Generating Indices for Multiparameter System and Other Questionss

from Alienated aka Ted Forrester

http://games.groups.yahoo.com/group/blackjackcardcounterscafe/message/22304

There is also a great post named "Side count adjustments" from November 18, 2002 on the Card Counting board at Green Chip.

If you're not already a member at Green Chip his posts alone are worth the money! ;)

Just when I was becoming halfway civil to Francis again, you had to dredge this up, huh? :-)

Rereading my response, it's clear to me that I have to get over my shyness and learn to state what's really on my mind. :-)

In any event, if you're reading this, Francis, PLEASE promise me to just let it be. PUH-LEEZE!!

Don

Search this archive for a post by alienated titled
"Calculating BC, IC & PE of Multiparameter Systems" posted "on 28-Nov-2002 04:37:57 (#1036)"

http://www.blackjackinfo.com/cc/5.php

it's the index generator, first link on the page

Courtesy of Dog Hand, I just reread that memorable thread and in hindsight I must say it was a nice crossing of (s)words.

Francis

Don,

Trust me, I am quite familiar with probability, statistics, and mathematics. I am sorry if the word 'seems" did not crop in. My suspicion, by the way, is that what I find with my returns is a result of 'wonging out.' But, as you say, no way to prove it. Too limited a sample size.

Separate question: do you know where I can easily get risk-adjusted indices for Wong Halves, or perhaps as easy which hand combinations are the ones typically subject to variance because of risk, as Bryce Carlson describes it.

My book by Griffin is packed away. I play seldom, and for recreation (the challenge, something like the way some play bridge).

Oh, and are there any non-linear counting systems that you know of, and that are workable? If so, where is the subject covered.

"Separate question: do you know where I can easily get risk-adjusted indices for Wong Halves, or perhaps as easy which hand combinations are the ones typically subject to variance because of risk, as Bryce Carlson describes it."

Norm Wattenberger's Casino Verite will generate risk-adjusted indices. Go to qfit.com to read about his products. BJA3 has r-a indices for Hi-Lo, but just for the Catch-22. On the other hand, I can't think of why you'd want any more. :-)

"My book by Griffin is packed away."

Not a good idea! Mine has been on my desk, in plain sight, for about 20 years! :-)

"I play seldom, and for recreation (the challenge, something like the way some play bridge)."

I understand.

"Oh, and are there any non-linear counting systems that you know of, and that are workable? If so, where is the subject covered."

I'm not sure what you mean by your question. What do you understand a "non-linear system" to be?

Don

1. Yeah, casino verite sounds great, but I'd need a windows emulator because I only own macs.

2. I seem to have 8 days winning, and then one big loss day (looking at when I play over time). Which may be being tired, sloppy on the loss days. Or just the law of large numbers. Your experience when counting? (Trying to analyze my play, the nature of counting, and expected results, etc., on a deeper level, so I can improve and understand what is not to worry about - it'll just happen - and what can be fixed/improved on with technique).

3. What is Catch 22?

4. Carlson makes the comment that ideal play would calculate indices that maximize the log of expected win - I assume he means log of winning probability minus log of losing probability. Not sure though. Assume it relates to the proof of Kelly's optimal betting criterion, or a lemma to it.

"1. Yeah, casino verite sounds great, but I'd need a windows emulator because I only own macs."

Not sure what is available. Perhaps others will help.

"2. I seem to have 8 days winning, and then one big loss day (looking at when I play over time)."

Again, it's all meaningless. You can keep records, but you aren't going to learn too much from them.

"Which may be being tired, sloppy on the loss days."

Do you never feel tired on a day when you win a ton of money? Do you never feel energized, only to get your head handed to you? You shouldn't play if you're tired, but nor should you blame your losses on being tired. You're going to lose extraordinary amounts of money feeling perfectly fine. It comes with the territory.

"Or just the law of large numbers."

Yup.

"Your experience when counting? (Trying to analyze my play, the nature of counting, and expected results, etc., on a deeper level, so I can improve and understand what is not to worry about - it'll just happen - and what can be fixed/improved on with technique)."

Suggest you read BJA3. It should have enough numbers to satisfy your curiosity. :-)

"3. What is Catch 22?"

Ditto. But, the short answer is: The I18 plus two indices for doubling A,8 vs. 5 and 6 and two for hard 8 vs. 5 and 6. So, I dubbed the original 18 plus the four new ones, the "Catch-22."

"4. Carlson makes the comment that ideal play would calculate indices that maximize the log of expected win - I assume he means log of winning probability minus log of losing probability. Not sure though. Assume it relates to the proof of Kelly's optimal betting criterion, or a lemma to it."

He's talking about risk-adjusted indices, which I also discuss in a chapter of BJA3. Christmas is coming; buy yourself a present! :-)

Don

Don,

I have no prob buying the book. Just may not get to me before I go away for Xmas, which is my most immediate desire for the info. I'll agree to buy if you give me the 22 indices and we discuss the log concept so I get it. Btw, I'm pretty sure that Thorp's ten count can be turned into a +/- count, fundamentally, because the log of numbers close to 1 (1 -x, where x is small) is approximately equal to -x).

But this is another discussiong beyond the first two. And I'm sick to boot. Annoying.

Finally, what I hear you saying is that, even with wonging out, the variance for any count is/can be quite large (meaning the variability in win rate per hour compared to the average win rate per hour). Wong's book essentially says that (with numbers not based on wonging ot). But of coruse, I want to understand all the calculations, because it adds to my confidence about what I am doing. Many, many questions, always, which is why I make a good lawyer. :)

Do the R-A indices figure in the bet. For example, one might on average be betting off splitting 10s against a 6 if the TC > 4 (using Halves). However TC > 4 implies a bet of at least 4 units. And since as a player, you encounter 10-10 against a 6 WITH TC > 4 very very rarely, it can be a very risky bet.

In short, I think properly calculated R-A indices ought to be explicit about the betting assumptions involved - or that is my strong intuition, and I'd have to see a proof to be convinced it isn't so.

The other issue is that R-A usually implies that there is an underlying risk preference function, but different people have different risk tolerances. This is the point, in part, of traditional insurance products. Without knowing the risk preference function, I can't believe one is calculating a "true" risk adjusted index but rather one that takes into account standard deviation of result, but places some implied assumption on that standard deviation. (One runs into this issue in certain stock market valuation models, where money managers optimize an objective function, ER - ks, which is Expected return minus some number k times s, the standard deviation of the ER. k, however, is not a priori known, but must be plugged in, so to speak, while ER, and s, are simply measured. The higher k, the more risk is important to the investment decision.

Same principle. (I hope it's becoming clear that math is not my shortcoming.....)

1) I "visit" here and contribute when I can, but I can be found on a more steady basis on advantageplayer.com, where I answer questions daily, especially on Don's Domain.

2) I'm not trying to "sell" you my book. It's just that, unfortunately, I don't always have the time to answer the same questions over and over again, so I do the next best thing and refer people to the exact location in my book where they can find the answer. Sometimes, people misconstrue these actions on my part as wanting to advertise, or sell, my book, but that is never the motive.

3) Trust your instincts! :-)

Don

I have purchased BJA (2d Ed. on amazon) for Tuesday delivery, before I leave for the holiday. All I really want are the Halves risk-adjusted indices. I'll read the rest.

Here is a hypothesis for you, or Stanford, or any similar person:

If you simulate various errors in counting (not play), I hypothesize that over a reasonable range of assumptions, Halves scores higher than AO2 (side count included) in a six deck shoe with comparable sets of indices. This requires a monte carlo simulation, and it requires certain assumptions about types of errors and their rates.

My hypothesis has to do with the nature of the counts, the ease of side counting Aces [which assumes the ability to estimate the amount of 1/4 decks that have passed], the variability they produce and the level (essentially meaning the number of different counts and cards counted).

I am sorry for not being more precise. I hobe this gives enough of a sense to rebut the idea that AO2 dominates Halves at all, if some real world conditions are interjected (that is, mistakes in counting every so often).

"I have purchased BJA (2d Ed. on amazon) for Tuesday delivery, before I leave for the holiday. All I really want are the Halves risk-adjusted indices. I'll read the rest."

I'm awfully sorry, but BJA is now in its third edition, and that version is nearly 50% larger than the second edition that you purchased. I'm surprised that you didn't see BJA3 on amazon or right here. In any event, there is no discussion of risk aversion in the second edition. It is one of the myriad topics that were added to the third edition, which is the one you really want.

Don