Memory and Counting

[Secretariat]

Interesting stuff Midnight and Jester. I did not know about the PAO and Master memory systems. I looked it up. I kind of do the same thing. I have 96 images stored (athletes numbers and jerseys). The numbers "sound' in my head for my first three columns. I use the image for the 4th or 5th column whether I use Tarzan or Hi Lo modified. I never use letters. I don't know how you (Midnight especially) do it quickly but I know that can be done. Good for you.

Midnight, you say that your system is simpler than Tarzan. It just sounds more complicated to me. Especially with insurance. I don't see how you can do it instanatenously like Tarzan. The only thing I dislike somehow is having to play with your chips to keep up. I first thought that you zeroed down the 23-45-67-89 columns and made some kind of visual adjustment with the tens and the aces.

One thing is sure the members here who labeled Tarzan's system as complicated are going to have their head spinning looking at your system. I do too, but I know that any complex system is just a matter of personal skills and dedication to training before being mastered.

And no argument on the shuffle tracking or sequencing. Those memory systems are the way to go.

One last thing Midnight. For betting with the Tarzan system, if you add the 6 side counted and oppose it to the Ace, it's exactly like betting Hi-Lo.

[Midnight]

Interesting stuff Midnight and Jester. I did not know about the PAO and Master memory systems. I looked it up. I kind of do the same thing. I have 96 images stored (athletes numbers and jerseys). The numbers "sound' in my head for my first three columns. I use the image for the 4th or 5th column whether I use Tarzan or Hi Lo modified. I never use letters. I don't know how you (Midnight especially) do it quickly but I know that can be done. Good for you.

Midnight, you say that your system is simpler than Tarzan. It just sounds more complicated to me. Especially with insurance. I don't see how you can do it instanatenously like Tarzan. The only thing I dislike somehow is having to play with your chips to keep up. I first thought that you zeroed down the 23-45-67-89 columns and made some kind of visual adjustment with the tens and the aces.

One thing is sure the members here who labeled Tarzan's system as complicated are going to have their head spinning looking at your system. I do too, but I know that any complex system is just a matter of personal skills and dedication to training before being mastered.

And no argument on the shuffle tracking or sequencing. Those memory systems are the way to go.

One last thing Midnight. For betting with the Tarzan system, if you add the 6 side counted and oppose it to the Ace, it's exactly like betting Hi-Lo.

Hey Secretariat, yes, I'm aware that you could derive Hi-Lo from Tarzan's original DC count but you need a keep a side of sixes and aces apart from the main triplet to do so. In my TOR Triplet DC count this need for extra side counts of individual ranks is eliminated. All the information needed is encoded in the main Triplet (outside of the side EDI count for unbalanced K values). My system is not new but in fact a variant of Tarzan's count where I made a few trade offs inorder to make the system more versatile. Eliminating the EDIs, unmooring the Ace and side counting it, and merging the groups {2,3}with{4,5} and {6,7} with {8,9} transforms TOR back to Tarzan original. Its all in the name as TOR stands for Tarzan on Roids. Its not a new or novel system outside of these adjustments. Similar to how TKO, TKO/A and ReKO are all variants of KO. Its basically the same concept. That's why I didn't change the name of it to something else completely. I give Tarzan 100% credit for the system as he was the first, as far as I know, to describe the process formally in the forums.


As I said TOR makes some trade-offs inorder to be more versatile so it lives up to its designation as a "swiss army knife system". Tarzan original's equal grouping in the main triplet is a simplification that allows cancellation without EDI's but it comes at a cost. That cost is not being able to switch the system to other games like Spanish 21, countable Blackjack side bets, and countable Baccarat side bets without having to overhaul it. With TOR I'm able to walk into a casino and attack any of the aforementioned three as well as regular blackjack, optimally and without having to change the core system. For example moving from Blackjack to Spanish 21 with Tarzan original is not possible because it breaks the equality between the main triplets that allows his "there's always a zero in one of the groups" simplification method. Also in Tarzan original just like with traditional systems, moving to Spanish 21 you have to recalibrate your eyes and deck estimation skills to deal with the smaller decks which can lead to errors. With TOR you always know exactly how many have come out, whether that is the exact number, how many quarter decks, half decks, 3/4 decks or full decks have been played. All you need is to use the Spanish 21 EDIs that are slightly different from the Blackjack EDI's:


SP21 EDIs (difference between Blackjack and SP21 EDI highlighted in red)
quarter deck (or 12 cards) 22-22-31
half deck (or 24 cards) 44-44-62
3/4 deck (or36 cards) 66-66-93
Full deck (or 48 cards) 88-88-12.4


Also, the larger groupings of Tarzan Original reduces the P.E. of the system as there is less information inherently when you put more ranks into a group. This is also why almost all traditional card counting systems have terrible to average playing efficiencies and are far from optimal as far as playing your hand is concerned (Eric Farmer did some research on this which he published in an article on his site). You essentially have only two groups (or rather one because this information is merged into a single running count) in a traditional system everything goes into a low group {2,3,4,5,6,} or a high group {tens and aces}. There really is no third group as far as playing is concerned because you have no information on {7,8,9} due to them not being counted. Get dealt any stiff hand with a traditional system like Hi-Lo and you’re basically flying blind so to speak. Hi-Lo plays relatively poorly to average for most hands except for ten splits and the like which most players don’t employ anyways because of their heat inducing effects. Or my favorite A7 v 2. Your best chance of successfully doubling this hand requires knowledge of {2,3} removed however you don’t have this information with any of the traditional systems or the 4 rank groupings of Tarzan original. You need more granularity to play more efficiently. However too much granularity is just as bad as it causes an exponential increase in the complexity (like Arnold Snyder’s “Snyder’s Folly” count). The increased granularity of TOR comes at the cost of slightly higher complexity as you state, but it’s a price I’m willing to pay to play close to optimally at a wide variety of games (including non-blackjack type ones like baccarat side bets). I also have a version of TOR where I further “atomized” the {6,7} and {8,9} groups into one rank groups. This turns the DC triplet (of form XX-XX-XX) into a DC quartet (of form XX-XX-XX-XX) but I have never used it because the complexity it adds is simply not worth it in my opinion (I estimate it only moves the Blackjack PE from ~.92 to ~.94). Like everything else, all it takes is practice, dedication and familiarity to become proficient at any skill. After months of practice, I can keep my TOR count almost at same pace I could keep Hi-Lo when used it first starting out.

Only a small niche segment in the AP community (I'm only aware of you, myself and Tarzan at this point) would ever bother with such "extra work", but like griffin said "for the compleat card counting fanatic" going this extra mile may be worth it. If your playing style is the Blackjack Apprenticeship, spreadzilla 1-40, where you play for the back off, or you only backcount the game and play only when you have the advantage, or if you never play double deck, or you only play Blackjack and not SP21 or side bets then Hi-Lo should be sufficient and moving to a DC based count would probably only modestly increase your win rate (in Blackjack that is). In all other scenarios, I predict (will be verified by sims) you should see quite an improvement for card counting with DC counts and should be able to make a killing with TOR-Shuffle-PAO and TOR-Sequence-PAO as the overall edge is much much higher for these approaches to advantage play.

[Secretariat]

1) Very impressive, MIDNIGHT. A BJ system that can be applied to other games and side bets.
Tarzan did quite well with the Lucky Lucky side bets, but no mention of Baccarat. Maybe Spanish 21.

2) Moses, Jack Jackson, BJ Genius007, Freightman were/are also quite interested, in P.E. Flash too. He estimated Tarzan PE to be at 0.94. Someone believed it was up at near .99. Can't remember who. Might have been Flash again.

3) Gronbog took a year to program and simulate the complex Tarzan count. When will we find out about your SCORE?

4) As far as I know the separations of 23s from 45s, and 67s from 89s, were not considered in the 2017 Gronbog sims.
But Tarzan was able to "see" the imbalances when they were significant.
That's why I am convinced that he performed at a higher level than his sims showed.

5) It looks like your TOR system is better defined than T-count for the programmer/simulator.
I would not be surprised if TOR yields the best SCORE ever seen, the best being Tarzan Advanced plus Key cards.

6) Regarding memory, did you consider, or try, visual "recording". In the sense that visual awareness is such that some cards are side counted visually and almost unconsciously every time they appear, so that at any time in a shoe, you can say how many cards of a specific denomination have been played?

7) How do you come up with a .92 PE for TOR?

[JesterCW]

Midnight, I’m starting down the rabbit hole of writing simulation code for my own personal understanding and challenge. I have one clarification that I think I understand but want to make sure — whenever I see a letter in your TOR system, that’s an EDI that’s been applied, right?

In the used example, the 8D is throwing me off — is this a negative?

I’ve gone and gotten myself downright confused. If you could be so kinda can you step your way from:
24 23 - 28 16 - 44 12
to
43-8D-42-E

I’m having trouble seeing how it works. Just take the {2,3} grouping of 24. How does this go to 4 using the half deck EDI? Sorry for the confusion.

nvm — got it. Brain hurts today, lol.

[JesterCW]

Also, as the 7 is included in the 67 group and for hi-lo the 7 would be ignored, how do you get an accurate running count? Just trying to think through it!

[Midnight]

Midnight, I’m starting down the rabbit hole of writing simulation code for my own personal understanding and challenge. I have one clarification that I think I understand but want to make sure — whenever I see a letter in your TOR system, that’s an EDI that’s been applied, right?

In the used example, the 8D is throwing me off — is this a negative?

I’ve gone and gotten myself downright confused. If you could be so kinda can you step your way from:
24 23 - 28 16 - 44 12
to
43-8D-42-E

I’m having trouble seeing how it works. Just take the {2,3} grouping of 24. How does this go to 4 using the half deck EDI? Sorry for the confusion.

nvm — got it. Brain hurts today, lol.

Hi Jester, since we use the half deck EDI and we applied it five times (means we subtracted it from our original DC count) you can just sum the EDI five times plus the TOR count that remains after the cancellation (simplification) and you will arrive back at the original Deck composition. I forgot to mention that a letter in the main triplet represents a negative number so in this example D is -4 since d is the fourth letter of the alphabet. See example below, if you sum the EDIs and the numbers of each group you will arrive back at our original:

43 - 8(-4)- 4 2 ............ TOR count after EDI reduction/cancellation
+44 - 4 4 - 8 2 .............EDI applied 1st time
+44 - 4 4 - 8 2 .............EDI applied 2nd time
+44 - 4 4 - 8 2 .............EDI applied 3rd time
+44 - 4 4 - 8 2 .............EDI applied 4th time
+44 - 4 4 - 8 2 .............EDI applied 5th time
24 23 - 28 16 - 44 12 ............ Original DC count

There was another way to check as well. Since we know that our EDI count is 5 based on E and we have the half Deck EDI's all we had to do was multiply each group number in the half deck EDI by the EDI count and then add the TOR count.

half deck EDI 44 -44 -82
EDI count = 5

Multiply EDI count by each half deck EDI value : 4 * 5 4* 5 - 4 * 5 4 * 5 - 8 * 5 2 * 5
Equals 20 20 - 20 20 - 40 20
Now add TOR count + 4 3 - 8 D(-4) - 4 2
Give you original Deck composition 24 23 - 28 16 - 44 22

The EDI reduction is replicating the effects of cancellation in a traditional card counting system. Without it, the numbers in each group would just keep growing until we got to 48 48 - 48 48 - 96 24 at the end of the shoe. This would render the system useless in practice as you can imagine the difficulty that would arise when trying to compute a running count with numbers this large. Now lets calculate the TKO running count for both the TOR count and the original DC count and show that they are equal. This is proof that applying the EDI does not change the running count in an arbitrary system in anyway whatsoever.

{ 2,3}{4,5} {6,7} {8,9} {Tens} {Ace}
TKO K values : 1 1 - 1 0 - (-1) (-1) ........(a)
TOR count: * 4 3 - 8 D(-4) 4 2 ........(b)
Multiply a*b then sum 4 + 3 + 8 + 0 + (-4) + (-2) + 10 ( remember as TKO is unbalanced we have to add in E = 5 (EDI Count) * 2 (unbalance of 2 per half deck for TKO)
RC = 19

{ 2,3}{4,5} {6,7} {8,9} {Tens} {Ace}
TKO K values : 1 1 - 1 0 - (-1) (-1) ........(a)
Original DC count: * 24 23 - 28 16 - 44 2 2 ........(b)
Multiply a*b then sum 24 + 23 + 28 + 0 + (-44) + (-22) + 10 ( remember as TKO is unbalanced we have to add in E = 5 (EDI Count) * 2 (unbalance of 2 per half deck for TKO)
RC = 19


Both running counts are equal in TKO even though the DC numbers in TOR and the original provided with the actual cards removed are very different. Therefore, the K values and EDI’s “map” the TOR count to TKO. If two players were at the same table and one was counting with TKO and the other was using TOR with TKO for betting (TKO K values) and they compared their running counts after each round they would be in exact agreement 100% of the time. This same principle applies to moving from TOR to any card counting system that can be represented by the groupings and unlocks the ability to customize everything for the hand or situation being placed. All you must do is select the k values with the highest correlation to the EORs for the play at hand to maximize the expectation. Also of course you wont be doing all the above in a real game. You would just add the first three numbers in your TOR count and subtract the last two. With practice you get superfast at it. That is not difficult. The difficulty in these DC counts is remembering the count from round to round and that’s where the PAO systems help and are the solution. For the TOR count above my ‘person’ for 43 is Thor there is no ‘action’ because of the D in the middle triplet, when I have a number and letter I represent the number as the number and the letter as an animal in this case a dinosaur because the letter is D, the ‘object’ for 42 in my PAO table is represented by a lifeboat. So my image for this Deck Composition is “THOR and 8 dinosaurs in a lifeboat”. The PAO images are so absurd that you can never forget them even if you leave the table and go get lunch you will still remember the last one hours later. If I was playing and I took a hit to whatever hand I had and got a five I would immediately update the image to 44 which is black widow in my system and the image would the be ‘Black Widow and 8 dinosaurs in a lifeboat’ or 44-8D-42.

[Midnight]

Also, as the 7 is included in the 67 group and for hi-lo the 7 would be ignored, how do you get an accurate running count? Just trying to think through it!

The variant of TOR we are using in our example does not map directly to Hi-Lo as we have {6,7} in the same group. This is why I use another Level 1 count for betting called TKO. If you wanted to use Hi-Lo you would have to "atomize" the {6,7} group which means put each rank in its separate group. The version we have been using is TOR{6,7}{8,9} the one you would need if you wanted to use Hi-Lo for betting would be TOR{6}{7}{8,9} which have EDI's as follows:

TOR{6}{7}{8,9} EDIs -
{6}{7}
quarter deck (or 13 cards) 22 - 11- 24-1
half deck (or 26 cards) 44 - 22- 48-2
3/4 deck (or 39 cards) 66 - 33- 3.12-3
Full deck (or 52 cards) 88 - 44- 4.16-4

As you can see because we have now separated the {6} and {7} into individual groups the middle triplet now represents {6} and {7} and everything else is shifted one place to the right. This also moves the ace one spot over and turns the DC triplet into a quartet. As stated in an earlier post with each atomization the complexity increases. Fortunately TKO has a higher betting correlation than Hi-Lo so the only reason to do this would be if you weren't comfortable with unbalanced systems or didn't want to learn it. Otherwise you could use the original TOR system and just estimate the Hi-Lo running count by assuming an even distribution of 6 and 7 in the {6,7} group and dividing by 2. This would be accurate 99% of the time and would only be off if there was a large and unequal distribution in {6,7}. Fortunately, both these versions of TOR are extremely similar with TOR{6}{7}{8,9} being slightly more powerful because of the higher resolution achieved by "atomizing" {6,7}. But that comes at a cost as stated.

[Midnight]

1) Very impressive, MIDNIGHT. A BJ system that can be applied to other games and side bets.
Tarzan did quite well with the Lucky Lucky side bets, but no mention of Baccarat. Maybe Spanish 21.

2) Moses, Jack Jackson, BJ Genius007, Freightman were/are also quite interested, in P.E. Flash too. He estimated Tarzan PE to be at 0.94. Someone believed it was up at near .99. Can't remember who. Might have been Flash again.

3) Gronbog took a year to program and simulate the complex Tarzan count. When will we find out about your SCORE?

4) As far as I know the separations of 23s from 45s, and 67s from 89s, were not considered in the 2017 Gronbog sims.
But Tarzan was able to "see" the imbalances when they were significant.
That's why I am convinced that he performed at a higher level than his sims showed.

5) It looks like your TOR system is better defined than T-count for the programmer/simulator.
I would not be surprised if TOR yields the best SCORE ever seen, the best being Tarzan Advanced plus Key cards.

6) Regarding memory, did you consider, or try, visual "recording". In the sense that visual awareness is such that some cards are side counted visually and almost unconsciously every time they appear, so that at any time in a shoe, you can say how many cards of a specific denomination have been played?

7) How do you come up with a .92 PE for TOR?

Hi Secretariat, glad to know its just not us taking the multiparameter approach. Yes I agree that because TOR is more numbers based it should be easy to code for a programmer. There is no "seeing" imbalances or anything like that. You get your count, apply the K values to map to your play count or bet count, and then you have the custom running count that most highly correlates to that the play, betting EORs, side bets, etc. There is really no guesswork or intuition involved like there was in Tarzan's original system.

I have never tried the visual recording approach to memorization but it sounds interesting. Can you describe how its done exactly?

I'm really excited to see what the sims reveal but have been so engrossed on working on methodology for TOR-Shuffle-PAO ad TOR-Sequence-PAO that I haven't devoted the time to pursuing it as I should. Professor Persi Diaconis's book "The mathematics of shuffling cards" is my constant companion and goes everywhere with me, lol. This conversation though has renewed my interest and I will try to get the sims done asap so we could know for sure how the system performs and get the independent verification of the experts.

[JesterCW]

Now...how do I deal with wanting to learn this before I go to a casino again? lol

I think there will still be some Hi-Lo in my future for the immediate, but I'm developing and practicing my PAO system and think I'm hooked on learning this methodology. The one part that I'm worried about a bit is being able to quickly count every pair and update with the EDIs in my head when needed, but I bet with practice this will seem like less of a task! Thanks Midnight. I'm excited to be working on this!

[Secretariat]

I have never tried the visual recording approach to memorization but it sounds interesting. Can you describe how its done exactly?

It's basically just watching for a specific denomination (let's say 9s) without really side counting and clogging your brain. But your brain records it almost unconsciously so that you can tell at any depth how many 9s are left in the shoe. I have tried it a few times (not in casino) and I was pretty much on track. I think there is an upside to it. Won't be my main focus. I was asking out of curiosity since you obviously have experimented quite a few tricks.

[Secretariat]

Now...how do I deal with wanting to learn this before I go to a casino again? lol

I think there will still be some Hi-Lo in my future for the immediate, but I'm developing and practicing my PAO system and think I'm hooked on learning this methodology. The one part that I'm worried about a bit is being able to quickly count every pair and update with the EDIs in my head when needed, but I bet with practice this will seem like less of a task! Thanks Midnight. I'm excited to be working on this!

If you go HiLo, and considering your dedication and ability, you may want to try side counting 789s as a group, or just 9s, or just 7s, or the surplus of 9s over 7s. And there's another level beyond that.

[JesterCW]

I like that. It’s a step in between HiLo and TOR. I now have homework to do, lol. Time to research the side counts and why I might choose one over another. Taking 789 as I’m already “ignoring” the group and turning it into a PAO side count could work, and could probably take advantage of the EDI reductions. That said, is keeping an Ace side count more beneficial? Etc. my work is cut out for me.

I’m also vastly overthinking heat and cover. Some seem to just full on attack (this seems better — maybe — if you have some money and want to do some damage before you get the boot). I most certainly don’t have that bankroll. I’d like to have longevity for now and am happy playing at lower EV while managing risk and growing my bankroll as I have other work — but also have 24/7 freedom in all my jobs, so I could travel at will. So, with longevity in mind it seems advantageous to take advantage of multiple systems or just buckle down and learn a new count and TOR alongside my PAO system asap (99% doing this). It’ll be good practice to keep a side count visually. I have a very modest spread planned (10-50 next trip, 10-150 once I’m a little more comfy). Red chips if I can find them. I’m going to go ahead and learn all the deviations as I’ve already learned 120 or 130 or something so the rest seem like just another Anki deck. I’m worried I won’t always be able to Wong out easily so figure I might as well know all the negative counts. Hopefully the spread of deviations won’t make me look more AP than not. I’m probably not splitting tens but will mess with A9. Maybe? lol. Overthinking.

Thanks again for this start down a path I think will be a bit harder and thus might get some more hours in!

[Secretariat]

I like that. It’s a step in between HiLo and TOR. I now have homework to do, lol. Time to research the side counts and why I might choose one over another. Taking 789 as I’m already “ignoring” the group and turning it into a PAO side count could work, and could probably take advantage of the EDI reductions. That said, is keeping an Ace side count more beneficial? Etc. my work is cut out for me.

I don' want to interfere with your ambition to upgrade to TOR using PAO.

With Hi-Lo and two side-counts Ace/789, I don't think you need mental imagery. If you add a 4th column, then it is probably required.

If I use Hi-Lo 3 columns I start a 6D shoe with 0-24-72. All three columns are just sounds in my head.
Like zero/twentyfour/seventy-two.

The first column his Hi-Lo, RC. The two others are just countdowns. After one deck the perfectly neutral count would be 0-20-60. Now if it's 5-20-60, then it's TC1.

5-24-60 would mean 4 extra aces with 5 decks left, so it's almost TC2.

Like Freightman often mentions, there's no need to play perfectly robotic TC. With five decks left, you could place a TC2 bet at 5-24-60 and you would not be far off.

I won't say a word about using EDI reductions with Hi-Lo as only Midnight masters the concept.
I think Hi-Lo modified RC/TC conversion is suffficient.


EDIT
I just looked back at the thread Jester. Mastering the main HiLo count, one column, would be my priority. Then side count the ace and master it before going 3 columns. Now I don't know what is the learning curve for the TOR system and in how many steps it can be learned. Only Midnight can help you for that.