[Dumoti]

So I stumbled across an article on Banzai Points. For those who are not aware, Banzai Points were developed in Australia by Richard Cowen. Basically, they are used for evaluating a hand for no trump purposes when the hands are 4-3-3-3 or 4-4-3-2 or some combination thereof.

Basically, the analysis revealed that an ace is worth about 5, a king about 3.97, a queen about 3.06, a jack about 1.93, and a ten about 0.95. So, as approximations, the Banzai Point Count recommends 5-4-3-2-1. I like it because it's both useful and simple... or so I thought.

As an initial trial, I thought that I would employ Banzai Points when I was responder, my partner had opened 1NT and my hand had the right shape. However, I still found it all confusing. My partner will naturally open a 15-17 NT. But how many Banzai Points is that? How many Banzai points would I need to invite or to bid game? If a standard point count has 40 points in a deck and a Banzai Count has 60, do I just multiply everything by 1.5? So you need 39 for game? Or how does that work exactly?

What are your thoughts?

[bravejason]

Initially, you can do a straight scaling to convert between the point count systems, but it will be off and you’ll have to experiment and practice with it to figure out the proper conversion factors, bearing in mind that each honor may have its own scaling factor.

Another approach is to deal hands and count the points both ways. If you do that enough, and perhaps run some analysis on it, you’ll figure out how 4321 translates into Banzai.

I’d definitely scour the Internet looking for a schedule of points. Presumably, someone has already figured out how many Banzai points are needed to bid and how many are needed for game and slam.

[smerriman]

bravejason, on 2018-July-09, 14:41, said:

Presumably, someone has already figured out how many Banzai points are needed to bid and how many are needed for game and slam.

[johnu]

I've never heard of Banzai points and can't really see the utility of using them. That being said, it would be better IMO to translate Banzai points to Work/Goren points by multiplying the Banzai point values by 2/3.

Card Banzai Work/Goren
A = 5 => 3.33 (3 1/3)
K = 4 => 2.67 (2 2/3)
Q = 3 => 2.0 (2)
J = 2 => 1.33 (1 1/3)
10 = 1 => .67 (2/3)

That way you can add your opponents values in Work/Goren points to yours to estimate partner's point count if applicable. This is also 1000% better IMO for disclosing your range to opponents who have never heard of Banzai points and have no idea what the equivalent of 22 Banzai points are in Work/Goren points. If you described to the opponents a hand as showing 30 Banzai points and couldn't give an equivalent Work/Goren point count, I would be inclined to give a procedural penalty to you.

[helene_t]

Banzai points are useless. It is quite easy to find out how honours should be evaluated for notrump contracts - simply do a regression analysis such as

E(tricks) ~ aces + kings + queens + jacks + tens

or

log(odds(3NT makes)) ~ aces + kings + queens + jacks + tens

You will find that traditional 4321 counts are fairly accurate, certainly much better than Banzai points.

For suit contracts it is more complicated, since it matters more if honours support each other and if they are in long or short suits. But in any case, for suit contracts the 4321 scale undervaluates aces, so Banzai points would be even worse.

[Dumoti]

Sorry that I didn't reply immediately. I posted, then life got in the way.

I do not agree that Banzai Points are useless. For example, opposite:

Kxx

Axx yields 2 tricks.
QJx also yields 2 tricks.

So it stands to reason that Axx is approximately equal to QJx for no trump purposes.

Yet Axx is 4 Work Points whereas QJx is only 3.

Smerriman, thank you for posting the chart. However, I was given to understand that Banzai points were only for 4-4-3-2 and 4-3-3-3. Accordingly, there would not be any 5-card suits.
We can easily argue that K-x-x opposite A-x-x is about the same as K-x-x opposite Q-J-x

But we cannot so easily argue that A-x-x-x-x opposite K-x-x-x is equal to Q-J-x-x-x opposite K-x-x-x.

[Stephen Tu]

Dumoti, on 2018-July-12, 15:45, said:

Sorry that I didn't reply immediately. I posted, then life got in the way.

I do not agree that Banzai Points are useless. For example, opposite:

Kxx

Axx yields 2 tricks.
QJx also yields 2 tricks.

So it stands to reason that Axx is approximately equal to QJx for no trump purposes.

Not really. There are a couple of issues, one is tempo. Say opponents have knocked out your stoppers, with the ace combo sometimes you take two tricks in the suit and make your contract. With the QJ combo the missing ace is their entry to run their suit, on which you have to discard winner(s).

One is the need for some help opposite; Axx opposite xx is 100% sure trick and stopper and allows you to hold up. QJx opposite xx, if it can be led through twice 24% of the time provides zero tricks and the other times you can't hold up and their suit will run next time they get in.

You might want to look at
http://bridge.thomas...com/valuations/He did a lot of computer simulations looking at how well various point counting schemes do at estimating # of tricks at NT and suits.

[Dumoti]

With all due respect to Thomas Andrews, most of his data have been generated using double dummy solvers.

For example, holding Axxx opposite Qxxx, the percentage play is to cash the ace and then lead low to the queen. You hope to find some combination of:

1. The stiff king.
2. The king onside.
3. The suit breaking 3-2.

However, double dummy, the computer will also pick up Kx offside because it will know to cash the ace and duck a small one to the now-bare king. At the table, there are no such guarantees.

Similarly holding Kxxx opposite Qxxx the computer will invariably pick it up for one loser whenever Ax is on either side. In real life, this cannot be so readily accomplished.

So most of his data are misleading and the virtues of Binky Points are uncertain.

[Stephen Tu]

Multiple studies from large numbers of online bridge records, and national championship/world championship records, have shown that double dummy analysis as an estimate is very close to actual single dummy results by humans below the slam level. The advantage that DD gives to declarer getting suit combos like this right when possible is basically offset by the defense never blowing a trick on opening lead and always finding correct shifts later.
Banzai was published around 1987, if it was really demonstrably more accurate, how come it is practically unheard of, and no top players using it?

[Dumoti]

Couldn't the same criticism be leveled at Binky Points? If it's so wonderful, why has no one at my club heard of them?

If the analysis at http://bridge.thomas...luesfor3nt.html is smack on and Work Points are off by 38.99 percent at predicting 3NT whereas 3 other methods are demonstrably better, why are these methods unheard of?

[Stephen Tu]

Binky points are obscure and not a practical method for calculating at the table unless you are a computer that can look up values from a lookup table. So it's understandable why no one uses them.

Basically Andrews showed that Work points are already pretty accurate when you adjust upwards for aces and tens. *which good players do automatically anyway*, when judging what to do in borderline cases, deciding to upgrade a hand to a 1nt opener or bid 3nt over 1nt, or whatever. Even if they aren't using the exact formula he settled on as most accurate (the "fifths" evaluation), it amounts to the same approximate thing in practice. Bid more having tens and aces. Bid less with mostly quacks. So they don't need to know the exact formula of the fifths evaluation when they are already approximating it by their actions.

Banzai devalues aces too much and overvalues Qs and Js, which is the opposite of what you want to do. And even more so at suit contracts, and you don't know in advance whether you will end up in NT or suit when you open a balanced hand.

[Dumoti]

Stephen Tu, on 2018-July-13, 13:58, said:

Basically Andrews showed that Work points are already pretty accurate when you adjust upwards for aces and tens.

Okay. Let's start right there. Andrews made 4 assumptions, which were:

1. South is "balanced" (which could include 5-3-3-2).
2. There is no 8-card major fit (Okay. Whatever).
3. North is not "freakish" shape (freakish undefined, but fair enough).
4. Neither hand has a doubleton queen or jack without a higher honor (no Qx, Jx), and north doesn't have any singleton kings.

These assumptions are far different from the ones used to develop Banzai points. These points are for two hands that are 4-3-3-3 or 4-4-3-2. 5-3-3-2 is explicitly excluded. Nor do Banzai points shy away from considering how well Qx will do opposite KTxx or some other holding. And Andrews' caveat that north shouldn't have any singleton kings(!!) is a far cry from Banzai's requirement that north be 4-3-3-3 or 4-4-3-2. Obviously, there will be no singleton kings in this scenario! But apparently Andrews WOULD accept a hand like:

♠x
♥Kxx
♦xxxx
♣AKxxx

The evaluation of that hand for NT purposes is well outside the scope of Banzai points.

So, Banzai points are not the right tool for evaluating every hand. But you shouldn't say that a hammer is a bad tool just because you can't use it to change the tires on your car. It's just not the right tool *for that job*.

Quote

Banzai devalues aces too much and overvalues Qs and Js, which is the opposite of what you want to do. And even more so at suit contracts, and you don't know in advance whether you will end up in NT or suit when you open a balanced hand.

I think, you need to go back and read the OP. The plan is NOT to use Banzai to decide when to open the hand. The plan is NOT to use Banzai when you are in a suit contract. The plan is NOT to use Banzai when you are contemplating an opening NT bid.

The plan is to use Banzai points when your partner has already opened 1NT and you are responding with a hand that is 4-3-3-3 or 4-4-3-2.

Now admittedly Banzai points may not work as well as advertised because responder will have no way of knowing whether the opening bidder has an undisclosed 5-card suit. But that's the point of trying them — to see how well they work in the field.

[Stephen Tu]

Look, you are free to use whatever you want. Go ahead and whip out Banzai points, play using the published guidelines for some number of months and see how you do in your partnerships, report back.

I don't see how you can not use Banzai to decide on initial range of opening balanced hands, then switch to Banzai later, because on some hands with an extreme number of aces/tens vs. quacks the difference between Work & Banzai will be quite pronounced and hard to show later when you end up in a suit contract. Like with Axx Ax Axxx Axxx which is a "weak nt" in Banzai but clearly a strong NT playing anything else. 1d-1h-1nt, how does partner with xx KQxxxx Kx xxx know he is facing that hand where you want to be in game or QJx Ax QJx QJxx which is equal banzai points where 2h is the limit and might not even make? I suppose the 4 ace hand can try to raise to recover from the underevaluation when partner bids 2H, but that jeopardizes the partial on weaker hands partner might have with no interest even opposite 4 aces, and with 5 cd only.
I don't see how you can accurately estimate combined Banzais when only one side of the partnership is using them. Sure you can be approximately right on average, but some hands will be pretty far off.

But just know this topic has come up before in this forum, multiple times, and most of us are highly skeptical of them. Most of us are doing just fine using 4321 and mentally upgrading aces and tens a bit and downgrading for lack of them.
Also with regard to Andrew's study, he did note excluding doubleton quacks and singleton Ks was controversial, and also included a table with the results without those exclusions under "less ideal situations".

[blackshoe]

Klinger and Jackson wrote a book about Banzai points. As I remember, a lot of people viewed it negatively.

[Dumoti]

Stephen Tu, on 2018-July-13, 14:59, said:

Look, you are free to use whatever you want. Go ahead and whip out Banzai points, play using the published guidelines for some number of months and see how you do in your partnerships, report back.

I don't see how you can not use Banzai to decide on initial range of opening balanced hands, then switch to Banzai later, because on some hands with an extreme number of aces/tens vs. quacks the difference between Work & Banzai will be quite pronounced and hard to show later when you end up in a suit contract. Like with Axx Ax Axxx Axxx which is a "weak nt" in Banzai but clearly a strong NT playing anything else. 1d-1h-1nt, how does partner with xx KQxxxx Kx xxx know he is facing that hand where you want to be in game or QJx Ax QJx QJxx which is equal banzai points where 2h is the limit and might not even make? I suppose the 4 ace hand can try to raise to recover from the underevaluation when partner bids 2H, but that jeopardizes the partial on weaker hands partner might have with no interest even opposite 4 aces, and with 5 cd only.
I don't see how you can accurately estimate combined Banzais when only one side of the partnership is using them. Sure you can be approximately right on average, but some hands will be pretty far off.

But just know this topic has come up before in this forum, multiple times, and most of us are highly skeptical of them. Most of us are doing just fine using 4321 and mentally upgrading aces and tens a bit and downgrading for lack of them.
Also with regard to Andrew's study, he did note excluding doubleton quacks and singleton Ks was controversial, and also included a table with the results without those exclusions under "less ideal situations".

Yes, I know the topic has come up multiple times. I've read all the threads about Banzai Points contained in the forum.

However, I am highly skeptical of your claim that you just use 4-3-2-1 and mentally upgrade for As and Ts and are "doing just fine."

Here are two hypothetical hands:

♠ ATx
♥ Axx
♦ Axx
♣ Axxx

♠ Kxx
♥ Kxx
♦ xxx
♣ Kxxx

North counts 16 HCPs plus 4 aces and a ten and upgrades it to 17. He opens 1NT. South invites to 2NT and North bids 3NT like a shot.
We will also stipulate that you don't get a diamond lead or, if you do, it breaks either 4-3 or 5-2 with the hold-up play cutting communication.
What are your chances of making the contract?

Well, they're not very good. You have 7 top winners. If clubs go 3-2 you can knock out the ♣Q for down 1. If the clubs are 4-1 or 5-0 then things get worse.

Of course, Banzai Points advocates will be quick to point out that 4333 opposite 4333, there's no way that aces are worth 4.25 if we're evaluating for no trump purposes.
So you can be skeptical all you want, but until you show me a 4333 opposite a 4333 in which Work Points gets it right but Banzai Points screws it up, I will simply classify your skepticism as unfounded.

Now that doesn't mean that I'm not aware of the technical problems with implementing Banzai Points. There's no way for the responder, in a vacuum, to know whether opener bid 1NT with a 15-17 and 3-2-5-3 or bid 1NT with Axx, Axx, Axx, Axxx, or bid 1NT with KQTx, KQTx, xx, KQTx.

In practice, however, we expect responder to be using Banzai Points as a check back when his hand is something like:
AJxx
Axx
xxx
xxx

This is a solid 9 HCPs and opposite opener's theoretical 17 HCPs, it could make 3NT. In fact, if we upgrade 0.25 per ace, then it seems even more clear cut. However, Banzai Points argue for caution with this hand.

Nevertheless, as a practical matter, at least when we look at this hand, we know for sure that partner didn't open Axx, Axx, Axx, Axxx.

[smerriman]

Dumoti, on 2018-July-14, 19:55, said:

So you can be skeptical all you want, but until you show me a 4333 opposite a 4333 in which Work Points gets it right but Banzai Points screws it up, I will simply classify your skepticism as unfounded.

I don't see how this could possibly be relevant? Of course 4333 vs 4333 is not good. But you don't know whether your partner has a 4333 or not.

If you want to say ATx Axx Axx Axxx shouldn't bid 3NT after partner invites with 2NT, you need to simulate all possible 2NT hands, not just give an example of one. Then figure out whether it is worth bidding game.

In this case a simulation says yes, so if you're suggesting you should pass, you're playing losing bridge.

[Stephen Tu]

You can always cherry pick some constructed deal to support your case. So what? Bridge is a game about what works on average, nothing works all the time. Andrews analyzed what works on average. Cowan, being hamstrung with much slower computers, and no double dummy software, from being a decade earlier, did an analysis based on suits in isolation, which doesn't really work because bridge isn't played that way, and tricks won later aren't as good as tricks won right away, because late tricks might go away on the opponents suits. Just because you'll have 9 tricks eventually doesn't do you much good when opps get six first.

I know who I believe is going to be right more often. If you want to overbid your quacks, and treat two jacks and a ten as equal to an ace, then I really want to play against opponents willing to bid so much holding those, and who underbid because "I had too many aces partner, aces are overrated".

You want to not invite with AJxx Axx xxx xxx because according to Banzai partner needs a super-max? Then at MP over a quick 1000 board sim, I win 39.4% when partner bids game and makes, tie the board 38.5% when partner has a min 15 and doesn't accept and can make 2nt+, and lose the board 22.1% when 2nt or 3nt fails. So that's an average of 58+% against 1nt passers. I'll take that.

[Dumoti]

smerriman, on 2018-July-14, 20:48, said:

I don't see how this could possibly be relevant? Of course 4333 vs 4333 is not good. But you don't know whether your partner has a 4333 or not.

Yes, but when partner opens 1NT his possible shapes are 4-3-3-3, 4-4-3-2, and 5-3-3-2. That means that 67.4% of the time he will have a hand that is the right shape for applying Banzai Points. So if responder also has a 4-4-3-2 or 4-3-3-3 hand, then applying Banzai Points might be worth the trouble and that is something that I wish to investigate. What's the problem?

Quote

If you want to say ATx Axx Axx Axxx shouldn't bid 3NT after partner invites with 2NT, you need to simulate all possible 2NT hands, not just give an example of one. Then figure out whether it is worth bidding game.

I think that you need to go back and read the original post. The plan is to experiment with Banzai Points as the responder when partner opens 1NT. The plan is not to try to mastermind what an opening 1NT bidder should do.

The point of the example is to debunk the claim that Aces + upgrades is superior to Banzai Points when the hands are the right shape.

Quote

In this case a simulation says yes, so if you're suggesting you should pass, you're playing losing bridge.

I'm not interested in watching you build up straw men and then take them down.

[Dumoti]

Stephen Tu, on 2018-July-14, 22:11, said:

You can always cherry pick some constructed deal to support your case.

Well, if it's so easy, then why didn't you do it when I asked you to?

Quote

So what? Bridge is a game about what works on average, nothing works all the time. Andrews analyzed what works on average.

This claim has already been debunked.

Quote

Cowan, being hamstrung with much slower computers, and no double dummy software, from being a decade earlier, did an analysis based on suits in isolation, which doesn't really work because bridge isn't played that way, and tricks won later aren't as good as tricks won right away, because late tricks might go away on the opponents suits. Just because you'll have 9 tricks eventually doesn't do you much good when opps get six first.

Yeah. I've played bridge before. I know how No Trump contracts work. Thanks, though.

Quote

You want to not invite with AJxx Axx xxx xxx because according to Banzai partner needs a super-max? Then at MP over a quick 1000 board sim, I win 39.4% when partner bids game and makes, tie the board 38.5% when partner has a min 15 and doesn't accept and can make 2nt+, and lose the board 22.1% when 2nt or 3nt fails. So that's an average of 58+% against 1nt passers. I'll take that.

Anyone can invent numbers. What I asked you to do was to invent two 4-3-3-3 hands in which the 4.25-3-2-1-0.5 system was superior.

Can you do it? Or can't you?

[johnu]

Dumoti, on 2018-July-09, 12:44, said:

What are your thoughts?

Other forum members gave their opinion of Banzai points. You don't agree. OK, but you're not going to convince anybody without some concrete evidence which nobody has.