Don, can you please direct or explain why flooring was your method of choice in BJA3? Does it make a difference in overall expectation for TC'ing balanced counts (I'm aware that flooring is necessary for TC'ing unbalanced counts)? Thanks.
Don, can you please direct or explain why flooring was your method of choice in BJA3? Does it make a difference in overall expectation for TC'ing balanced counts (I'm aware that flooring is necessary for TC'ing unbalanced counts)? Thanks.
> Don, can you please direct or explain why
> flooring was your method of choice in BJA3?
> Does it make a difference in overall
> expectation for TC'ing balanced counts (I'm
> aware that flooring is necessary for TC'ing
> unbalanced counts)? Thanks.
Norm, Cacarulo, and Karel Janecek all told me that this was the best methodology for generating and using indices. So, I believed them! :-)
Don
Shame shame Don.. aren't you the one who always challenge conventional thinking? ;-)
So, Norm, Cac, and Karel? What are your reasons behind using Flooring?
> Norm, Cacarulo, and Karel Janecek all told
> me that this was the best methodology for
> generating and using indices. So, I believed
> them! :-)
> Don
> Shame shame Don.. aren't you the one who
> always challenge conventional thinking? ;-)
No, not always. And, I certainly wouldn't challenge something that those three all agreed on. You'd have to be crazy!
> So, Norm, Cac, and Karel? What are your
> reasons behind using Flooring?
Truncating is out, because of the problem with the zero interval; it's too big. All values from -0.999 to +0.999 become 0, so that interval is twice as wide as all the others, and that's not a good idea.
Between flooring and rounding, we get different frequencies and edges for the TC values, and I believe flooring works out better, but I don't remember why.
Don
...but I don't think that flooring nor truncating are satisfactory soloutions. It can lead to errors of up to 0.9 index points whereas simple rounding can produce errors of 0.5 points at worst.
I can understand that you have to simplify things for the reader and therefore use integers but I would not recommand this to professionals.
I personally use decimals both for indexes and for calculating the TC.For example, an RC of 4 with 2.5 decks to go gives a TC of 1.6. Its not +1 nor is it +2.
Being inaccurate on both ends will not necessarily eliminate the error. It might even square it.
Francis Salmon
When you play, during play, are you actually dong the division to get the 1.6 or is it simply rote (i.e. 2 plus 2) for you?
Also, what do you do with 1.6? Have you simmed your own indices to the tenth place?
Thanks so much.
> I personally use decimals both for indexes
> and for calculating the TC. For example, an
> RC of 4 with 2.5 decks to go gives a TC of
> 1.6. Its not +1 nor is it +2.
So what would your bet be at a TC of 1.6? What about 3.4? Would you add an extra 0.4 untits to your bet or do you use the decimals strictly for playing decisions?
I do the same thing when I play DD, but I have yet to find a good way to use the extra information after the decimal point. My indices are all whole numbers. I would love to get some advice on this.
-Sonny-
> When you play, during play, are you actually
> dong the division to get the 1.6 or is it
> simply rote (i.e. 2 plus 2) for you?
No. The equivalent to dividing by 2.5 is multiplying by 0.4.So I multiply the RC by 4 (we all had to learn the multiplication table by heart in primary school!) and then put a decimal point.This is much quicker and easier than dididing.Likewise with 5 decks remaining I would multiply with 0.2. That's how I get decimals from the calculation, and it would be a pity to truncate them,wouldn't it?
> Also, what do you do with 1.6? Have you
> simmed your own indices to the tenth place?
There are a number of indices at and around 1.6,e.g. doubling 8v6 (1.6), doubling soft17 versus 2 (1.5),standing on 12v3(1.4) and in negative territory not doubling soft13v6(-1.6) or soft14v5(-1.4),not doubling 10v9 (-1.7),hitting 12v5 (-1.7) or 12v6(-1.3), not doubling 9v3(-1.2) and so on.
Furthermore, TC +1.6 signals under reasonnable conditions a slight advantage, so I will raise my bet here.
I don't believe in simulations for determining indices.I got them through my own CA-program, which can calculate EVs depending on TCs. I'm quite happy with one position after the decimal point but I could have it more precisely any time.
> Thanks so much.
You're welcome
Francis Salmon
> So what would your bet be at a TC of 1.6?
> What about 3.4? Would you add an extra 0.4
> untits to your bet or do you use the
> decimals strictly for playing decisions?
An extra 0.6 means an extra advantage of 0.3%.What you bet depends also on the depth of penetration (floating advantage)and of course on your bankroll.
In other words,I don't correlate my bets strictly with TC, I correlate them with the actual advantage.
> I do the same thing when I play DD, but I
> have yet to find a good way to use the extra
> information after the decimal point. My
> indices are all whole numbers. I would love
> to get some advice on this.
I'm not really familiar with DD as I play exclusively shoe games but I certainly think that you should use more precise indices.My CA-program was designed with the infinite deck approach.They work extremely well with multiple decks but their might be some slight distortions for DD because of the increased influence of the initial hand.
Francis Salmon
But then we're back to the age-old question of how complex is too complex? Is it easier to remember 4.3 or just 4? Since the gain is negligable to non-existent(see Fuchs and Vancura on rounded indices), and I'm sure the errors in deck estimation, as well as the inherent inadequacies built into (human level) counting systems to begin with, more than offset going into such detail over index #s. I can't justify memorizing indices to anything more than whole numbers.
My original question is more to satisfy my curiosity than have any meaningful implications. I figured that:
1) Truncating can produce the most errors (per Don, the 3 wise men ;-), and Francis),
2) Flooring was excellent for unbalanced TC, since it reconciles the fact that 0 is no longer the pivot (Brett Harris said it best when he wrote regarding unbalanced counts, "if indices are calculated differently on either side of zero, (it is) like calculating indices differently for Hi-Lo TC=+1 and TC=+3"), but since we are talking about a balanced count where 0 IS the pivot,
3) Rounding seemed the most logical choice.
> ...but I don't think that flooring nor
> truncating are satisfactory soloutions. It
> can lead to errors of up to 0.9 index points
> whereas simple rounding can produce errors
> of 0.5 points at worst.
> I can understand that you have to simplify
> things for the reader and therefore use
> integers but I would not recommand this to
> professionals.
> I personally use decimals both for indexes
> and for calculating the TC.For example, an
> RC of 4 with 2.5 decks to go gives a TC of
> 1.6. Its not +1 nor is it +2.
> Being inaccurate on both ends will not
> necessarily eliminate the error. It might
> even square it.
> Francis Salmon
> An extra 0.6 means an extra advantage of
> 0.3%.What you bet depends also on the depth
> of penetration (floating advantage)and of
> course on your bankroll.
I could be wrong, but it seems like you already have too much ambiguities from deck estimation, using a (level 1?) count, and especially not being able to bet accurate Kelly to utilize the extra info.
> In other words,I don't correlate my bets
> strictly with TC, I correlate them with the
> actual advantage.
Nice~ Let's assume you have a $10,000 bankroll, then you'd bet $100 at 1% advantage, $110 at 1.1% advantage, ... etc. right? There might be some cover potential in this since you're varying your bets even at a TC of 1. I'm going to have to look further into this.
> I'm not really familiar with DD as I play
> exclusively shoe games but I certainly think
> that you should use more precise indices.My
> CA-program was designed with the infinite
> deck approach.They work extremely well with
> multiple decks but their might be some
> slight distortions for DD because of the
> increased influence of the initial hand.
> Francis Salmon
I seem to recall having this same discussion with you on another forum several years ago.
My opinion has not changed. :-)
You are an experienced pro and I respect that, and you obviously have no problem with the mental gymnastics involved in using this degree of precision. Your system works for you.
However, you are creating an island of precision in a sea of approximations. Card counting is based on averages, approximations, and probablilities over millions of hands. Most of the advantage, especially in shoe games, comes from simply chunking out the big bets when we have the edge.
It certainly doesn't hurt anything to be this precise, assuming that your brain is up to the task, but you are deluding yourself if you think you are gaining any significant EV as a result.
I think I even gave you numbers to document the gain in EV,obviously in vain :-).
One thing is sure: the fact that so many players take interest in these decimal indices shows that people are not happy with the whole number solution.
Francis Salmon
Posting Permissions
-*-
Bookmarks