Question for SSR: CVCX 4.0
  • SSR,

    What do you think of the new CVCX? What is the difference between hands to double and N0? Why are these numbers nearly identical? Shouldn't they be exactly identical? Thanks.

    MJ
  • MJ1 said:
    SSR,

    What do you think of the new CVCX? What is the difference between hands to double and N0? Why are these numbers nearly identical? Shouldn't they be exactly identical? Thanks.

    MJ


    I personally like _all_ of Norm's software. My most recent upgrade was to CVBJ version 4.

    I believe the main change for cvcx/cvdata version 4 was adding support for multiple cores (the laptop I am using has two CPU cores so it will run the sims about 2x faster).

    To be honest, I have not studied the exact math to distinguish the difference between N0 and "hands to double". The main thing is that N0 is the number of hands you need to play to reach your "expectation" (more specifically, to be ahead by 1 standard deviation, or being ahead at least 2/3 of the times you play to N0 hands). For example, after playing N0 hands, you should be right on top of your expectation, assuming you play with a 1% edge, by the time you reach N0 hands you should be ahead .01 * N0 * average bet. And since you have played enough hands to greatly reduce standard deviation, one SD above or below where you end up will still have you in + territory...

    Doubling your bankroll is a different measure. That depends on your edge, and the number of hands you need to double your starting money given the aggressiveness you want to use (full-kelly, 1/4 kelly, etc). It is likely you are just looking at some "noise" in the computation, but perhaps Norm can chime in and help since he wrote the thing. But in a casino you are constrained to bet sane units (at least multiples of $.50, although almost everyone uses multiples of $5. That causes you to either overbet or underbet most hands, which could/would change the number of hands to double your bank, but would have nothing to do with N0 which is not about bankroll at all, just about enough hands to reduce standard deviation to an acceptable level.

    Note that N0 does not guarantee you you will earn your expectation. It is possible to play for a lifetime and end up a loser after playing 4*N0 hands. It just isn't very probable. But definitely possible.
  • stainless steel rat said:
    The main thing is that N0 is the number of hands you need to play to reach your "expectation" (more specifically, to be ahead by 1 standard deviation, or being ahead at least 2/3 of the times you play to N0 hands).


    Thanks SSR, but I thought playing N0 hands gives the player at least a 5/6 chance of being ahead, and a 1/6 chance of being behind. After all, if you play N0 hands you are playing a sufficient number of hands to overcome 1 standard deviation of negative variance. The area to the left of the bell curve after 1 SD is about 16%. Thus, shouldn't the player have around a 5/6 chance of being ahead?

    Just curious, why didn't you upgrade your CVCX to the newest version?

    MJ
  • MJ1 said:
    Thanks SSR, but I thought playing N0 hands gives the player at least a 5/6 chance of being ahead, and a 1/6 chance of being behind. After all, if you play N0 hands you are playing a sufficient number of hands to overcome 1 standard deviation of negative variance. The area to the left of the bell curve after 1 SD is about 16%. Thus, shouldn't the player have around a 5/6 chance of being ahead?

    Just curious, why didn't you upgrade your CVCX to the newest version?

    MJ


    I did upgrade - the cvcx upgrade was free. I just didn't pay a lot of attention to the details of what was new. Some features norm added are not things I use very much. My main goal has always been to run sims to find the optimal betting ramp for a specific game.

    You are correct about the 5/6. Was not thinking about the 16% way out to the right that could also happen.

    But as I mentioned, that's a completely different issue from doubling your bankroll because bets have to be in sane increments, and after simulating 2 billion rounds, there is a significant accumulated error in floating point math. I suspect that since the sims are based on kelly betting, that if the thing could bet precise amounts that were accurate to multiple decimal places, that the two numbers would be closer. But getting them exactly the same would still be difficult. All you have to do is sum up .1 a million times to see the problem. Computers can't represent .1 exactly, and it ends up being .09999997 which is pretty close, but not when you add it up a million times and expect to get 100,000, and not 99,999.97... Whether that is the problem or not is unknown for me however, not having access to Norm's source code to see what it is doing.
  • Where I am getting stuck is I do not even see why N0 and hands to double should even be anywhere close to each other. They are two totally different concepts. Correct me if I am wrong, but one has absolutely nothing to do with the other.

    In my first post, I suppose I took it for granted that they should equal without really thinking about what they represent.

    MJ
  • MJ1 said:
    Where I am getting stuck is I do not even see why N0 and hands to double should even be anywhere close to each other. They are two totally different concepts. Correct me if I am wrong, but one has absolutely nothing to do with the other.

    In my first post, I suppose I took it for granted that they should equal without really thinking about what they represent.

    MJ


    I don't expect them to be close to each other. I had assumed you had some set of rules where the math says they are the same (hands to double, hands to reach N0) but the sim was not agreeing.

    N0 is how many hands you need to play to end up ahead 5/6 of the time approximately. The lower this number the better the game and the lower your RoR.

    Hands to double is a different measure but is related to the above. Except now you are measuring how many hands are required, given some specific bank and betting policy, to double your bankroll.

    The two numbers are definitely inter-connected, but I would not expect them to be the same very often at all. I'll try to play with the new CVCX some to see what it does with this stuff...
  • SSR,

    Regarding hands to double, I did the math and it seems like this figure only represents the number of hands required such that the accumulated earnings equal HALF (50%) of the starting BR, not double.

    For example, if my starting BR is $10k, the HTD figure will represent the number of hands that must be played so that earnings equal $5k. The problem here is that this is not doubling the BR, it is merely increasing it by 50%.

    This is a simple calculation, please check it yourself and let me know what you come up with.

    Thanks,
    MJ
  • MJ1 said:
    SSR,

    Regarding hands to double, I did the math and it seems like this figure only represents the number of hands required such that the accumulated earnings equal HALF (50%) of the starting BR, not double.


    Good catch. Originally I was using the Kelly bank; not full bankroll. I assume you have Kelly set to 0.5, which is why you came up with 50%. The latest update uses the full bankroll in the HTD calculation.
  • QFIT said:
    Good catch. Originally I was using the Kelly bank; not full bankroll. I assume you have Kelly set to 0.5, which is why you came up with 50%. The latest update uses the full bankroll in the HTD calculation.


    Hi Norm,

    Thanks for stopping by. Always nice to hear it straight from the horse's mouth.

    Post edited by user.

    MJ
  • As I said "The latest update uses the full bankroll in the HTD calculation." There is a free update at www.qfit.com/downloads.htm.

    norm
  • Ok Norm, I downloaded the update and am pleased to report that the HTD figure is now performing as one would expect for different levels of Kelly. :) Thanks for the update!!

    Further, I am also glad to say that N0 is now working correctly when DA is selected. Before, whenever DA was selected, N0 would actually increase!! This update solves several problems.

    On a side note, you might want to revise the update description, because it makes no mention of HTD or N0/DA but mentions something about the KISS count.

    There was something else I would like to ask. Can CVCX be used to determine win rates for reverse wonging? For example, let us say the Hi-Lo counter plays off the top of the shoe and exits at TC <= -1 and looks for another table. If we select back counting (play 100% hands) and the DA, shouldn't that provide the win rate for reverse wonging? <br />
    Thanks,
    MJ
  • That might work for reverse wonging. But, it is only an estimate. Accurate numbers are not possible post-sim. CVData would be required for accurate numbers.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!