Strategy advantage
  • sage01 said:
    SSR:
    If you have the time, I would like to take you up on the offer to run some sims with CVCX. Let’s try about 10 + runs of anywhere from 10K to 55K hands. I assume they all run as one long session. Also does it plays the hands & not just assumes a % gain? What is the logic of the software? How does it go about the problem? I would be interested in the variance of the final results
    Let’s try, 6 decks, I play $10 and max bet is $50 (1 to 5 spread) if that is too small I could change tables and move up to a 1 to 6 spread. It is a good game, about 85% pen, (less than 1 deck behind the cut), double any two, split to 4 including aces, double after split and late surrender. Stand 17, BJ pays 3 to 2. Bjmath.com puts it at EV= -.402225%
    Looking forward to the results


    I am going to try to do this today. Here's what it does however. It uses a pseudo-random number generator to generate a stream of cards. The idea is to use random numbers to populate the "shoe" (you specify the number of decks with the CV software) as though it were randomly shuffled (and you can even modify this to use the mechanics of a casino house shuffle if you want). Then it just deals these cards to the penetration you specify, then re-shuffles and goes on. So it really does "play" the hands one at a time and records the results, rather than just using a percentage calculation. 20 billion rounds takes a few minutes on m core-2 laptop...

    results will follow in a bit.
  • stainless steel rat said:
    I am going to try to do this today. Here's what it does however. It uses a pseudo-random number generator to generate a stream of cards. The idea is to use random numbers to populate the "shoe" (you specify the number of decks with the CV software) as though it were randomly shuffled (and you can even modify this to use the mechanics of a casino house shuffle if you want). Then it just deals these cards to the penetration you specify, then re-shuffles and goes on. So it really does "play" the hands one at a time and records the results, rather than just using a percentage calculation. 20 billion rounds takes a few minutes on m core-2 laptop...

    results will follow in a bit.


    OK, first glitch. Norm's software has a 50,000,000 round _minimum_ (I suspect you can figure out why that is based on my earlier comments). But for fun, I decided to run several 50M round sims to see how much the results vary. I will only post the hourly win rate since that is the "interesting number". Rules were as you gave, except you didn't mention strategy so I used hi-lo with the I-18 BS departures plus the fab-4 surrender indices...

    run 1: $5.32 per hour
    run 2: $5.29 per hour

    There was enough agreement between the two 50M round sims that running more seems pointless.

    if you use a more reasonable spread of 1-12, this goes up to $12.80 per hour, or if you use 1-20 which is what I use on shoes, you get to $23.00 per hour. Shoes are tough, because you don't get many big bets out... That makes a big spread essential for those rare + counts/ big bets...

    One thing CVCX/CVData provides is the percentage of hands played at each true count value...

    70% of the hands are played at a TC <= 0... 11% at a TC of 1, which is a nearly even game, leaving 19% for the higher counts where you have an advantage, with the most being played at +2, and +3...<br />
    Hope that helps. At 50K games multiple runs would vary more. at 10K they would vary more. Significantly more. 50M runs take a few seconds. 50 billion runs take minutes and provide _very_ stable results for comparisons between various strategies.

    But note the spread hourly win rate. That's not guesswork, and it is critical to understand that when actually playing. With CVCX/CVData, you _know_ the expectation you should have for a given bet spread. It also gives you RoR values for your bankroll...

    For example, if you start with $3000, and play long enough using your 1-5 spread, you will go bankrupt one of every eight times you go out to play. Using 1-8 is the optimal spread for that bankroll and game, giving you a 12.7% RoR compared to a 13.2% RoR at 1-5 (not a lot of difference, but a difference). 1-20 will bankrupt you about once every four sessions. More money solves that... taking $5,000 with you and spreading 1-20 has a 1 in 10 chance of going bankrupt.

    Notice how much information I gave? It took very little time, just a click here and look at the results. Click again and the results are updated... So you can play "what if?" all day to see exactly what happens when you change something.

    For example, what if you memorize _all_ of Wong's hi-lo BS indices???

    Using my last result, 1-20 bet spread, produced an hourly win rate of exactly $22.12... using all indices takes this to $23.69. Think memorizing all 100+ playing indices is worth the effort?

    This is the way to get your questions answered accurately, quickly, and without a lot of noise and arguments to boot. :)

    I suppose the reason for the 50M round minimum is two-fold... (1) using smaller numbers produces less accuracy; (2) running 50M rounds takes seconds (and I do mean seconds, on my 2.0ghz core-2 duo laptop, it takes 3-4 seconds max). Smaller numbers would be interesting just to see the variance between two different 10K round sessions, but it isn't possible...
  • SSR- Not to beat a dead horse, but:

    If the task were an unrestricted teacher to class problem, I think we all know who turned-in their paper first and won the contest hands down. I see that as an application of common sense that does not preclude the use of technology in any way when necessary. Nor does it justify the folly that I have submitted as a solution. If anyone on this small planet knows the value of common sense coupled with science it is me (modesty out the window).

    If Grifter's offer were to anyone and not in a BJ sense, I would have a valid claim. However, as is, I would not lay claim to the $1000.00 because I have not defined a real and usable BJ progression. In this context, I disagree will all efforts, mine included.

    Once again, not taking sides, but making a half-A** attempt to make my observations clear.
  • Ray said:
    SSR- Not to beat a dead horse, but:

    If the task were an unrestricted teacher to class problem, I think we all know who turned-in their paper first and won the contest hands down. I see that as an application of common sense that does not preclude the use of technology in any way when necessary. Nor does it justify the folly that I have submitted as a solution. If anyone on this small planet knows the value of common sense coupled with science it is me (modesty out the window).

    If Grifter's offer were to anyone and not in a BJ sense, I would have a valid claim. However, as is, I would not lay claim to the $1000.00 because I have not defined a real and usable BJ progression. In this context, I disagree will all efforts, mine included.

    Once again, not taking sides, but making a half-A** attempt to make my observations clear.



    Before we go on, here is my _original_ quote that started this ridiculous discussion:

    SSR said:


    I don't know enough about it to question the results. My point is that 55,000 rounds seems like a lot, but the variance is large enough that one could end up ahead or behind. And the real problem with using this approach is that the 55K hand outcomes are fixed. It would be quite easy to define a progression to beat that string of results, even though it would not work in real games...



    Notice the _clear_ statement of "even though it would not work in real games..." That was, and has continued to be overlooked. And it is not ambiguous or unclear in its meaning IMHO. I _clearly_ said that it would be easy to define a progression to beat that string of results. That is true. Grifter said "impossible and I should apologize to Sage or else prove him wrong with $1,000 at risk to the winner." I gave a trivial progression plan, since I know the results of each hand before it is played. You did _exactly_ the same thing. I said it would not work in a real game. You said exactly the same thing. As did Sonny.

    So exactly _who_ is "wrong" in this fiasco??? Grifter just half-assed read what I wrote and went ballistic, as usual, when I'd bet now he wished he had read more carefully to have avoided this entire waste of time. No a progression to beat a specific of hands is not hard to develop, at least for someone that uses computers to solve similar problems every day (very similar to gene-splicing in fact, or searching large chess trees). No it will not be useful for any set of hands except the specific case it was designed for. Yes it might require 55,000 different bets. Or it might just require one on the first hand followed by walking away. Or a real progression (which I was certainly thinking of when I wrote the response) that would still be based on specific patterns in the specific 55,000 rounds mentioned. But there is no certainty that a short progression will fit, so I left the door open for the ultimate escape that it might require a progression as long as the series of hands.

    And things went several light-years south of South beyond that point... again I think more in the spirit of arguing as opposed to information exchange...
  • Wow, go out of town for a couple days to play poker, and come back to find out I'm the blame for everything....lol.
  • .....As far as Grifter's "last warning" goes. He can stick _THAT_ where the sun don't shine.....

    The remainder of this post was deleted......Totally redundant.......Poster was also warned last year about his posts.

    Moderator

Howdy, Stranger!

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