Hi, I hope my long list of questions below in using CVindex is not out of line! Thanks.
(btw, if you read through it you might find that I'm complicating things a lot...... I don't know, maybe I'm having the wrong idea here.)
------- 1. General setup ------------
Index generation requires settin up the TC at which the max bet is called for.
However, given a fixed set of rules, different playing styles have quite different optimal ramp (determined using cvcx).
For example, play-all or Wonging, play one spot or play two spots, number of players at the table, etc.
Some optimal betting schedule will have max bet out as early as TC = 3, but some as late as TC = 6.
Is it advisable to run sims for each situation?
A similar consideration goes for DIFFERENT set of rules which make things worse...... there are so many different rules, S17 or H17, RSA or not.... should I really go ahead and run CVindex for all the rules/playing style combination (that I will engage in)?
------- 2. parameter-feedback -------
After a set of r-a indices is generated, one runs a standard sim to use cvcx on the new strategy (base + r-a modification) to determine a new optimal betting spread.
If the new spread has the max out at a new TC, different from the setup in generating the r-a indices, what should I do?
It feels like one needs to run this entire process to let things converge.....thinking of the many combinations mentioned above in question 1., I don't know how long it would take @@
------- 3. ratio (kelly factor) -------
It seems that the option 'ratio' is the inverse of Kelly factor, right?
(ratio = 1 means full Kelly, ratio = 2 means half Kelly, ratio = 3 means 1/3 Kelly etc)
However, for a given fixed set of rules, for different playing style (play-all or Wonging) one doesn't necessarily play at the same Kelly F.
That is, similar to the problems above, there are several 'ratio' I will play at. Shall I generate r-a indices for each 'ratio'?
------- 4. Generating 30+ all at once ---------
Is it okay to generate more than 20 or 30 indices all at once (in one sim)? Is it better to maybe generate 4 or 5 indices first, plug them back into the 'base' strategy then generate more? (each time I will have pre-sim and it will be multi-pass and beat to death)
Obviously the 'base' strategy that is used affects the indices being generated. I would imagine the 1st index generated will actually affect the outcome of the 2nd one, IF the 1st one is included into the 'base' strategy (is it?). So on and so forth.
I read somewhere in the threads of the former blackjackinfo board that only the relevant hands are simulated for that specific index, while still satisfying some frequency consistency etc. I don't know what that means exactly and how CVindex runs exactly (I guess I don't really need to). I would just like to know that I can generate as many risk averse indices as I like, even the whole table of every entry of the hit/stand/double/split/surrender.
p.s.
I know in general the use of r-a indices should modify only a few of the higher index values for when the big bet is out, but as described above with max bet out at TC = 3 or 4, many of my new indices are different from the 'base' strategy.
------- 5. auto-fill ---------
When running a CVindex sim, only some indices are checked to be generated, but the resultant strategy table is full.....that is, other indices are sort of auto-filled.
Where do these number come from? I suppose they are NOT usable, right? Apparently the CVindex sim report has the detailed EV tables for only those selected.
For example, I checked splitting 3 vs. 2 to be generated, and I get the full list of splitting 3 vs. 2~7, which looks really strange:
33: 0, 3, 0, -2, -9
[vs 2, 3, 4, 5, 6]
It was multi-pass, beat to death with pre-sim. (hopefully the 'multiple draw after Ace split' wasn't checked....honestly I'm not sure as there was this bug that automatically checks it from time to time)
------- 6. Surrender when no surrender ---------
In the 'Rules' tab, I didn't check the desired late surrender, but I selected some surrender indices to be generated later on.
Are the resulting surrender indices legit? (the base strategy has its own surrender table, which is of course different from the generated one)
-------- 7. DAS --------------
okay, this one has not much to do with r-a indices or CVindex, but just the format of strategy table itself.
The indices for DAS is incorporated as the 'modification table', yet there is only a single pair of modification tables (for SD/DD and shoe games, on the 5th and 6th column) for both S17 (1st and 2nd column) and H17 (3rd and 4th).
I'm using Hi-Lo and the DAS indices for S17 and H17 are surely very different and requires a total of 4 modification tables.
I also notice this same problem while using cvbj. Part of the playing strategy that should change still remains the same when I check/uncheck H17. (just to be clear, this is with the existing '94 basic hi-lo under the name Stanford Wong, of course.)
Is this by design or am I missing something? It seems to me that the only way to get around this is to define two separate 'single ' strategies (one for S17 and one for H17), instead of a 'advanced' strategy which fails to include distinct DAS index tables.
Reply With Quote
Originally Posted by Norm
Bookmarks