Oh. Thought you were giving the advantage of the particular insurance play, which didn't make much sense. Sorry.
Don
Why not work on this set of commonly used rules as below? They are 6-deck, S17, DAS, SPL-3, SPLA-1, no HSA, and surrender, which still exist at high limit tables and are referred by many blackjack authors as a standard. Also, refine the TC range into (-4, -3, -2, 0, +1, +2,+3, +4, +5, +6, +7) to concentrate on higher bets where you may spread. If you try to reproduce published data, chances are you will not because many definitions are ambiguous.
Last edited by aceside; 12-25-2023 at 08:29 AM.
Merry ????? and many happy ????????'s to all,
The scripts are a work in progress and I always try to reality check this kind of thing by seeing if I can duplicate known values. Once it is all working, it can be used for anything; my goal is UBZII.
The TC values were the data points I used; I spread them out because it takes an hour or so for each one and I was just trying to get a useful range. Still working on being able to generate decks of a specific count (this is what goes incredibly slowly at high counts by just walking through the deck); I think I found a way to speed it up by generating specifications for a bunch of decks (100,000) ahead of time. Then shuffling the cards still in the shoe, several times for several more decks; then shuffling the cards and restoring the ratios in and out of the shoe and repeating the process. Creating a deck by just exchanging arbitrary numbers of positive and negative cards would be the simplest way but it doesn't work for unbalanced counts. A formula for card ratios would be nice, but complex for multilevel counts. (In case anyone is interested in all this . . . .)
[The question marks were supposed to be emojis.]
Last edited by BackCounter; 12-25-2023 at 11:07 AM. Reason: emojis got stripped
I’d say we just stick to HiLo. I just looked up UBZ II and found what it stands for. It looks like that this system was discussed about 12 years ago and then disappeared. It is probably not effective for card counting shoes because the most important cards, aces, are significantly undervalued in this system.
"Compromise strategy" makes it sound like it suffers on one end or the other. In Blackjack Attack, it beats KO every time, 1, 2, or 6 deck, and usually beats Hi Lo. It sometimes beats even balanced Zen. The compromise, I suppose, would apply to fully ace-reckoned level 2 counts.
Some strategies aim at shoes and some at single deck (unless you want an ace side count). George used to alternate between hand held and shoes every six months. He didn't like switching strategies; so he developed UBZ as something that worked well for both.
"I don't think outside the box; I think of what I can do with the box." - Henri Matisse
Hello; Don, you have a lot of footsteps to follow in! The progressive levels of confidence that you used was a big help. I run ten threads for each player hand-upcard combination and use the mean of the results. I used each thread as a measurement and generated 99% confidence intervals for each action, then if the top two actions overlapped I redid it with more iterations. Starting at 100,000, then 1,000,000, then 10,000,000 if necessary. It takes less than an hour for one data point and typically all but a few make 99%.
The data points used integer true counts only. Generating decks ahead of time with the desired count saves a lot of time; it just takes a few minutes until the count gets to about 10, then there is an exponential-like increase, close to two hours for 14. I use the pattern of card counts, how many 0, +1, and -1 cards are before and after the deck position, then populate and shuffle them internally based on the pattern to get the decks, then shuffle the cards after the deck position to get more. 100,000 iterations looking for a count of 14 generates only about 25 deck patterns. I hope this does not introduce any sort of bias.
Then running a sim to get the hand frequencies, advantage, insurance, variance, and covariance takes another 45 minutes or so; it takes 1,000,000,000 hands to get consistent data. 11 data points took a total of about 18 hours of running time. Not as efficient as Norm, but hey, I did it. (Using floating-point math and lots of multi-dimensional arrays.)
The problem: the basic strategy comes out correct, but I do not know if the graphs for all the counts come out right. For 8,8 vs. T, the lines for split and stand have a reverse slope from the ones in the back of the Hi-Lo book. And the indices do not all match. And the variance/s.d. seems to increase faster than it should at high counts. That's what makes me think there is a bias somewhere.
Four decks, and the rules specified for the illustrious 18 in BJA3.
bs2.jpg
graphs2.jpg
indices2.png
ramp2.png
Sorry to just be getting to this. Busy last few days. I can't comment on your programming methodologies; I'll leave that to Dave. Most of the indices look pretty close, but doubling 10 vs. T and standing on 15 vs. T are surely too low. Can't say why. You should never be writing "Hit" anywhere. Indices are for standing, doubling, splitting, insuring, and surrendering when the TC is equal to or above the index. NEVER use "hit" in such a chart.
I suppose it does no harm to furnish the covariance, but you don't use it for anything, so you?
Don
Hello, and thank you very much for the reply. I was looking at the negative indices in reverse, and printing the "backward" actions; I guess that doesn't make any sense. But two (3,3 and 6,6 vs 7) are actually hit (green):
337.png
I use the covariance to calculate Kelly for two hands.
Bookmarks