Index Ranking with CVDATA

[RustyNCrusty]

Using CVDATA what’s the best method to calculate index contributions for index ranking purposes?

For a given index set will the sum of the individual index gain contributions equal the gain of the total index set over a baseline of basic strategy while accounting for SE? Assume the same betting strategy is employed in all sims. Two methods tried: a) subtraction of each index from the total set to determine contribution and, b) adding the index to a Basic Strategy baseline to find contribution. I expected method (a) would be the most accurate but when reconciling the sum of contributions method (b) had the least error and the rankings were not the same between the two methods.

[Norm]

You would need to run 20 Billion hands for each index. And, some indices affect other indices. It's incredibly time consuming.

[RustyNCrusty]

Yes, 20 billion hands per sim, ranking 63 indexes took 2.5 days at roughly an hour per....then repeat to test the second method. Yes, 20B were required to get the SE down to about .001 in order to have decent resolution for the ranking and even then..... As you note one can see index interactions. Norm, thanks for your comments....provides confirmation that the time consuming nature of this is correct as experienced.

[ZenMaster_Flash]

If I understand the O.P.'s question.

The Ranking of the importance of indices has long been 'solved'

See the 6th ed. of Theory of of Blackjack

[RustyNCrusty]

ZF – Yes, you are correct it was done in BJA3…in Chapter 5 I18 + Fab 4 for HiLo 4D S17 DOA nRSA? LS 75% pen and referenced again in chapter 10 where 6 & 8 deck are called out. The method of computing the rankings in BJA was not a direct sim and was for a narrow set of rules. Even DS comments that the “above study is not a definitive work”. Is there something new in BJA6? So, to satisfy my own curiosity I want to see how direct sims compare with BJA under various rule sets.

[Three]

ZF – Yes, you are correct it was done in BJA3…in Chapter 5 I18 + Fab 4 for HiLo 4D S17 DOA nRSA? LS 75% pen and referenced again in chapter 10 where 6 & 8 deck are called out. The method of computing the rankings in BJA was not a direct sim and was for a narrow set of rules. Even DS comments that the “above study is not a definitive work”. Is there something new in BJA6? So, to satisfy my own curiosity I want to see how direct sims compare with BJA under various rule sets.

The advantages per index varies by count as well since they each have their own correlation to matchup EOR values although I am not sure if it is enough to change the order of the indices importance. The frequency of the matchup wouldn't change although the frequency you use the index play could have small differences. The main fluctuations among counts would be from the correlation to the EOR's for the match up resulting in stronger and weaker indices for different counts. For those that like graphical representations the tighter bell curve for correlation of the index to combinatorial analysis would be a stronger index. Weaker counts have the shoulders of these bell curves twice as wide as strongly correlated counts. This translates into more severe penalty for use of the index when it is not advantageous to do so. About 3/4 of the illustrious 18 are hurt by ace reckoned counts as an example. This may affect the order but I am not sure that the correlation weakness is enough to change the influence of frequency of the matchup or the index use as weaker correlation often makes the index used less frequently.

[DSchles]

Several years ago, for the 25th anniversary of the I18 (September 2011), "Zenfighter" undertook a "new look" at the I18 and Fab 4 and sent me all of his findings. I wish I had had the courage to undertake the writing of a new Appendix to BJA3 (which might have made it BJA4!), and that would have carried the findings ("The I18 Revisited"), but I didn't.

In any event, there were very few differences, although there were a couple that tweaked the order slightly.

You can knock yourself out doing this kind of study, and I commend you for the effort, but you aren't going to unearth any major findings. Promise!

Don

[RustyNCrusty]

DS - I agree that I'm unlikely to find any major findings...it's like the adage of the two economists and the $20 bill on the ground...

For my own edification I will plow for a while.

What percentage of ideal gain using all indexes do you estimate Catch 22 to provide? Would you expect this estimate to be applicable across all rule sets?

[Three]

In any event, there were very few differences, although there were a couple that tweaked the order slightly.

You can knock yourself out doing this kind of study, and I commend you for the effort, but you aren't going to unearth any major findings. Promise!

Don

If you learn all your indices and don't get lazy about memorization it is really a moot point isn't it.

[Three]

What percentage of ideal gain using all indexes do you estimate Catch 22 to provide? Would you expect this estimate to be applicable across all rule sets?

Just learn all your indices not just 22. The most important will be used a lot. Look at the hand matchup EOR's for correlation to your count. It is usually worth learning those indices used with your biggest bets out regardless of how common they are used but if the correlation of the count to the EORs is poor you may be better off playing basic or at least there isn't much gain that isn't more likely based on the luck of how whacky those, not so neutral, neutral cards are. If the correlation is poor enough 1 extra or deficit neutral key card can have more influence than a huge TC excess or deficit over/under the index.

[RustyNCrusty]

Tthree – Sounds like good advice and very practical…particularly on correlation. I’ll take my vitamins – Thks.

[DSchles]

"What percentage of ideal gain using all indexes do you estimate Catch 22 to provide?"

Has been discussed often. About 80-85% for the shoe game. Less for SD.

"Would you expect this estimate to be applicable across all rule sets?"

No, because, for the doubling indices, it matters if DAS is permitted or not.

Don

[ZenMaster_Flash]

Would someone please chime in here.

moses imagines that if he hits a 12 vs. 5 or 6

or hits a 13 vs. a Deuce, etc. etc.

that the pit will identify him as a Card Counter.

I have tried to explain to him that he is mistaken.

He also holds that if he requests a $5 table be upgraded to

a $10 table that he will be exposing himself to extra scrutiny.

Would someone please chime in here ?