Index variation inconsistencies?

Hi all,

I am a little confused why there are different Index variations for the Illustrious 18 to those given by Stanford Wong in Blackjack secrets to those given by Gamemaster on the lessons on this site - when they are all referring to TC.

Does anyone care to elaborate, please?

Thanks in advance and regards

Laurence.

Although they all refer to TC, the different TC's are not weighted the same and indices will change depending on your spread. The I18 for someone with a 1:4 spread Wonging will be slightly different than the I18 for someone with a 1:20 spread play all.

For example, if you're Wonging out at +1, none of the indices below +1 have any value to you. As your spread grows, so do the relative value of the indices which are very high (TT vs. 5, for example).

The only way to get a perfect list is to input your own spread and do all your calculations yourself. If you can't do that, just use I18 as a general list of suggestions rather than a concrete list of rules.

The bet spread does not affect the indices. The index variations may be due to how they are derived: floored, rounded, or truncated.
Indices will also vary if generated for max ev or risk averse.

BJC

I thought that Wong's earlier works were written without the powerful simming capabilities we have today, and that some of the indices are slightly off - not so ?

If I recall correctly the indices were modified in later editions to reflect flooring instead of rounding, but I had read it a long time ago, so I'm open to hear from the others.

BJC

Whoops, sorry, I misread the OP. I thought he was asking why I18 lists are different.

I agree with BJC. Has to do with method used for coming up with the indices. But, the difference in results would be very minimal, unless you have some strange version of Ill18 that are way off.

i'm remembering a bit about the I18 from when i read about them in D. Schlesinger's Blackjack Attack.

like the I18 that he presented were i believe for a four deck shoe, s17 game 75%pen and i believe a spread of 1-12 using hi/lo count .
as Schlesinger put it this was meant to be a sort of a compromise for multiple deck games, 2 decks thru 8 decks.

so as i see it, absent grinding out indices for a whole slew of game, the I18 is a kind of a fudge factor thing, sort of in the same sense that some generic basic strategy might be that is used for virtually all games.

it's a fuzzy sort of approach, lol.

just a point of clarification, i believe maybe not spread, but the RAMP (or at least point of max bet) does affect RA indices, but as BJC stated, not EV-max indices.

Rukus

Your excellent with the math behind the the numbers, but how would the ramp effect the indices (max ev or RA) when the indices are generated without a betting strategy in place?

Aren't the indices generated using formulas not inclusive of the betting schedule?

Thanks

BJC

Another question...

Not to add another question to the mix, but are the values for indices (any) different for different systems? For example, say Zen Vs Hi/Lo? Or would it just be that the method of getting to whatever count (tags) would be different?

I only use one system so I never thought about it before?


Homeschool


P.S.....sorry not trying to hijack your thread

Homeschool:

Sorry to say you are wrong! It is like this:

"Guns don't kill people. People from Texas kill people."

BJC, good to hear from you.

If i properly understand the process (that at least QFIT uses) to calculate RA indices, we need estimates of EV and Variance at each TC, at least relative to every other TC. These, particularly variance, are affected by the ramp and are derived in the "pre-sim" that runs before the index generation begins. To further refine the indices, once they are generated, a new "pre-sim" can be run with these new indices to get even more accurate EV and Variance estimates before re-generating the indices yet a second time. These is no formula used for RA index generation in CVData, it is done purely by simulation (in fact I believe QFIT's CVData uses simulation for the EVMax versions as well). Now that I think about it, i guess i always just assumed the RA process used my betting strategy I had pre-selected on that relevant tab...

QFIT would be best to comment on this, but in the meantime i believe this is how his CVData works. At least i hope i had the proper understanding all these years

The count at which you make your max bet can affect RA indexes which are above that count. But the effect is minor and this is really esoteric.

The betting ramp, other than max bet, is not relevant. I allow you to specify a betting strat when generating indexes in case you check the option that automatically runs a sim after the index generation.

jump in here, lol.......

not so much maybe the indice values but the value of the indices relative to should they be in the I18, sort of thing, might be the point.
that maybe could be betting related, sort of decisions. ie. to include the indice or not and the power or value of it, sort of thing.

i dunno anything about the zen count.
thing is the I18 as presented by D Schelsinger was geared for hi/lo true count sort of counts.
so yes depending on the count your using you'd either need to convert the indice values of the I18 to fit your count or generate the actual indices with your count system in mind.

BJC and KewlJason are correct.
Additionally,
You may not be comparing the index numbers for the exact same game. Haven't read Wong's Secrets, but gamemaster's index numbers are only for 6d, S17, DAS and are Risk Averse. Unless Wong specified these are risk averse numbers and specifically for the aforementioned game that could also account for the difference.

If the numbers are only off by 1, you may want to read Snyder's hi-lo lite intro and accompanying articles to put your mind at rest. (Unfortunately, counting is a little closer to chemistry than physics.)

Hope this is helpful!

the wise frog is correct. for zen, the conversion rate is about 1.7 zen TC:1 HiLo TC, so if a HiLo index has a value of 10, the zen count index is likely around 17. this is just a rule of thumb but not 100% accurate. you should use published indices or generate your own for your own count.