[straube]

Picked up Better Balanced Bidding which was co-written by Ron Klinger. In it, they describe Banzai points which are claimed to be better than Goren when both hands are balanced.

Banzai pts

A=5
K=4
Q=3
J=2
T=1

I found it easier to add Goren pts and then add 1 pt for every honor (including the ten) at the end.

To convert Banzai pts to Goren pts, multiply by 2/3.

Some strange conclusions.

They might pass Axxx Axxx Axx xx because it only has 15 Banzai hcps which equals 10 Goren pts. They might open with QJTx QJTx KJx xx because it has 18 Banzai hcps which equals 12 Goren pts.

They gave lots of examples how it might be better. I was partly persuaded (especially for reaching 3N) but it seemed like Banzai pts over-corrected.

Thoughts? Anyone use Banzai pts?

[whereagles]

If there's one thing the 4321 scale is good for is when both hands are balanced... eheh. No need to mess it up.

[gwnn]

passing with three aces is nonsense (:

[Bbradley62]

So, the point of banzai points is to decrease the value we assign to A and K while increasing the value we assign to J and T?

[straube]

Bbradley62, on May 29 2010, 07:44 PM, said:

So, the point of banzai points is to decrease the value we assign to A and K while increasing the value we assign to J and T?

They say that "4-3-2-1 overvalues Aces and Kings in relation to Queens, Jacks, and Tens in balanced hands."

They give this real hand...

KTxx Qx Axx KJ98 opposite QJx AJTx T9xx Qx

Meckwell bid to 1N while users of Banzai pts bid to 3N.

Banzai evaluates as 19 pts + 17=36 or the equivalent of 24 Goren pts (i.e. a slightly aggressive 3N)

They give a constructed deal...

A9xx Axx Kxx Axx opposite Kx Kxx Axxx xxxx

Banzai evaluates as 19 pts + 13 pts=32 or the equivalent of 21.7 Goren pts. They feel that the first hand is not worth a strong NT but note that the field will be in 3N on this deal.

[rogerclee]

Goren + judgment > Banzai

I think it would not be that hard to argue Goren > Banzai, but I am very confident in the first claim.

[cherdanno]

Goren > Banzai.

[matmat]

Goren played bridge. Did Mr. Banzai?

[Siegmund]

Thomas Andrews has an article about 54321 points. Called them Cowan points, or something like that.

There was a case that, in a certain limited set of circumstances concerning balanced hands that are committed to NT, the 54321 approach is good. There isn't much of a case for using an extremely notrump-only hand evaluation method for the opening bid, even for opening bids of notrump.

[straube]

Siegmund, on May 30 2010, 12:35 AM, said:

Thomas Andrews has an article about 54321 points. Called them Cowan points, or something like that.

There was a case that, in a certain limited set of circumstances concerning balanced hands that are committed to NT, the 54321 approach is good. There isn't much of a case for using an extremely notrump-only hand evaluation method for the opening bid, even for opening bids of notrump.

That makes a lot of sense. When you open a balanced hand you don't know whether partner is balanced or not. Three aces goes with just about anything. I'm thinking the authors make a mistake using Banzai points to open, but perhaps it has some usefulness for hand evaluation opposite an opening hand that is known to be balanced. They derived these points looking at only 4432 and 4333 hands.

The appendix gives some rationale for how they determined the point values. Someone ran hands for numbers of tricks and then did a least squares analysis for 5 variables (the 5 top honors). They could have picked any number of variables. For the 5, they came up with 5, 3.97, 3.06, 1.93 and 0.95. I'm a bit concerned just looking at how easily these numbers can be rounded...almost like they were engineered.

[jjbrr]

taking tricks is where the money's at

[Cascade]

Siegmund, on May 30 2010, 05:35 PM, said:

Thomas Andrews has an article about 54321 points. Called them Cowan points, or something like that.

There was a case that, in a certain limited set of circumstances concerning balanced hands that are committed to NT, the 54321 approach is good. There isn't much of a case for using an extremely notrump-only hand evaluation method for the opening bid, even for opening bids of notrump.

Richard Cowan is a mathematician/statistician at the University of Sydney. Possibly retired now.

You can see a summary of his original analysis here

54321 count

[kenrexford]

I have always felt that notrump game contracts fall into one of two main categories. I call them T and Cloud.

A "T" notrump contract is one with a long suit to establish (the stem of the T) and stoppers in the various suits (the top of the T). 4321 seems to under-guess a T 3NT, but adjustments help. Looking for long suit adjustments and controls adjustments.

A Cloud notrump is one where you end up with 9 before they end up with 5, but a lot of work needs to be done, in a lot of suits. These contracts often can be set double dummy, and these sometimes fail if you miss the line. Banzai Points might in theory better handle Cloud notrump contracts. Not sure.

[gwnn]

straube, on May 30 2010, 06:05 AM, said:

The appendix gives some rationale for how they determined the point values. Someone ran hands for numbers of tricks and then did a least squares analysis for 5 variables (the 5 top honors). They could have picked any number of variables. For the 5, they came up with 5, 3.97, 3.06, 1.93 and 0.95. I'm a bit concerned just looking at how easily these numbers can be rounded...almost like they were engineered.

1.03
0.91
1.13
0.98
0.95

For the additional utility (I know there's a neat word for this in economics...) of each honour. I feel pretty surely that this should be a monotonous series, not something that has been arbitrarily perturbed from a bunch of ones.

[ONEferBRID]

I ran across this one sometime ago:

A = 4.5
K = 3.25
Q = 1.75
J  =  0.5
       10
- - - - - - - - - - - - - - - - - - - - - -
So...for example, that relatively balanced 18 point hand that I posted in the 2/1 forum ( Rebid --Part 1 ) would upgrade to 20 ( 3 Aces and 2 Kings ) to help "justify" a 2NT opening.

[straube]

Cascade, on May 30 2010, 04:42 AM, said:

Siegmund, on May 30 2010, 05:35 PM, said:

Thomas Andrews has an article about 54321 points. Called them Cowan points, or something like that.

There was a case that, in a certain limited set of circumstances concerning balanced hands that are committed to NT, the 54321 approach is good. There isn't much of a case for using an extremely notrump-only hand evaluation method for the opening bid, even for opening bids of notrump.

Richard Cowan is a mathematician/statistician at the University of Sydney. Possibly retired now.

You can see a summary of his original analysis here

54321 count

I think his study is flawed. He broke the problem into looking at trick expectation for individual suits. That likely doesn't take into account the frequency that 2-2 vs 3-2 vs 4-4 etc actually occur. It doesn't take into account defensive help. It undervalues aces for the control they give; we're not playing each suit one at a time...we're trying to play our suits and not their suits. It doesn't take into account that honors can be guards...unable to take tricks if we lead the suit but able to take tricks if the opponents do so.

He ought to instead have looked at actual deals, recorded how many tricks were taken and then solved for the top 5 honors.

[3for3]

It is like real estate, location*3

Jacks and Tens in insolation are of dubious value. Add them to a Queen and they are clearly more value than Goren.

ATx
Jxxx
KQx
AQx

Versus

Axx
QJTx
KQx
Axx

Surely the second hand isfar better than the first.

In other words, J/T are more context dependent than the higher honors.

Danny

[babalu1997]

all counting methods start with aset of hands which justtify their creation

like ltc, then there are adjustment

like ltt for contested auctions, then there are adjustments

like ahen, goren points, then there are adjustments

points, schmoints

use your brain space for other things

[NickRW]

Constructed examples prove pretty much nothing. 54321 will work well enough for you in *some* low level NT contracts - the "cloud" type that Ken mentions, but it is diabolical for higher level and suit contracts.

Nick

[Cascade]

straube, on May 31 2010, 04:02 AM, said:

Cascade, on May 30 2010, 04:42 AM, said:

Siegmund, on May 30 2010, 05:35 PM, said:

Thomas Andrews has an article about 54321 points. Called them Cowan points, or something like that.

There was a case that, in a certain limited set of circumstances concerning balanced hands that are committed to NT, the 54321 approach is good. There isn't much of a case for using an extremely notrump-only hand evaluation method for the opening bid, even for opening bids of notrump.

Richard Cowan is a mathematician/statistician at the University of Sydney. Possibly retired now.

You can see a summary of his original analysis here

54321 count

I think his study is flawed. He broke the problem into looking at trick expectation for individual suits. That likely doesn't take into account the frequency that 2-2 vs 3-2 vs 4-4 etc actually occur. It doesn't take into account defensive help. It undervalues aces for the control they give; we're not playing each suit one at a time...we're trying to play our suits and not their suits. It doesn't take into account that honors can be guards...unable to take tricks if we lead the suit but able to take tricks if the opponents do so.

He ought to instead have looked at actual deals, recorded how many tricks were taken and then solved for the top 5 honors.

I am not sure what was done but he may well have taken into account the relative frequencies of those suit lengths:

"I found the best values for the honour cards by a weighted least squares analysis" - my emphasis.