True Count: Round, Floor or Truncate? CF. CV Blackjack and CVData

[Boardwalker]

Are the following correct:

To FLOOR any mixed number, drop the fractional or decimal part and subtract 1 from the integer part. 1¼ and 1¾ floored are both equal to 1. -1¼ and -1¾ floored are both equal to -2

To ROUND a positive mixed number, drop the fractional or decimal part if it is < .5 and keep the integer part unchanged; drop the fractional or decimal part if it is >= .5 and add 1 to the integer part. 1¼ rounded is equal to 1. 1¾ rounded is equal to 2.To ROUND a negative mixed number, drop the fractional or decimal part if it is < .5 and keep the integer part unchanged; drop the fractional or decimal part if it is >= .5 and subtract 1 from the integer part. -1¼ rounded is equal to -1. -1¾ rounded is equal to -2.

To TRUNCATE any mixed number, drop the fractional or decimal part and keep the integral part unchanged. 1¼ and 1¾ truncated are both equal to 1. -1¼ and -1¾ truncated are both equal to -1

My understanding of floor for negative numbers seems different than the rule that appears on page 28 of the current (2011) Casino Verite Blackjack manual ("Fractions are always removed."). The rule that appears on page 17 of the current (2010) CVData Reference Manual is ("Rounds a number down."). My understanding is that these rules are inconsistent for flooring negative mixed numbers, e.g. CVBJ will floor -1¼ and -1¾ to -1, while CVData will floor -1¼ and -1¾ to -2, and, that the rule in the CVData Reference Manual is correct. Is my understanding correct? Do CVBJ and CVData calculate negative true counts differently, if floor is the method of converting them to integers.

It is my understanding that developers of card counting strategies calculate playing index numbers with an understanding that the user will specifically floor, round or truncate, when doing true count division. Should true counts be floored, rounded or truncated for the following published index numbers:

High-Low from Stanford Wong's Professional Blackjack (The 1981 and 1994 books have some different indices.)

High-Low from Don Schlesinger's Blackjack Attack

High-Opt I and High-Opt II from Lance Humble's World's Greatest Blackjack Book and from the system reports that he sold separately

Arnold Snyder's High-Low Light and Zen Count from his latest edition of Black Belt in Blackjack

Lawrence Revere's systems

Ken Uston's systems

I realize that some of these systems are quite old and how they were developed and intend to be used may not be known or knowable.

[Norm]

Flooring rounds down. ProBJ 1981 floored, 1994 truncated.

BJA Chapter X indices floor.

Most books don't specify.

[Boardwalker]

Thanks and more questions:

Should the true count be rounded, floored or truncated for betting?

Which High-Low indices are the best: PBJ 1981 floored, PBJ 1994truncated or BJA CX floored?

Which method of obtaining integer true count should be used if the book doesn't specify, e.g. the Hi-Opt systems?

Thanks in advance.

[Norm]

Should the true count be rounded, floored or truncated for betting?

Same as playing

Which High-Low indices are the best: PBJ 1981 floored, PBJ 1994truncated or BJA CX floored?

BJA

Which method of obtaining integer true count should be used if the book doesn't specify, e.g. the Hi-Opt systems?

Who knows?

[Dog Hand]

Are the following correct:

To FLOOR any mixed number, drop the fractional or decimal part and subtract 1 from the integer part. 1¼ and 1¾ floored are both equal to 1. -1¼ and -1¾ floored are both equal to -2. <snip>

Boardwalker,

This is not correct for positive numbers. As you can see, you failed to "subtract 1" for your two positive examples.

Instead, think of "floor(X)" as "what is the largest integer that is less than or equal to X?"

Hope this helps!

Dog Hand

[moo321]

Almost completely immaterial difference.

[Three]

There are practical differences between these options.

Round:
All true count buckets are the same size. You must do the math to 1 more significant digit, the first decimal, to see if it is higher or lower than .5 (at least in the cases it is not obvious to you). This will slow you down a little which costs you money over time.
Truncate:
The true count bucket for TC 0 is twice the size of any other bucket (-0.99999 to +0.99999). With the other styles playing an index of 0 requires no math since you just look to see if it is negative or not. This is not true with truncating. Then the frequency of this over sized bucket includes both the most frequent buckets for any other style of determining a TC. It is pretty easy to use as you stop your math at the decimal.
Floor:
This has equal sized buckets for all TC's. It stops the math at the decimal. It is the best from a practical stand point.

The important thing is to use the same system that your indices and ramp where designed for. If you don't know or would prefer another style of determining the TC use CVData to generate your own indices for that style of determining a TC and generate your ramp to also use the same technique.