Sonuvabish
Joined: Feb 5, 2014
• Posts: 1342
April 3rd, 2014 at 4:35:19 PM permalink
Quote: AxiomOfChoice

A count where some parameters get cubed before comparing to the index number.

x + y - z > 3 is linear.
x^2 + y - z > 3 is quadratic
x^3 - 2y^2 + 4 < 12 is cubic

etc, etc. Because baccarat uses at most 6 cards to complete a hand, perfect play would require no more than an order-6 polynomial (I think?)

Why would you use exponents to count? All you're doing is curving the line. Am I missing something?
AxiomOfChoice
Joined: Sep 12, 2012
• Posts: 5761
April 3rd, 2014 at 4:37:09 PM permalink
Quote: dwheatley

I think I see the 6th order poly you would have to use, but it would have ridiculously complicated coefficients that depend on the deck composition. Why are we even talking about polynomial counting systems?

I'm not claiming that you should actually use the pages-long perfect count. I'm just saying that there is no reason to limit yourself to a linear system when the game is non-linear, the game is slow, and they give you a pencil and paper. Eliot's linear count captured about half the value -- it seems likely that you could do better with a simple quadratic or cubic count.
Sonuvabish
Joined: Feb 5, 2014
• Posts: 1342
April 3rd, 2014 at 4:42:40 PM permalink
Quote: AxiomOfChoice

I'm not claiming that you should actually use the pages-long perfect count. I'm just saying that there is no reason to limit yourself to a linear system when the game is non-linear, the game is slow, and they give you a pencil and paper. Eliot's linear count captured about half the value -- it seems likely that you could do better with a simple quadratic or cubic count.

That's not the type of multi-paramter counting that I am familiar with.

Half the value? Can a quadratic count double blackjack counting efficiency? If so, how can it's power be tested?
AxiomOfChoice
Joined: Sep 12, 2012
• Posts: 5761
April 3rd, 2014 at 4:53:45 PM permalink
Quote: Sonuvabish

That's not the type of multi-paramter counting that I am familiar with.

All the multi-parameter counts that you are familiar with are linear, though.

Quote:

Half the value? Can a quadratic count double blackjack counting efficiency? If so, how can it's power be tested?

With a simulation? I mean, that's what playing efficiency is (linear systems are fine for betting efficiency as long as penetration does not get too deep -- the non-linearity of blackjack doesn't hurt your counting system until you're down far beyond anything that a casino will deal you, I think)

You are also not going to double anything, I don't think. Lots of simple counts have betting efficiencies in the 97-99% range. I don't remember the playing efficiencies -- I know that they are bad, but not THAT bad.

It's all irrelevant, anyway, I think. I don't think that too many people are going to be able to do this quickly, in their head. Baccarat gives you lots of time between hands (in high-limit squeeze games) and a pencil and paper. Unfortunately they don't seem to offer 9-1 so I suspect that it's moot anyway.

It was an interesting theoretical discussion, though.
Sonuvabish
Joined: Feb 5, 2014
• Posts: 1342
April 3rd, 2014 at 5:12:32 PM permalink
Quote: AxiomOfChoice

All the multi-parameter counts that you are familiar with are linear, though.

With a simulation? I mean, that's what playing efficiency is (linear systems are fine for betting efficiency as long as penetration does not get too deep -- the non-linearity of blackjack doesn't hurt your counting system until you're down far beyond anything that a casino will deal you, I think)

You are also not going to double anything, I don't think. Lots of simple counts have betting efficiencies in the 97-99% range. I don't remember the playing efficiencies -- I know that they are bad, but not THAT bad.

It's all irrelevant, anyway, I think. I don't think that too many people are going to be able to do this quickly, in their head. Baccarat gives you lots of time between hands (in high-limit squeeze games) and a pencil and paper. Unfortunately they don't seem to offer 9-1 so I suspect that it's moot anyway.

It was an interesting theoretical discussion, though.

Correct, I am familiar with linear approaches to multi-paramter counting. I figure they are a waste of time. And I am the one who 'wastes' his time with the Ultra count. Here are my stats: BC: .992 PE: .649 IC: .88. Impressed or are you a bigot?

So are you basically saying that blackjack approximates linearity, whereas baccarat does not? I don't see why you couldn't do a quadratic in your head: 1, 4, 9, 16, 25...you're just squaring a number. May need a a cue in remembering what next to square. Not sure how this actually would work in a count system.
endermike
Joined: Dec 10, 2013
• Posts: 584
April 3rd, 2014 at 5:15:45 PM permalink
I am going to make one vague statement and then add more depending on how some analysis goes. Higher order counts (meaning exponents and interactions) can provide more information than linear ones. I make no statements to orders of magnitude on value or feasibility of use in casinos.
AxiomOfChoice
Joined: Sep 12, 2012
• Posts: 5761
April 3rd, 2014 at 5:17:06 PM permalink
Quote: Sonuvabish

Correct, I am familiar with linear approaches to multi-paramter counting. I figure they are a waste of time. And I am the one who 'wastes' his time with the Ultra count. Here are my stats: BC: .992 PE: .649 IC: .88. Impressed or are you a bigot?

So are you basically saying that blackjack approximates linearity, whereas baccarat does not? I don't see why you couldn't do a quadratic in your head: 1, 4, 9, 16, 25...you're just squaring a number. May need a a cue in remembering what next to square. Not sure how this actually would work in a count system.

I think that the non-linearity becomes a serious problem at very, very deep penetrations. If you can find an 8-deck blackjack game that only cuts off 14 cards, it might be worth looking into.
Sonuvabish
Joined: Feb 5, 2014
• Posts: 1342
April 3rd, 2014 at 5:26:19 PM permalink
Quote: endermike

I am going to make one vague statement and then add more depending on how some analysis goes. Higher order counts (meaning exponents and interactions) can provide more information than linear ones. I make no statements to orders of magnitude on value or feasibility of use in casinos.

Interactions? Like the individual EOR effects depending on your particular hand, or for the particular order of cards that came out? How do you count that? Feasibility isn't really the issue at the moment, but magnitude is. We're not even sure if it works. If you can shed some light, please do.

AOC: No, the deepest dealt games I have played were about 2/3 of a deck left in 6D and about 1/4 deck in DD. Nothing over 95% penetration. If that is what is required for gain, then it is clearly unworkable.
teliot
Joined: Oct 19, 2009