.bridgebase.com/forums/public/style_emoticons/default/tongue.gif) I do not mind ♠ 8 lead, but personally I would lead a club. People open with 5 card major, even with 5-4-2-2 hands a lot. And responder sometimes skips the stayman and bids 3 NT. Of course all the things i wrote are for IMP scoring. At MP I would still never lead a ♥ but I would proabably lean towards spade 8 a little more.

 

I don't know who is Bird and who is Anthias. But i know Helgemo, Meckwell,Weinstein, Hamman, Versace, Lauria, Forrester and all other top players and their achievements and I watch them frequently on BBO VG every national and BB and EC. I haven't seen them leading their doubleton or singleton vs NT in this type of auctions, even when they have no hcps in their hands. Perhaps they do now and then, but in general they do not unless they smell something burning.

[Vampyr]

Why do so many believe that passive leads require better defense?

 

It's not so much that; it is just that after making a B&A lead, you must first, if you are leading a major, hit on the right one, and then defend double-dummy for the rest of the hand. I don't expect often to have enough knowledge of the unseen hands to defend double-dummy, and as for the ox opposite...

[mike777]

Why do so many believe that passive leads require better defense?

I understand if passive leads at notrumps are shocking for you, changing habits require some time.

I for my part have given up fourth best leads, because I have my doubts whether they help the defense more than declarer.

I simply lead attitude.

So if I lead a high spot card, partner knows I am not interested. If I lead a middle one I don't know what is best, and if I lead lowest I am serious.

Simple and often effective and declarer can not tell whether I led lowest from four (because no switch looked attractive to me) or five or six and I may not lead lowest from 5 or more if a switch in another suit looks attractive.

When I lead high and do not hit partner's suit, which does happen surprisingly often (confirmed by the book mentioned here), declarer will usually win and start establishing his tricks.

So at most 2 candidate suits usually remain to switch to when partner wins a trick.

Is a halfway intelligent partner looking at his hand and dummy and what he knows about declarers hand so likely to misguess which suit he should try next?

Even if partner wins the first trick and needs to switch at trick two, which is obvious most of the time, he has a high chance to get this right.

What is important is that your lead agreements show your interest in the suit led and partner can distinguish an 8 spot from a 2 spot and this is not even taking into account there is a signalling convention called Smith-Peter.

Not so hard for those past the beginner stage.

 

Rainer Herrmann

 

Rainer thank you for taking the time to respond in a very thoughtful way.

 

 

My answer is your first line says it all.

Your second line says it fine.

 

 

I mean my answer to be most respectful and my answer is the counter argument.

My answer is look at the facts, the evidence that is counter to your very strong pointsl

 

----

 

 

Perhaps I am not clear.

1) you make a powerful argument.

2) now make an argument against it.

3) I find these types of discussions much more powerful

[Jinksy]

On p 70 of chapter 4 ,"Leading from a weak hand," Bird & Anthias. Winning Notrump Leads. they stated their conclusion as "From a weak hand it is generally right to lead your shorter major, whether or not it includes an honor." I feel that this is a good precept.

 

I have the book next to me, and it matches my memory - it's frustratingly light on test cases for this question. Of the hands they discuss in the chapter (inc quiz hands), very few of them compare two <4 card major suits, and of those that do only one has as much as a king outside the suits.

 

On that hand, (#5 from the quiz), the 3-card major slightly outperforms the 2-card one, but unfortunately is headed by JT, so it's hard to tell whether the honours or suit length contribute more.

 

It makes a lot of sense to me that on slightly stronger hands, a longer-short major would work though - partly because with greater expected partnership strength you don't necessarily need to set up a 5 card suit so it will do better just by being passive, but mostly because if opener has such as Kxx or Ax in the suit, he'll only be able to hold up once.

 

Then if you've led from a doubleton successfully (ie hit P's suit), declarer will be able to put you in safely if he needs to set up his side suit, but if you've led from a tripleton successfully, neither hand is safe.

[FM75]

I have been studying data science of late. As a result I am far less impressed by Bird-Anthias than I was after their first book. While I believe that some of what is in there makes sense, had they been data scientists they would have published their results in a fashion that their research would be reproducible.

 

Their second book was even worse, with clear errors every few pages. The publication of the second book was horribly sloppy. Does that suggest that their research was as well. Frankly, I have to say 'yes'.

 

I tried reproducing some of just the statistics from the first book, not even including the double dummy analysis. It was hard to match - and of course not reproducible, because they do not give the seeds used int their random number generators. They also do not give the specific rules for point count, distribution, etc.

 

Their research would not warrant publication by any data science publication - and they would be broadly criticized for the unreproducible results.

[whereagles]

FM: you are demanding from the book the same level of detail of a research article. They are not the same thing.

 

And reproducing results doesn't require using the same seeds. If the results from different seeds are the same up to a certain confidence level, stuff has basically been reproduced.

[Jinksy]

FM: you are demanding from the book the same level of detail of a research article. They are not the same thing.

 

No, but the higher the precision in either the better, right?

 

That said, I'll take B&A's data against anyone's instinct unless that instinct is backed up by comparable data.

[rhm]

I have been studying data science of late. As a result I am far less impressed by Bird-Anthias than I was after their first book. While I believe that some of what is in there makes sense, had they been data scientists they would have published their results in a fashion that their research would be reproducible.

 

Their second book was even worse, with clear errors every few pages. The publication of the second book was horribly sloppy. Does that suggest that their research was as well. Frankly, I have to say 'yes'.

 

I tried reproducing some of just the statistics from the first book, not even including the double dummy analysis. It was hard to match - and of course not reproducible, because they do not give the seeds used int their random number generators. They also do not give the specific rules for point count, distribution, etc.

 

Their research would not warrant publication by any data science publication - and they would be broadly criticized for the unreproducible results.

What they did not do is to state their precise assumptions about the unseen hands they made for the different bidding sequences used in the book.

But the book contains a very large number of simulations and I guess the average reader would have got bored by the technical specifications.

But at the end of the second book they explained their method in general and also what they did to prove that their results are robust.

For example whether a single major suit raise shows 5-9 or 6-10 does not affect the opening lead results substantially can be shown by redoing the simulations under the differences.

 

Their general conclusions agree with my experience about simulations.

 

There are a number of print errors in the book.

I understand that nobody gets rich (except maybe Culbertson) from writing Bridge books and the investment costs like proof reading must somehow reflect the sales potential.

There are very few Bridge books on the market where I have not found errors.

 

If you found a marked difference in your own simulation tell me the hand and the bidding and I will do a simulation myself.

 

Rainer Herrmann

[whereagles]

No, but the higher the precision in either the better, right?

 

In theory yes, in practice only if the reader is interested in it :)

[inquiry]

I don't have it, nor have I seen a review, but this new program from the guy that brought us "double dummy solver" might be useful in this discussion. See

 

http://www.bridgecaptain.com/LeadCaptain.html

 

I am not100% sure it is available yet, either. But it would be interesting to compare with other Sims when it is.

 

 

 

 

 

 

 

 

[SteveMoe]

I find the ♠ lead appealing and the ♥ lead not so much.

By leading a short major we are trying to hit partner's 5-card suit. (Yes other factors could also lead to success).

Leading from a doubleton almost requires the opponents hold 3-3 in that suit (or partner to hold 6+ with 10 HCP - If partner had a chance to bid over 1NT they would have with that Yes, I know that does not apply here). So, the 3-card major holding plays to the required assumption that partner can clear the suit with one loser and get back in later.

 

(PS - Nice link, Inquiry. Thanks! Sims are so touchy - not sure I am comfortable with the idea of a black box approach, but the tool if it works will be helpful)

[mcphee]

Not being enough of a genius to know better I lead 4th best C and see ZERO reasons to pick any other card. Partner may have Jxx and will be able to return the suit on gaining the lead. With my trash it is highly ppossible partner will get in before they have extablished 9 tricks.

[BillPatch]

I ran a double dummy simulation of 1000 on Jack and the race is too close to call between clubs, spades, and hearts.

defeats contract:

S 10 11.4%

S 8 11.6%

S 3 11.2%

H 10.9%

low C 11.7%

I didn't record A or 10 of clubs.

[jallerton]

I ran a double dummy simulation of 1000 on Jack and the race is too close to call between clubs, spades, and hearts.

defeats contract:

S 10 11.4%

S 8 11.6%

S 3 11.2%

H 10.9%

low C 11.7%

I didn't record A or 10 of clubs.

 

It will be a lot less close in practice, when partner does not know all four hands and may not be able to read that you have led from a short suit. Some of the time when a major suit lead would beat the contract, partner would have doubled for the lead. A club lead is clear cut in my view.

[lamford]

It will be a lot less close in practice, when partner does not know all four hands and may not be able to read that you have led from a short suit. Some of the time when a major suit lead would beat the contract, partner would have doubled for the lead. A club lead is clear cut in my view.

I agree with the club lead, and recall a Woolsey article where he thought that "three apart" leads were bad, so I might lead a spades if my clubs were AJxxx, KTxxx or Q9xxx. I cannot find that article now, but did a test of it once, where I agreed with the findings. I think the double-dummy test is fair, as it is not clear whether that favours the declarer or defender. I would expect that leading a five-card minor instead of a three-card major would gain, and it probably matters whether the responder would bid Stayman with a four card major and 4333 shape or even 4432 with one major. So, the parameters you set for the 3NT (and the frequency of a five-card major in 1NT) make a huge difference.

[BillPatch]

By the way, I was using Jack6 and its version of BWS2001. On the default settings and also at the world championship level settings, Jack when asked(Ask Jack)chose the poor 8 of diamonds for the opening lead. Assigning maximum time and power to the query, Jack chose the 6 of clubs.