A blast from the Past. An old thread post on "debating Level 1 systems" on wongs free

[BoSox]

The whole thread can be read on BJ21 free site on index page 54.

Flash's Post In A Previous Thread Needs Further Discussion
Posted By: BoSox <Send E-Mail to this Poster>
Date: 26 Nov 12, 11:53 am

Flash's post was posted on November 23, 2012, titled "The Facts And A Healthy Measure Of Skepticism," needs a response. Flash was talking about a Level 1 counting system and the problems associated with it.
Flash wrote "The game with GOOD conditions is still a game where your card counter edge is going to be below 1% as there plays you cannot make like doubling soft 20 and splitting faces, and your spread needs to be "polite" to maintain longevity and we cannot estimate cards/decks played as accurately as a computer does."
Flash, you are lumping together every player who only uses a Level 1 count system in their arsenal, such as his or her level of experience, which games they are playing, their level of wagering, their spread used, and in which casinos they are playing for tolerance levels, etc.
Flash, you are also (whether you realize it or not) are attempting, very poorly, to discredit the excellent book Black Jack Attack, Playing the Pros Way, by Don Schlesinger. Specifically, Chapter 10, The World's Greatest Blackjack Simulation, considering the research that has gone into it by the author, as well as the respected John Auston. What is wrong with you?
"Playing efficiency with a Level 1 counts tends to hover just above .50%."
True playing efficiency is an important aspect of CC, which is 13% of the gain attributable to card counting and cannot be ignored. Flash, you conveniently forget to mention Level 1 counts like the High Low have a 97% betting corr. efficiency and gains 87% attributable to card counting, along with basic strategy alone.
BoSox

Password:

Messages In This Thread

• Flash's Post In A Previous Thread Needs Further Discussion (1157)

[BJGenius007]

The debate is pointless in the matter of the best counting system. It is just a matter of the taste and the circumstance. People can play faster using Hi-Lo. With sufficient spread (20-1 in shoe), it works. With more complicate counting system, players slow down the game but they could have higher return at the same 20-1 spread or reduce spread at sweat shops.

[ZenMaster_Flash]

Botox,

So now you stoop to cherry-picking old posts (not by me) in your crusade of character-assassination. You are being dishonorable.

[BoSox]

Botox,

So now you stoop to cherry-picking old posts (not by me) in your crusade of character-assassination. You are dishonorable.

As far as me being on a crusade of character-assassination against you, that couldn't be further from the truth. I actually admire what you are doing trying to help out new players. I do not consider myself dishonorable, the last thing I am trying to do is hurt anyone here. I have no problem following the site's rules, I hope you do to. In my above post if I have the wrong Flash please explain.

[Three]

True playing efficiency is an important aspect of CC, which is 13% of the gain attributable to card counting and cannot be ignored. Flash, you conveniently forget to mention Level 1 counts like the High Low have a 97% betting corr. efficiency and gains 87% attributable to card counting, along with basic strategy alone.
BoSox

But what you really want to do is bet accurately. BC is not a measure of how accurate your betting is. There are stats that tell a bit about how accurate you are betting like n0 with the same spread and RoR for the same BR. The short run is a function of n0 or if you like getting to the long run is a function of n0. There is a great old thread that compared betting accuracy or at least the bell curve around your advantage estimate for different approaches. The Hiopt2 bell curve was half as wide as the Hilo bell curve. In other words Hiopt2/ASC bet much more accurately by always being relatively close to the actual advantage with the TC advantage estimate while Hilo was off by more than Hilopt2/ASC worst case scenario a significant amount of the time. In term of BC we are talking about 1 point difference but in terms of betting accurately the difference is huge. The accurate bets lets you beat the game with a small n0 and a smaller spread.

[BoSox]

But what you really want to do is bet accurately. BC is not a measure of how accurate your betting is.

Tthree, do you deliberately go out of your way to insult someone.You will say you are only pointing out differences in systems, maybe it's me but I believe it goes further than that. The Hi LOW BC is more than efficient enough for me. Here is the glossary term:

Betting Correlation. A term used to illustrate the efficiency of a counting system's ability to inform the player when to increase or decrease his bet. It is usually expressed as a decimal, such as 0.95. This would mean that a count with a 0.95 betting correlation is correct 95% of the time in determining the proper bet size.

In term of BC we are talking about 1 point difference but in terms of betting accurately the difference is huge.

Does the Hi Low system measure up to some of the better systems NO it does not. However to say that the Hi LOW system hurts in the betting aspect, to all of the new players starting out on this board is asinine. Try making this argument on Wong's site and see what happens.

[ZenMaster_Flash]

Incidentally, the three year old post from BJ21
criticizing me was posted by an intelligent but
(admittedly) mentally ill person (Aspergers Syndrome).

He attacked all of my posts and finally drove me away.

(His math abilities were too respected for him to be barred)

I'll not let that happen here !

[BoSox]

ZenMaster Flash, for the record I have always admired your work and your intelligence. I think you would be, and all ready are, a great asset to any board.

[ZenMaster_Flash]

Botox,

Thanx for that ! I appreciate hearing it; as just like me,

you can be a bit caustic in your comments from time to time.

[Three]

Betting Correlation. A term used to illustrate the efficiency of a counting system's ability to inform the player when to increase or decrease his bet. It is usually expressed as a decimal, such as 0.95. This would mean that a count with a 0.95 betting correlation is correct 95% of the time in determining the proper bet size.

BC is a measure of how well your count tags correlate to full deck EOR's for the various ranks. Each count will have that degree accuracy for the average of all situations that fall into a TC. It doesn't say anything about the range of the advantage within the bin that make up that advantage or the SD of the bell curve for the TC bin. These are what betting accuracy is about. Accurate betting has your average for a bin be pretty close to the range of actual advantage within the bin. In other words your bets are never that far off the actual advantage. But less accurate betting can have you betting big when you actually are at a disadvantage or betting min when you actually have a big advantage. Just play around with a composition dependent analyzer and calculate the TC for various counts as you try to come up with deck compositions that fool the count. Some counts are easily fooled while others are always pretty close to the actual advantage predicted by the CDA. The difference is how you treat the low cards. Counting them all the same will make your count easily fooled despite having the average advantage for the bin being close to the average for all the possible deck compositions that fall within the bin. What you want is a counting system that groups similar advantage situations together in the same TC. Some counts are better at this than others and BC is not a predictor of which counts are good at not being fooled by the count when it comes to actual advantage for the current deck composition.

[Three]

Betting Correlation. A term used to illustrate the efficiency of a counting system's ability to inform the player when to increase or decrease his bet. It is usually expressed as a decimal, such as 0.95. This would mean that a count with a 0.95 betting correlation is correct 95% of the time in determining the proper bet size.

A correlation is not a percent. Your interpretation of correlation is wrong. Read the following quote from BJ math cautioning the placing of percent to a correlation coefficient:

"Another caution should also be observed in interpreting correlation coefficients. Because they are coefficients, they cannot be interpreted as percentages of agreement. A coefficient of r = 0.30 does not indicate that there is a 30% agreement between the two sets of scores. Also, it is not proper to say that a correlation of r = 0.40 is twice as strong as a correlation of r = 0.20 just because the coefficient is twice as large. Coefficients of correlation are useful only in judging relative strengths of association and for indicating significant relationships between two variables".

You are lacking in some basic understanding of the terms you use. BC is simply the (sum of the count tags for each rank times the rank's EOR) divided by (the square root of ((the sum of the count tag squares) times (the sum of the EOR's squares for each rank)):

sum of(C(i)*R(i))/((sum of (C(i)^2))*(sum of (R(i)^2)))
Where C(i) is the count tag for rank "i" and R(i) is the EoR for rank "i".

Correlation coefficients measure the degree of a linear relationship between two things. EoRs are full deck EoRs and the EoRs for whacky deck compositions would be very different. Often understanding something can be helped by taking examples to the extreme. Let's see the change in advantage for dealing all the aces out of a 2 deck game H17, DAS, RSA1 one at a time:

Full deck (no aces removed): -.3333%
1 ace removed: -.6044%
difference -0.2711%
2 aces removed: -.8815%
difference -0.2771%
3 aces removed: -1.165%
difference -0.2835%
4 aces removed: -1.457%
difference -0.292%
5 aces removed: -1.756%
difference -0.299%
6 aces removed: -2.062%
difference -0.306%
7 aces removed: -2.379%
difference -0.317%
All 8 aces removed: -2.704%
difference -0.325%

The EoR of 1 card of any rank will depend on what cards have already been removed. As you can see it is not linear. The EOR of the ace increased by 20% just by removing aces. T's have a bigger affect on the value of the ace than other aces. Every card removed affects EoRs a little. The effect is fairly negligible except in really whacky deck compositions which is where we tend to be making big bets. Fortunately a simulation would take this into account to a degree. The degree depends on how well it groups similar situations together. The more a count tends to group similar situations together the more accurate the simulation is at adjusting for whacky deck compositions. This can be seen in tighter bell curves around the advantage estimate for each set of situations that are grouped together. BC does not measure the ability of a count to group similar situations together but betting accuracy depends on this ability. Your count needs to get more complex to start tightening the bell curve by tending to group more similar situations together in the same betting bin. A simple count simply has to group a wider range of situations into the same group. n0 is the measure that best shows how good a count is at grouping similar situations together.

The long run is a multiple of n0 where you expectation at least 2 or 3 SD's. I have heard people use a much higher figure. N0 is an exponential function of SD so we are talking 4 or 9 times n0 or even higher (16, 25, 36, 49, 64, 81, 100 times n0 depending on the multiple you are squaring). This is why having a low n0 is so important to consistently keeping bad runs short, light and infrequent while having the positive runs to negate them. This is simply the math of the game. The approach you choose affects all these things. If you are wining about losing it may simply be what ride or the random walk to the long run that your count choice, game selection and playing style comes with. When choosing a count you should make sure the ride or random walk to the long run fits your BR and comfort tolerance for bad runs for the games you play and your playing style. If it does then your count choice is fine for you. If not try to get a better understanding of what causes the random walk and make the necessary adjustments to one or more of the following:
1) game selection (The rules, pen, casino tolerances etc that you are playing)
2) Playing approach (Wonging, number of indices, or more advanced techniques that can be used in conjunction with your count that alter things like TC frequencies played, advantage at various TC's or accuracy of bets and plays (Not necessarily BC and PE), etc).
3) Your count choice (Try to tighten up those bell curves around each decision you make. If you don't no how to do that minimizing n0 has a high correlation to the same)

Note: Most players could help their expectation a lot from the list above without switching counts. But if you are using a simple approach switching counts will go a long way to change the random walk to the long run. If you are already using a level 2 count or some other advanced strategy changing counts may be less effective at changing the random walk. If you try to understand what things change and the effect of the change you will have a much better chance at switching that random walk for something that suites you better, that is if you would prefer a higher EV and more predictable short term results.

Another thing I hear KJ say he does that really makes the walk even more random is having different betting levels for different casinos. In the real world we must all make choices to keep from getting 86'ed but you should understand the effect of those choices and make the ones that you are most comfortable with. KJ knows the effect of the choices he makes. He is fine with it, or at least that is what he keeps saying to us.

Anyway too many people don't understand what drives the random walk and they don't understand what they can do to make it more predictable. You go on a trip and will play 40 hours, lets say 100 rounds/hour, for a trip total of 4000 rounds. If your approach to the games you will be playing has a n0 of 40K rounds you expect to take 10 trips to be fairly certain you will be ahead. If your approach to the games you will be playing has a n0 of 16K rounds you will be fairly certain you will be ahead after just 4 trips. Playing different bet schedules for different casinos will trash your overall n0. So as you can see short term results can be made a lot more predictable. People that have taken measures to rein in the leash for the random walk (the walking the dog metaphor for expectation versus actual results) usually don't complain much about their losing runs. Eventually, if they play enough, they will have a particularly bad run but it won't be so much expected as a part of the deal.

[ZenMaster_Flash]

While Three's post is accurate, he would do a
boon to 95+% of the readers by posting a
summary that will simplify the definitions and
present "facts on the ground" (non-academic)
views of: B.C., P.E., and I.C.

Your intellectual power needs no introduction;
but it may be challenging for you to re-do your
posted material.

[BoSox]

Originally Posted by BoSox
Betting Correlation. A term used to illustrate the efficiency of a counting system's ability to inform the player when to increase or decrease his bet. It is usually expressed as a decimal, such as 0.95. This would mean that a count with a 0.95 betting correlation is correct 95% of the time in determining the proper bet size.

A correlation is not a percent. Your interpretation of correlation is wrong.

RIght, or wrong I can not take credit for the quote. In front of the quote I wrote it was a glossary term. I will admit I have believed the quote for a very long time, as it has served me well.