Software simulator and unexpected results

[Kasi]

but as far as the CVBJ log, yes it records splits as two or more hands.

I thought maybe it did and thanks much for the corroboration it actually does.

All that means, to me, is if one played 200,000,000 "hands" with CVBJ, one could not accurately equate the results of 200,000,000 what I perceive as "rounds" in CVCX.

Simply put, it seems to me, and has for a long time, CVBJ and CVCX actually have different underlying assumptions as to the definition of a "round".

Probably because noone actually plays 200,000,000 "hds" or "rds" in CVBJ and actually tries to compare the result to a CVCX sim.

In other words, it won't make much difference whether one assumes a few hundred thousand "rds" or hds" if using CVBJ and most people won't be using CVBJ for the purpose of comparing the actual results of CVBJ to CVCX. Mostly just using it for guesstimating skill factor of counting and betting.

Hopefully, QFIT can either confirm or deny my assumption here that CVBJ treats "split hds" as "2 hds" but CVCX treats it as one "rd". To me, it is pretty clear (less than crystal clear lol) that the 2 softwares treat "split hds" vs "rds" differently.

If so, it's not like it bothers me that CVBJ and CVDATA/CVCX may have different assumptions as to what is a "hand" or a "rd". As long as I know that the output from each software is based on a different assumption and I cannot accurately compare the output of one to the other.

On the other hand, I've always wished all his softwares actually had the same definition of "rd" so I didn't have to worry about adjusting lol.

Silly me, I started off assuming all 3 of his software products had the same underlying assumptions as to what a "round" was. I no longer feel they all share the same assumption.

Hoping to hear from the MAN himself lol

Then I adjust my reality to what he says lol..

 

[QFIT]

CVCX/CVData are simulation software and designed to determine risk, bankroll requirements, advantage, etc. This requires data by round, from the original amount you put on the table until the round is complete. In the case of multiple-hand play, yes the entire original amount is used as this is the only way that hand covariance can be taken into account. Risk is based upon the initial outlay including multiple hands, but additional funds (e.g. splits, doubles, insurance) are figured into wins and losses and therefore standard deviation.

CVBJ is practice software. It is designed to aide the learning process and reduce error rates. It must deal with separate hands.

Comparing results in CVBJ with CVCX/CVData is useless. Win/Lost results in CVBJ have no meaning, unless you play it for decades. The game in CVBJ is a combination drill, not a simulation. The error data in CVBJ is what you should look at.

 

[sagefr0g]

....

Comparing results in CVBJ with CVCX/CVData is useless. Win/Lost results in CVBJ have no meaning, unless you play it for decades. The game in CVBJ is a combination drill, not a simulation. The error data in CVBJ is what you should look at.

i don't mean to be contentious, just i must say this is the type of statement that i hear over and over on this site and it perplex s me.:whip:

i mean i think i get the point underlying this sort of statement, point being i guess that, that is just how statistical stuff and probability stuff is by it's nature, to a great extent pointless beyond 'decades' of data, sort of thing.

so is it just my gambler's fallacy infested mind that makes me believe there is meaningful, useful information to be gleaned from short interludes of data that is a greater part of some whole?

like ok, i know nothing of statistics or probability, just ok, i know there are operations such as exit polls in elections sort of things.
a relatively small amount of data, part of some potential whole amount of data from which predictions are made, sort of thing.

and ok, like with orthodox counting theory sort of thing, i mean don't we take just a pitifully small amount of data at a time and plop down our bets based on that data. i mean what? is that wrong if your not intending to play the game for decades? if not intending to play for decades, if having a hope of more immediate gratification then it's a meaningless and hopeless endeavor?:whip:

 

[QFIT]

i don't mean to be contentious, just i must say this is the type of statement that i hear over and over on this site and it perplex s me.:whip:

i mean i think i get the point underlying this sort of statement, point being i guess that, that is just how statistical stuff and probability stuff is by it's nature, to a great extent pointless beyond 'decades' of data, sort of thing.

so is it just my gambler's fallacy infested mind that makes me believe there is meaningful, useful information to be gleaned from short interludes of data that is a greater part of some whole?

like ok, i know nothing of statistics or probability, just ok, i know there are operations such as exit polls in elections sort of things.
a relatively small amount of data, part of some potential whole amount of data from which predictions are made, sort of thing.

and ok, like with orthodox counting theory sort of thing, i mean don't we take just a pitifully small amount of data at a time and plop down our bets based on that data. i mean what? is that wrong if your not intending to play the game for decades? if not intending to play for decades, if having a hope of more immediate gratification then it's a meaningless and hopeless endeavor?:whip:

A couple points:

1. One thing you learn from polls is that they are wrong a lot.

2. Long-term results are an excellent source of information about possible short-term results. Short-term results are a terrible source of information about long-term results.

 

[sagefr0g]

A couple points:

1. One thing you learn from polls is that they are wrong a lot.

2. Long-term results are an excellent source of information about possible short-term results. Short-term results are a terrible source of information about long-term results.

excellent answer imho, and thank you for taking the trouble.

one more question if i may.

i don't really know anything about exit polls either, are they scientific in nature and are they considered mathematically proper methods?

edit: hmm i found this source, i guess it's worth a read?
http://books.google.com/books?id=irBeIsO8wZgC&pg=PA43&lpg=PA43&dq=statistics+++exit+poll&source=bl&ots=KGHuLuUBxc&sig=inz4Oasb10Lpuxgyyyppr99mscc&hl=en&ei=_0LjSu-_BI_L8QbAsaTrAQ&sa=X&oi=book_result&ct=result&resnum=7&ved=0CBgQ6AEwBjgK#v=onepage&q=&f=false

 

[QFIT]

Exit polls are scientific in nature, but can still be wrong as was embarassingly expierenced by CBS and NBC when they declared Gore the winner in Florida. One of the problems, besides the normal sampling error, is that some types of people refuse to take part in polls and some people lie, creating a bias. The bias is stronger these days as cell phones generally are not a part of polls, creating an additional bias.

 

[sagefr0g]

Exit polls are scientific in nature, but can still be wrong as was embarassingly expierenced by CBS and NBC when they declared Gore the winner in Florida. One of the problems, besides the normal sampling error, is that some types of people refuse to take part in polls and some people lie, creating a bias. The bias is stronger these days as cell phones generally are not a part of polls, creating an additional bias.

seems as if i have no end to questions, lol.

can't there be some credence to knowledge derived from the methodology of exit polls if such methodology was used on small (maybe random samples) of data such as the kind of data we use for blackjack stuff?

 

[QFIT]

seems as if i have no end to questions, lol.

can't there be some credence to knowledge derived from the methodology of exit polls if such methodology was used on small (maybe random samples) of data such as the kind of data we use for blackjack stuff?

Altogether different animal. In BJ, we are dealing with a very small edge. A typical poll has an error rate of +/-3%. When any poll is so close that it is anywhere near the small edge in BJ, the poll is called a statistical tie.

 

[Kasi]

In the case of multiple-hand play, yes the entire original amount is used as this is the only way that hand covariance can be taken into account. Risk is based upon the initial outlay including multiple hands, but additional funds (e.g. splits, doubles, insurance) are figured into wins and losses and therefore standard deviation...
Comparing results in CVBJ with CVCX/CVData is useless. Win/Lost results in CVBJ have no meaning, unless you play it for decades.

Thank you QFIT for answering both my questions.

I think that's what I thought lol.

 

[sagefr0g]

....

Simply put, it seems to me, and has for a long time, CVBJ and CVCX actually have different underlying assumptions as to the definition of a "round".

Probably because noone actually plays 200,000,000 "hds" or "rds" in CVBJ and actually tries to compare the result to a CVCX sim.

...l..

i think if you port the cvbj log to excel, you can then analyze the splits, ie. the number of splits, and the number from those splits, then adjust that according to the total number of hands cvbj logs, then compare those results with cvcx.

well like you was saying, make an adjustment, if you wanted to compare that stuff.