PE and BC

[lij45o6]

So, in my time that I have available, (sparingly,) I have taken up the initiative to research other counts and to dive back into Griffin's ToBJ. Something that came to me is the difference between Playing Efficiency and Betting Correlation. From my understanding, a PE that approaches 0.7 means that the system performs well with adjusted strategies alongside the count, ex: indices; while a BC that approaches 1.0 means that the system shows any given advantage by the given probability, ex: BC of 0.93, the count shows an advantage correctly 93% of the time. Please correct me if I am wrong! Also, how would these correlate with single deck and multi-deck games. From past discussions, PE plays better with a greater number of decks like 6 and 8, while BC plays better with single and double. I have also noticed the difference between ace reckoning and ace non-reckoning (neutral) counts and their effect on PE and BC. An ace reckoning gives a high BC and a low PE, ( assuming such numbers are "average" relative to other counts,) while a ace neutral count gives a higher PE and a lower BC, relative to the native ace reckoning system. Why is that?

[Three]

ex: BC of 0.93, the count shows an advantage correctly 93% of the time. Please correct me if I am wrong!

BC is a correlation coefficient not a percentage. It only measures relative relationships not absolutes. Placing a percentage there shows an absolute relationship which correlation coefficients don't do.

[lij45o6]

Gotcha.

[Freightman]

J

So, in my time that I have available, (sparingly,) I have taken up the initiative to research other counts and to dive back into Griffin's ToBJ. Something that came to me is the difference between Playing Efficiency and Betting Correlation. From my understanding, a PE that approaches 0.7 means that the system performs well with adjusted strategies alongside the count, ex: indices; while a BC that approaches 1.0 means that the system shows any given advantage by the given probability, ex: BC of 0.93, the count shows an advantage correctly 93% of the time. Please correct me if I am wrong! Also, how would these correlate with single deck and multi-deck games. From past discussions, PE plays better with a greater number of decks like 6 and 8, while BC plays better with single and double. I have also noticed the difference between ace reckoning and ace non-reckoning (neutral) counts and their effect on PE and BC. An ace reckoning gives a high BC and a low PE, ( assuming such numbers are "average" relative to other counts,) while a ace neutral count gives a higher PE and a lower PE, relative to the native ace reckoning system. Why is that?

You've got it reversed. PE plays better with pitch, keeping in mind that your soreads are much less. BC plays better with 6-8 deck shies, since you can get away with a higher spread. If you can tweak an ace reckoned count with side counts, which will increase your PE - thins now get interesting.

[Three]

From past discussions, PE plays better with a greater number of decks like 6 and 8, while BC plays better with single and double.

I think you have this backwards. But it is due to casino betting tolerances on betting for pitch games not that BC is not as strong in pitch games. PE is stronger with the fewer cards in play. Each card removed has a bigger affect. I am not going in great detail because I know you understand this. At least I think you do.

while a ace neutral count gives a higher PE and a lower PE, relative to the native ace reckoning system. Why is that?

Typo alert. One of those PE's should be BC. The lower BC for ace neutral counts is true only if you don't side count aces. You can get higher BC with ace neutral counts than ace reckoned since you can weight the ace at an accuracy of 1 decimal and memorize the multiples of the ace tag. The ace carries a 20% heavier weight than the T for betting. Decimal accuracy as a count tag in the main count is not practical so ace reckoned counts that adjust for surplus and deficit aces can weight the ace properly while still being practical even though most don't do it that way.

[lij45o6]

I think you have this backwards. But it is due to casino betting tolerances on betting for pitch games not that BC is not as strong in pitch games. PE is stronger with the fewer cards in play. Each card removed has a bigger affect. I am not going in great detail because I know you understand this. At least I think you do.

I see. So, as the cards are removed, you get a "more accurate" picture of the remaining pack. Still gonna read Griffin's book over again just to make sure.

Typo alert. One of those PE's should be BC. The lower BC for ace neutral counts is true only if you don't side count aces. You can get higher BC with ace neutral counts than ace reckoned since you can weight the ace at an accuracy of 1 decimal and memorize the multiples of the ace tag. The ace carries a 20% heavier weight than the T for betting. Decimal accuracy as a count tag in the main count is not practical so ace reckoned counts that adjust for surplus and deficit aces can weight the ace properly while still being practical even though most don't do it that way.

Typo fixed. From testing ace reckoning vs ace neutral counts, it looks like the neutral ones have a greater PE than BC. All I see is when the ace is ignored that BC drops and PE rises. Unless I am missing something associated with side counting the ace under an ace neutral count. Decimal accuracy would be a pain in the ass, unless it is halves. As you said, the EoR for 10 is -0.51 and A is -0.61, a 20% difference. How exactly would side counting NOT affect the weight of the running count of a system, since you are ignoring a large part of the change in value? How does side-counting benefit the AP?

[Three]

As you said, the EoR for 10 is -0.51 and A is -0.61, a 20% difference. How exactly would side counting NOT affect the weight of the running count of a system, since you are ignoring a large part of the change in value? How does side-counting benefit the AP?

First I will tell you how aces are side counted to adjust the betting RC. You count the number of aces you have seen. Each quarter deck you expect to see 1 ace so you determine how many quarter decks have been seen and if you have seen more aces than that the remaining cards have an ace deficit of the difference between the quarter deck expectation and what was observed and counted. If you gave seen fewer aces than expectation the aces are in surplus. You temporarily adjust the RC by the count tag for aces times the difference. If aces are in surplus you add the difference to the RC but if aces are in deficit you subtract the difference. So you still get the ace in this temporary RC that you use to get the TC. Then you must remember/revert back to the actual RC to move on.

With decimal ace tag (optional): you remember the multiples of the count tag. Like 1.2 for a level 1 count you remember 0, 1.2, 2.4, 3.6, 4.8, 6, 7.2 etc. If you have a whole number divisor for calculating the TC you can floor the decimal before dividing the temporary RC. Half deck you can floor to half deck accuracy. I is up to you if you want to go to quarter deck when dividing by quarter deck accuracy. This all sounds harder than it is. I think in terms of multiples of the divisor to compare to the RC when calculating a TC. If you are doing quarter deck division just use the multiples.

Like if 2.25 decks remain: The multiples for 0 thru 4 are 0, 2.25, 4.5, 6.75, 9. You can use multiples of 9 and then the remainder until you get used to it. Like 33 is 3 nines, or 27, (the 9 is 4 times 2.25) with a reminder of 6 (Multiple that is equal to or less than 6 is 4.5 which is the 2x multiple. So you have 3*4+2 for a TC of +14. You can get really fast with a little practice. This is much easier for most people than division. Quarter decks you consider the 4th multiple and the 1st thru the third for the remainder. Half deck you consider the 2nd multiple and for the remainder the deck estimate.

So with 4.25 decks remaining the fourth multiple is 17 and the 3 multiples for the remainder is 4.25, 8.5 and 12.75. The remainders are easily calculated on the fly as they never cause the decimal to change the non-decimal part of the number. Only the 4th multiple adds 1 to the non-decimal. We just took the tough out of working with decimals.

[ZenMaster_Flash]

"How does side-counting benefit the AP"?

In an Ace-reconed count your True Count leads you to

some very improper plays, as well as inaccurate betting.

For betting purposes, surplus/deficient ACES significantly

alter your chances of having an upcoming profitable hand.

For playing decisions, lets consider these hands:

S17 6 deck You want to double your 11 vs. Ace

and your positive TC has surplus Aces - a card

that hurts you on a doubled bet. How about 9 v. 2?

Now what happens when you have deficit Aces and

you are staring at 10 v 10 at a very high TC. "Risk

Aversion" whispers in my ear, saying "Be Cautious"

I respond, "But I am using a Risk Averse index for

this hand."; but reducing the chance of transforming

my 10 into a 21 is costly. Ditto for splitting 9's v. 7 or Ace.

Would you make some of these plays with surplus Aces at

what looks to you like a sub-optimal T.C. ?

[seriousplayer]

BC is a correlation coefficient not a percentage. It only measures relative relationships not absolutes. Placing a percentage there shows an absolute relationship which correlation coefficients don't do.

When you say "It only measure relative relationships not absolutes". Relative is "in proportion to something else". Percentage is a number used in comparative relation to a whole. Absolute is not relative or comparative.

[seriousplayer]

In an Ace-reconed count your True Count leads you to

some very improper plays, as well as inaccurate betting.

For betting purposes, surplus/deficient ACES significantly

alter your chances of having an upcoming profitable hand.

For playing decisions, lets consider these hands:

S17 6 deck You want to double your 11 vs. Ace

and your positive TC has surplus Aces - a card

that hurts you on a doubled bet. How about 9 v. 2?

Now what happens when you have deficit Aces and

you are staring at 10 v 10 at a very high TC. "Risk

Aversion" whispers in my ear, saying "Be Cautious"

I respond, "But I am using a Risk Averse index for

this hand."; but reducing the chance of transforming

my 10 into a 21 is costly. Ditto for splitting 9's v. 7 or Ace.

Would you make some of these plays with surplus Aces at

what looks to you like a sub-optimal T.C. ?

Ace netural counts are not 100% accurate either. There are compromises in PE, BC and IC. Unless you can get all three ratios to 1.00 side counting aces. There will be inaccuracies. How do you correct the compromise of ace netural counts (side counting aces)? Do you add more side counts?

[lij45o6]

Serious, Flash, Freight, and T3: Thanks for the advice. Starting to clear up the sky a little here.

Dogman: You've received some very good advice in these threads and an excellent summary with questions by Seriousplayer. Two things, the game has to be fun and profitable FOR YOU. One man's fun is anothers misery and/or what one considers junk another considers gold. Now I read in Griffens book where the minimums in pitch games should be in the mid 600's PE and at least a 970 BC. But he didn't go into the balance between the two depending on indices played and bet spreads allowed. Thus the confusion begins. The brainpower exercised in casino play is where the variations of "fun" come into play. In order to determine "profit" everyone would have to take a truth pill. Some claim to count two or more tables while playing at one in order to always play in a postive situation. Good idea, but seems a little farfetched to me and certainly would not be fun to employ for me, personally. Another count, Hi Opt II, requires the side count of Aces and 7s in order to achieve maximum PE and BC. This sounds taxing over 6 decks but perhaps doable to those with a strong memory and mathmetically inclined in a double deck. There are threshold and advantages in the indices. Learning the advantages is key, the thresholds are a bonus. The amount of profit you receive from this game will likely be determined by the amount of patience and discipline you put into it. If you think your mind can handle two counts at the same time, then Felt and Mentor combined would more than serve the purpose of high PE, BC, and IC mentioned in Griffens book. Giving more credit to the 5 and less to the 3 in these two counts would serve the purpose of improved PE without giving up any BC. But an approach that is too complex for the indivual is not prudent either. It takes away from the fun of the game and probably the profit when one busts a spring from not seeing an immediate profit to their extensive mental labor.

My philosophy is what is best for you. I don't really care WHAT you use it is HOW you use it: whether it be a level 1 with I18 or a level 3 with ASC and Full indices. As long as you are taking an optimal advantage from the situation. That is why I asked about PE and BC. I do have fun being at the table; what I like a lot is making money while playing and taking up a higher level would be in my economic favour as well as helping me hide my ass, to a degree. Statistical calculus in Griffin's book is a little beyond me even for someone who has taken SV/MV calc. I get the idea he is making as he is clear on how he goes about his calculations, it is trying to derive his work to find a balance between PE and BC that favours both ( high PE and high BC) and reduces missing information from an "inaccurate" count. "Inaccurate" count and the derived PE/BC are unavoidable. * Inaccurate meaning missing information or misleading information a count gives about a situation. The real trick for me is finding a count that allows an easy side ace conversion without sacrificing processing power in my head but at the same time allows me to derive as much information without sacrificing future income. The counts you gave I will be researching, just as I have been looking at Zen and HO2. Thanks.

[Three]

When you say "It only measure relative relationships not absolutes". Relative is "in proportion to something else". Percentage is a number used in comparative relation to a whole. Absolute is not relative or comparative.

No, there is no quantifiable relationship between to different BC's. All you can say is the higher one has a better BC than the lower BC. That doesn't even say it bets more accurately just that the count tags correlate to the full deck EOR's better. Hiopt2/ASC does a much better job than VAPC/ASC at getting accurate advantage estimates for the bins that will affect your bets yet has a lower BC. VAPC counts the low cards almost identically which costs a good bit in betting accuracy where you will be making bet adjustments. Relative as I tried to use it is a greater than, or less than, or equal to relationship. That is as far as you can go with correlation coefficients which is what BC is. Correlation coefficients measures how linear the relationship is between to things. The trouble with BJ is almost nothing (except insurance) has a linear relationship.

But the point is that correlation coefficients do not produce quantifiable results but only relative results so even if BJ was linear you still couldn't make the leap to quantifiable differences based on the difference in BC. It only can give relative (better, worse or the same) relationships and even then one being better than the other doesn't necessarily mean the better one bets better. It only means the relationship between the count tags and the full deck EoR's is more linear. That most often means that the one with the better BC bets better (more accurately producing more profits when an optimal ramp for your RoR, spread and BR that is based on the advantage estimates for the deck compositions that are lumped into the same TC bins). What you want is a count that does a better job of grouping similar advantage situations into its TC bins. BC really doesn't say much of anything about this ability for a count. All a count can do is make certain similar advantage deck compositions more likely to end up in the same bin and ones of greater difference in advantage to end up in another bin. The more vanilla the count tags treat low cards similarly the more dissimilar advantage situations will end up in the same TC bucket. This means lower optimal bets and less profit generally speaking.

[Three]

Ace netural counts are not 100% accurate either. There are compromises in PE, BC and IC. Unless you can get all three ratios to 1.00 side counting aces. There will be inaccuracies.

Even then there will be lots of inaccuracies because these are limits set for linear systems. Insurance is about the only linear thing that is linear in BJ so IC is an accurate measure of insurance performance. BC measures how linear the count tags correlate to full deck EoR's not how accurate it bets. Since betting accuracy is not linear there is no reason to think that a higher BC necessarily indicates more accurate betting and a more profitable system with each system bet optimally while spread, BR and RoR held constant. I forget which system it was but by the metrics mentioned in the quote you would expect it to be the strongest system out there but it did not perform that way. Anyway my point is when you use a metric that measures how linear a relationship is to gage how well something that is non-linear performs you can't expect it to be 100% accurate with its comparative predictions of actual betting performance.

If you side count aces it is as strong or stronger than a similar ace reckoned count for betting.