How to use True Count

#1
I understand how to calculate True Count (TC) - divide Running Count (RC) by estimated decks left. However, I don't understand how to use TC in a practical sense.

For example, in an eight deck game if the RC is +8 and there are 4 decks left then the TC is +2. Now when the dealer deals do I re-start the count at +2 or +8?

Do I estimate the TC at the start of every new deal?

Thankyou
 

KenSmith

Administrator
Staff member
#2
Continue your running count at +8, and recalculate the true count whenever you need to make a betting or playing decision.
 
#3
The true count is the count you use to make your playing decision. You always count with the running count but when its your turn you need to convert to the true count to get the true ratio of high cards to low cards in the deck. The running count alone is useless
 
#4
Is TC counted too strongly?

standard toaster said:
The true count is the count you use to make your playing decision. You always count with the running count but when its your turn you need to convert to the true count to get the true ratio of high cards to low cards in the deck. The running count alone is useless
I have a question on the TC. It seems to me, with my limited experience, that variations in count are more localized than a TC number would imply. In other words, if I get a +10 count in the early portion of a 6 card deck I would not expect to wait until the end of the deck to get a 0 balance... I'd expect it to occur relatively soon. What's your take on this?
 

sagefr0g

Well-Known Member
#5
bd99 said:
I have a question on the TC. It seems to me, with my limited experience, that variations in count are more localized than a TC number would imply. In other words, if I get a +10 count in the early portion of a 6 card deck I would not expect to wait until the end of the deck to get a 0 balance... I'd expect it to occur relatively soon. What's your take on this?
this link shows how true counts tend to behave:
http://www.blackjackincolor.com/truecount1.htm
 
#6
sagefr0g said:
this link shows how true counts tend to behave:
http://www.blackjackincolor.com/truecount1.htm
sagefr0g, thanks for the link. Some very useful info there but it doesn't quite address my question on localized variance. Basically I see emperically that a high + count (ignoring TC) will quickly reset to 0 or near zero long before the end of a 6-deck game. So if this is true wouldn't one potentially miss out of a local spike of say +10 in the beginning of a shoe because 6 (or 5) decks are outstanding?
 

sagefr0g

Well-Known Member
#7
bd99 said:
sagefr0g, thanks for the link. Some very useful info there but it doesn't quite address my question on localized variance. Basically I see emperically that a high + count (ignoring TC) will quickly reset to 0 or near zero long before the end of a 6-deck game. So if this is true wouldn't one potentially miss out of a local spike of say +10 in the beginning of a shoe because 6 (or 5) decks are outstanding?
well, i think if say you have a RC=10 in the beginning of a shoe and six or five decks still to be dealt, you'd have a TC=1 or TC=2, so you'd probably be betting up some on that. so if the TC then drops to zero you wouldn't be missing anything.

maybe i don't understand fully what your meaning by "variations in count are more localized than a TC number would imply." but if you look at this link http://www.blackjackinfo.com/bb/showpost.php?p=117353&postcount=18
you can see how different ways of calculating your TC can give different perceptions of what's going on with the TC, sort of thing. (ie. calculating TC=RC/(#cards left to deal) as opposed to TC=RC/(#decks left to deal) )

then there is the true count theorem. you might find this interesting:

The True Count Theorem
by Abdul Jalib

--------------------------------------------------------------------------------
Exerpt take from the following reference in DejaNews:
Subject: Proof that everyone has the same shot at getting tens
From: [.. (Abdul Jalib M'hall)
Date: 1996/07/30
Message-ID: <..>
Newsgroups: rec.gambling.blackjack


--------------------------------------------------------------------------------
The True Count Theorem
----------------------
The following is a mathematical theorem:

Theorem: the expected value of the true count after a card is revealed and removed from any deck composition is exactly the same as before the card was removed, for any balanced count, provided you do not run out of cards.

See the proof at the end of the article. Expected value is a precise mathematical term defined as the mean average, which is computed by summing the probability of an event times the value of that event, over all possible events. So the expected value of the true count after drawing a card is the summation of the probability of drawing each card times the value of the true count after drawing that card.

One consequence of this theorem is that the expected value of the count after any number of cards have been revealed and removed will be the same as before, and so the expected value of the count after a round has been dealt will be the same as before. And when you get a constant number of rounds, the expected true count after each is zero. Since the true count starts at zero, the overall expected value of the count is zero, when you get a fixed number of rounds.


The Running Count
-----------------

The theorem applies only to the true count, not to the running count. The running count does not obey the same laws of as the true count.

With regards to the effects of other players at the table and the tendency of the round to stop with a big card, much confusion stems from a mistaken assumption that the behavior of the true count is the same as the behavior of the running count.

The running count must be zero at the end of the deck. Therefore, drawing cards in high counts tends to cause the running count to fall, and drawing cards in low counts tends to cause the running count to fall. But the expected true count is unchanged.


The Proof
---------

Theorem: the expected value of the true count after a card is revealed and removed from any deck composition is exactly the same as before the card was removed, for any balanced count.

Proof:

Let Wi be the Weight of the card of rank i, i.e., the count value.

Let Ni be the Number of cards of rank i already revealed, counted, and removed from the deck.

Let Si be the Starting number of cards of rank i in a full deck.

Let Li be the number of cards of rank i currently Left in the deck, i.e., Li=Si-Ni

Let C be the number of Cards remaining in the deck, i.e., C=sum_over_i{L}

Let R be the Running count.

Let T be the True count, i.e., T = R/C

Assume that the count is balanced, i.e., sum_over_i{W*S}=0.


Need to show that: T = sum_over_j{((R+W[j])/(C-1))*(L[j]/C)}, that is to say that the true count after a card j is revealed, removed, and counted, averages out to the same as the true count before. The average is computed by adding up the true counts after each card is drawn weighted by the probability of each card being drawn.

sum_over_j{((R+W[j])/(C-1))*(L[j]/C)} =

(1/(C(C-1)))sum_over_j{(R+W[j])L[j]} =

(1/(C(C-1)))sum_over_j{(RL[j]+W[j]L[j])} =

(1/(C(C-1)))sum_over_j{(RL[j]+W[j](S[j]-N[j])} =

(1/(C(C-1)))sum_over_j{(RL[j]+W[j]S[j]-W[j]N[j])} =

(1/(C(C-1)))(sum_over_j{RL[j]} + sum_over_j{W[j]S[j]} - sum_over_j{W[j]N[j]}) =

(1/(C(C-1)))(R*sum_over_j{L[j]} + 0 - R) =

(1/(C(C-1)))(R*C-R) =

(1/(C(C-1)))(R(C-1)) =

R/C = T

. . . QED

Again...

Theorem:
The expected true count after a card is revealed and removed from any deck composition is the same as before the card was removed, for any balanced count, provided you do not run out of cards.

Corollary:
The expected true count after any number of cards are revealed and removed from any deck composition is the same as before the cards were removed, for any balanced count, provided you do not run out of cards.

Corollary:
The expected true count after a round is the same as before the round, for any balanced count, provided you do not run out of cards.


this link has some more stuff on variation of the counts.
http://www.blackjackincolor.com/truecount8.htm
and this one:
http://www.blackjackincolor.com/penetration9.htm
 
Last edited:
#9
After some more playing I think I can express my thought on TC a little more clearly with an analogy...

Statistically, over the long haul one would likely calculate the best long term return in the stock market by buying an index fund and holding it for a long time. But with some effort buying and selling shorter term stocks could make a lot more money.

The idea here is that even though over 1000s of hands the benefit of TC may be evident I'm wondering if more localized higher betting based more on the actual count would be more rewarding. Could the standard deviation on temporary positive spikes on actual counts bear this theory out? I typically see 2 or more positive count "rallies" in a shoe so wouldn't it be best to bet based on those rather than the over all shoe TC?

Again, I'm a newbie so ...
 

Sonny

Well-Known Member
#11
bd99 said:
The idea here is that even though over 1000s of hands the benefit of TC may be evident I'm wondering if more localized higher betting based more on the actual count would be more rewarding.
What's the difference between the true count and the actual count?

-Sonny-
 

Canceler

Well-Known Member
#12
Sonny said:
What's the difference between the true count and the actual count?
The following is a total guess on my part:

bd99 may be talking about True Count vs. Running Count. Specifically, a situation where the RC rockets upward, but the TC does not, because of a large “decks remaining” divisor. He thinks he may have an advantage here even though the TC indicates he may not. This strikes me as dangerous.

On the other hand, this comment in a different thread by QFIT about KO...

QFIT said:
Counting the seven improves playing efficiency, betting correlation and insurance correlation. It is also helped by the fact that more precise betting is possible with running counts in unbalanced systems.
...makes me think bd99 might want to consider using an unbalanced count.
 

EasyRhino

Well-Known Member
#13
Note that unbalanced counts "protect" against overbetting too early by starting with a substantially negative initial RC, which must be overcome before the betting starts.
 

sagefr0g

Well-Known Member
#14
bd99 said:
After some more playing I think I can express my thought on TC a little more clearly with an analogy...

Statistically, over the long haul one would likely calculate the best long term return in the stock market by buying an index fund and holding it for a long time. But with some effort buying and selling shorter term stocks could make a lot more money.
well i dunno about the stock market. but how i look at it, some stock and the decisions you make on going short or long would rationally be based on what you know about the business and the outside the business influences that can help the business or hurt it. theoretically one might know a heck of a lot about some business venture but likely never could one come to terms with all those outside influences. the point being one can't accurately normalize a stocks statistics. you try and you end up getting wild unexpected fat tails, sort of thing.
this sort of thing isn't a problem for a given blackjack type of game because the frequency for which true counts present is normalized over time. see the chart in this link: http://www.blackjackinfo.com/bb/showpost.php?p=117223&postcount=3
so in simplistic terms, comparing the stock market and blackjack one needs to be careful because of the apples and oranges effect. :)

as a side note, i'd like to make a correction in a prior post i made:
http://www.blackjackinfo.com/bb/showpost.php?p=120212&postcount=7
(ie. calculating TC=RC/(#cards left to deal) as opposed to TC=RC/(#decks left to deal) ) should have read (ie. calculating TC=RC/(#decks left to deal) as opposed to TC=RC/(#decks [rounded down] left to deal) )
bd99 said:
The idea here is that even though over 1000s of hands the benefit of TC may be evident I'm wondering if more localized higher betting based more on the actual count would be more rewarding. Could the standard deviation on temporary positive spikes on actual counts bear this theory out? I typically see 2 or more positive count "rallies" in a shoe so wouldn't it be best to bet based on those rather than the over all shoe TC?

Again, I'm a newbie so ...
to me it's human nature to wonder about things like this. it's our normal state of being to be more comfortable with and want to make decisions about controlling events in the short term by our decisions and judgments, while relying on preconceived plans that rely on carefully crafted probabilities and statistics of things essentially going on behind our backs tends to be unnatural and generally leads to feelings of discomfort.
so yes it's very tempting, the idea of seeking profit from standard deviation. it's so 'out there in your face' sort of thing. problem is, standard deviation is really just luck and lady luck has two faces, so that over time those two faces make us end up with expectation (what ever that maybe)
instead of luck.
so if i'm understanding what your getting at, well your not alone, all the above said, i still find such stuff fascinating as you might guess by the following links:
http://www.blackjackinfo.com/bb/showthread.php?t=11094
http://www.blackjackinfo.com/bb/showthread.php?t=12979
 
Top