I fought the law

[keylime]

Gang, I don't use LOTT. LOTT, like Bergen raises, are NOT losing trick count (LTC) sufficient.

 

Give you a dramatic example of this. At one of Calgary's fine DBC's (the Aurora), my student and I defended a 3♥ contract via a Bergen sequence that went off three, doubled:

 

1H - P - 3C (constructive, 4 trumps)

3H - X! (GREAT DOUBLE partner) - P - P

 

Trumps were two-two, colors equal, and we STILL took +800. Why is that?

 

1. The hands were 3-5-3-2 opposite 3-4-3-3. Flat on flat.

 

2. The Bergen raiser had A TON of losers. For their 8 hcp hand they rewarded partner with a 9.5 loser hand opposite a 7.5 loser hand.

 

3. Partner found one great double.

 

4. I was leading trumps anyway (pard did too at every chance possible (twice)).

 

LOTT frankly isn't LTC sufficient. Isn't in many cases. Furthermore, that leads me to another question:

 

Over a Bergen raise setup, are trump leads effective as opening lead choice? I tend to think so.

[MickyB]

Keylime...how did your partner find that double?!

[Chamaco]

1. The hands were 3-5-3-2 opposite 3-4-3-3. Flat on flat.

 

2. The Bergen raiser had A TON of losers. For their 8 hcp hand they rewarded partner with a 9.5 loser hand opposite a 7.5 loser hand.

 

--- cut ---

 

LOTT frankly isn't LTC sufficient. Isn't in many cases.

The example you used seems constructed on purpose, but good LAW followers would not apply it that way:

 

1) opener should not have a 7.5 losers hand but better off guaranteeing 7 losers

 

2) 4333 hands should consider their support shorted , e.g. 4 card support with 4333 is considered 3 card support

 

3) LOTT and LTC *can* be put together ny chhosing to use the invitational 4 card support with 8 losers rather than using straight hcp count (and using mixed raise with 9 losers). The only guarantee is that opener should guarantee a no of loser not greater than 7.

 

Having said that, it is obvious that the type of disaster you mention should not occur.

 

More generally, it is very easy (and unfair in my opinion) to try to shed bad light on one theory by applying it mechanically in specific cases.

LOTT/ LTC and any other theory should be adjusted with the aid og judgment and commonsense.

And even then there will be times when it fails, just like other widely accepted evaluation tools :-)

[luke warm]

just because one believes they are efficient and uses bergen raises doesn't mean judgment is checked at the door, like a coat... a 4333 hand with 8 hcp doesn't qualify as a bergen constructive... i personally would raise to 2M with that, tho i admit i might give thought to raising (later in the auction) to 3M in certain circumstances

 

i've used bergen raises for a long time and never had that kind of horror story happen... the only theoretical (to my knowledge) failing is the lead double

[keylime]

My partner had a light takeout double with 10 hcps and she wasn't going to let 'em play three cheap. I was quite happy.

[Gerben47]

I think both Lawrence and Cohen have some good points but I think it is sad they are using statistics to make the same data say what they want.

 

Here's the deal:

 

LOTT is right on 38% of the deals, one trick more about 24% of the time, one trick less about 24% of the time, and then about 7% for each of the extremes.

 

Now Lawrence says: 38% is not very good.

and Cohen says: To make a bidding decision you need to take a 1 sided test (which is correct), so to make the correct decision you have an accuracy of 38+24+7 = 69%, that's pretty good!

 

Then Cohen shows some hands where the number of tricks stays the same, and Lawrence shows some hand where the number of tricks changes. Big deal. We know both exist.

[Chamaco]

I think both Lawrence and Cohen have some good points but I think it is sad they are using statistics to make the same data say what they want.

Ditto.

 

Blind belief in the mechanical application of the LOTT on each single deal is as ridiculous as discarding the LOTT it if it fails by +/- 1 TOTAL trick about 24+24% of the time.

 

But it is sad to see some credited authors play with numbers to claim one theory is wrong.

[mikestar]

The law is less accurate with larger numbers of trumps for one simple reason: all trumps are not created equal.

 

The eighth trump is quite valuable. Compare 4-4 or 5-3 to 4-3 or 5-2. The ninth trump is almost as valuable: less chance of bad breaks, more opportunites for ruffs. The tenth trump is worth something, but clearly not as much as the ninth. The eleventh trump is worth a little bit, but not as much as the tenth. The twelfth trump is worth very little extra. The only thing the thirteenth trump is good for is making sure they won't set your grand slam by ruffing the opening lead.

 

Of course, this assumes that the added trumps are divided between the partners: if we start with a 7-0 fit and give all the extra trumps to the long hand, that thirteenth trump is worth a lot more. But this is length value rather than trump value per se: Give the trumps to the short hand and you will not feel the difference between a 7-5 fit and a 7-6 fit, and odds are good that you won't feel the difference between 7-4 and 7-5.

 

Both Cohen and Lawrence are simplfying a larger, more complex truth about the value of trump length and distribution--they are simplifying it in different ways.

 

I take strong exception to Cohen's assertion that The Law is superior to expert judgement. I take equally strong exception to Lawrence's assertion that the Law is too unreliable. Until one has expert judgment, the Law beats the hell out of guessing.

[ArcLight]

This is a direct quote by Marty Bergen, from his excellent book, "More Points Schmoints", page 113 on the Law of Total Tricks:

 

There is no doubt that it is more accurate for competitive bidding than the judgement of the best player in the world.

 

Your trump length is far more important than distribution or HCP.

[fred]

Until one has expert judgment, the Law beats the hell out of guessing.

Agree, but I think that most people who blindly follow laws and rules never gain expert judgment - they don't need to because their laws and rules always provide them with an excuse if their actions don't work out (shades of Walter the Walrus again - "I had to double I had 18 points!").

 

In my opinion the best way to develop expert judgment is to get as much experience as possible trying to excercise your non-expert judgment. Look over your results carefully. Try to be objective. Ask experts what they think. Listen and think about their answers even if your instincts suggest they are crazy. See what you can learn.

 

Other useful practices include:

 

- Kibitz successful players (either on BBO or in person)

- Watch vugraph

- Read "The Master Solvers' Club" in The Bridge World magazine

 

Fred Gitelman

Bridge Base Inc.

www.bridgebase.com

[whereagles]

I can't stand much of vu-graph. I prefer being there and play myself :)

[luis]

In a recent travel I decided to buy one bridge book to handle the idle times at airports.

I had this book in my hands and finally decided not to buy it, instead I took the 2003 World Championship book that is to me like watching a movie of the championship.

The reason I didn't like the book is that I couldn't find anything constructive in it, we know the law doesn't work in many hands, we know it's a tool that you can use along with your judgement so what is the point in showing when it doesn't work, what's the point in stating numbers and percentages ? I'd have liked a book about when to detect competitive auctions where the law should not be used instead the book looked just like a compilation of things we already know marketed under a rebel title.

 

I agree 100% with Fred about judgement and one thing I learned is that a bridge player never stops developing his judgement the more you play the more you watch and the more you read more information you have to take better decisions.

[ArcLight]

>The reason I didn't like the book is that I couldn't find anything constructive in it, we know the law doesn't work in many hands, we know it's a tool that you can use along with your judgement so what is the point in showing when it doesn't work, what's the point in stating numbers and percentages ? I'd have liked a book about when to detect competitive auctions where the law should not be used instead the book looked just like a compilation of things we already know marketed under a rebel title.

 

The book discusses hand evaluation as well. Its not just an anti-LAW book. It shows how your hand increases in value in some cases, but not others. If it was just an anti-law book I dont think it would have as much value.

 

Since I dont have any Bridge Worlds, maybe I should buy the Master Solvers club CD.

 

How does one identify successful players on BBO? Look for World Class players?

(I was trying to Kib Ben the other day but one of the players never showed up, must of been scared of him!)

 

 

(Luis, you say you don't like the book. If I understand your post correctly, you didn't read the book, did you? If not, then perhaps you mean "I was not interested in the book because ...")

[luis]

I didn't read it, I glazed over every page at the bookstore, while my wife was trying to figure out different ways to kill me. So I think I had a good view of the contents and how they were developed to decide if I was going to like the book or not.

I understand it can have some words about hand evaluation but there were too many pages showing how the law doesn't work in several situations and I decided It wasn't wise to spen money in a book that had so many pages devoted to show something that we all know like trying to demonstrate something that doesn't need a demonstration. I estimated I would use about 30% of what the book had so it wasn't a good buy for me.

 

If you write a book about a 1NT structure you wouldn't spend a lot of pages showing how many different hands can open 1NT would you? :-)

[ArcLight]

>If you write a book about a 1NT structure you wouldn't spend a lot of pages showing how many different hands can open 1NT would you? :-)

 

Actually, you would be surprised!

 

1) I'd discuss opening 1NT with a 5 card major, many players wont do that.

 

2) I'd discuss opening 1NT with 5-4-2-2* hands where I might have a rebid problem if pard bids a higher ranking suit, and I dont want to reverse.

 

 

 

By 5-4-2-2 I dont mean 5♠-4♥-2♦-2♣, I just mean the pattern 5-4-2-2 regardless of suit.

[inquiry]

I didn't read it, I glazed over every page at the bookstore, while my wife was trying to figure out different ways to kill me. So I think I had a good view of the contents and how they were developed to decide if I was going to like the book or not.

I understand it can have some words about hand evaluation but there were too many pages showing how the law doesn't work in several situations and I decided It wasn't wise to spen money in a book that had so many pages devoted to show something that we all know like trying to demonstrate something that doesn't need a demonstration. I estimated I would use about 30% of what the book had so it wasn't a good buy for me.

 

If you write a book about a 1NT structure you wouldn't spend a lot of pages showing how many different hands can open 1NT would you? :-)

My feeling is this will be a cocktail table kind of book... You buy it (or borrow it), so you can discuss the pro's and con's of the various arguements. It is a given that the law is oftne wrong, ...We see plenty of examples of it being wrong everyday. So 30% to proving that is way, way, way too much. But I would be interested in reading their replacement fo the law and digesting their arguements in support of their differen view, which no doubt will not be right all the time either, but after all, nothing and nobody is right all the time.

 

Ben

[Chamaco]

I decided It wasn't wise to spen money in a book that had so many pages devoted to show something that we all know like trying to demonstrate something that doesn't need a demonstration.

Now that I have it at home, I can write about my impression of the book.

 

1) it gives a number of examples.

Maybe too much for an expert, but not too much for a non expert.

The more examples, the merrier, for a low-midlevel player.

 

2) the use of statistics is not documented in depth enough for testing whether an alternative hypothesis may work.

Basically the authors are "playing with the numbers" of statistics to demonstrate their hypothesis, without giving a chance to others to use the ame numbers for the contrary.

The numbers are given only is a summarized way, but the way they summarized them can be subject to criticism (see my post above related to the more detailed numbers given in Wirsgren 's web site).

So their method is only apparemtly "quantitative/objective".

 

3) Most of the concept given by Lawrence, except the SST+WP criterion, were already given in his previous books

 

4) Almost all of the hands Lawrence shows where LOTT fails, have associated MIRRORED DISTRIBUTION and in some cases duplication of honors, and in sme others 4333 shape.

These factors are mentioned by Cohen too.

The problem is that duplication of shape or strength is very difficult to diagnose in most sequences, and- no matter if you use LOTT-based system or SST+WP system or whatever, it will lead to a number of tricks lower than expected.

 

So, duplication of shape/strength is not a problem of LOTT only, but of all system.

And, 4333 hands are usually discounted and downgraded in total tricks even by LOTT-players.

So basically Lawrence is applying the LOTT in a biased way.

 

5) LOTT is not a system to estimate the number of tricks you can actually make; it tries to estimate TOTAL tricks (e.g. even if you go down, a better alternative opps contract could be cold).

WP+SST is a system to estimate the number of tricks one side can take, so it is oriented towards constructive bidding; SSP+WP in itself says nothing of the cost-benefit of taking a sacrifice.

So it makes no sense at all to try to compare the 2 methods: one method may justsay "we'll make 4H and not 5H", the other may say "who cares ? 4S for opps is cold", and they may be both right.

 

-------------------------------------------------------------------------------

 

My own final comment: it seems to me that most people criticizing harshly the LOTT had discussions with some "LOTT-BOT", using it automatically, without any adjustments.

This can be seen by the examples they bring:

- 4333 hands taking less tricks

- hands with duplication of shape/strenght taking less tricks

- hands very pure (double fit) or with a long running suit (squeeze) taking more tricks.

- some will just show how bad things can go if you bid LOTT-based with a lot of wasted values

 

But all these are well known minus/plus factor to people who do not use the Law mechanically, there is no need to say "Look at how things go bad without applying judgment".

every decent player knows or should know that.

 

I think it is fair to say that most expert do not need the Law, they have already grown their "pattern recognition" engine.

 

But, if you accept that 30% of the time 25 hcp count in misfit do not make 3NT, yet you bid it, you may also accept that 40 % of the time the LOTT predicts EXACTLY the no of total tricks, and about 20% is predicts one TOTAL trick less and 20% of the time 1 TOTAL trick more.

(Note that 1 total trick means about 1/2 a trick for each side)

 

These are pretty good numbers, documented by Anders Wirgren himself.

 

However, just as Milton Work point count should be viewed only as a starting point t asses the potential of the hand, LOTT seems to provide a good starting point in determining the right level of competing.

 

Sure, do not use the LAW blindly:

- Cohen's suggestion to "use adjustments" is just another way to say "apply your judgment": recognize wasted values, double fits, duplication of shape/strength, etc