Generating RA indices using CVData?

rukus

Well-Known Member
#1
couldnt find a previous post covering this, so apologies if it's a repeat (and please provide a link if the question has already been answered!).
i finally went out and picked up CVData from Norm W. Was leaning towards SBA but decided Norm will be more likely to continue updating his suite of tools in the future.

anyway, i cannot find a simple generic set of step-by-step instructions for generating RA indices for any count. after playing around with the program/help file, i have some ideas of how to proceed but don't want to waste time generating indices only to discover i missed a step along the way (ie garbage in = garbage out..). so does anyone have some solid instructions that anyone can use to generate RA indices (or even E[] maximizing indices should they desire) using CVData for a given counting system?

thanks in advance,
rukus
 

nightspirit

Well-Known Member
#2
rukus, take a look in the user guide pg. 20. Chapter Index Generation. First you must define a new strategy. Thats also explained at pg. 28. Normally EV-Maximizing indices are simmed automatically, for Risk Averse indices check the box at the Indexes (+-8) screen.
 

rukus

Well-Known Member
#3
i had taken a look at that tour in the manual.

ive tried this for say AOII and am not reproducing those found in BjFb (which i would like to do before generating my own). i assume this must be because i am choosing different #decks and/or penetration levels than carlson did. would that be the reason?

another question - if i define a new playing strategy in CVData as basic strategy + the AOII tag values and generate E-maximizing indices, i get slightly different values than those generated when i choose the pre-configured "Advanced omega II" playing strategy included with the software. any ideas why this would happen? for complete info on what i am doing, ive left out ace-sidecount definition and am just using flat-betting strategy during index generation, as the manual mentioned that all options related to bet strategy (which spread and ace-sidecounting are) are ignored for index generation purposes.

thoughts?

thanks,
rukus
 

nightspirit

Well-Known Member
#4
rukus said:
i had taken a look at that tour in the manual.

ive tried this for say AOII and am not reproducing those found in BjFb (which i would like to do before generating my own). i assume this must be because i am choosing different #decks and/or penetration levels than carlson did. would that be the reason?
Yes, this can be a reason. Some develop multideck indices using 4 decks, as compromise between 2 and 6 decks and use them for both. I don't know how sensible balanced counts are for the penetration level, but unbalanced counts are very sensitive to that. You should design the indices for the conditions you play most. Then you could rund a sim to compare them with the original ones.

rukus said:
another question - if i define a new playing strategy in CVData as basic strategy + the AOII tag values and generate E-maximizing indices, i get slightly different values than those generated when i choose the pre-configured "Advanced omega II" playing strategy included with the software. any ideas why this would happen? for complete info on what i am doing, ive left out ace-sidecount definition and am just using flat-betting strategy during index generation, as the manual mentioned that all options related to bet strategy (which spread and ace-sidecounting are) are ignored for index generation purposes.

thoughts?

thanks,
rukus
I also used the flat-betting strategy. I never dealt with side count strategys and couldn't find detailed info in the manual now. When you select a playing strategy can it be that you must select a SC-strategy too? Or is the SC strategy only for betting purposes? Maybe you that set up a betting strategy.

Yes i got it! When you define a betting strategy there is a colum called "Betting Side Counts". There you can add the tag value for the card you like to count. See also Reference Manual pg.15.

Hope that helps! :)

Hmm, i looked into CVData and i'm a little confused know. I think you are right. The SC can't be implemented for the index generation. It's seperat, only for betting.
 
Last edited:

jack.jackson

Well-Known Member
#5
rukus said:
i had taken a look at that tour in the manual.

ive tried this for say AOII and am not reproducing those found in BjFb (which i would like to do before generating my own). i assume this must be because i am choosing different #decks and/or penetration levels than carlson did. would that be the reason?

another question - if i define a new playing strategy in CVData as basic strategy + the AOII tag values and generate E-maximizing indices, i get slightly different values than those generated when i choose the pre-configured "Advanced omega II" playing strategy included with the software. any ideas why this would happen? for complete info on what i am doing, ive left out ace-sidecount definition and am just using flat-betting strategy during index generation, as the manual mentioned that all options related to bet strategy (which spread and ace-sidecounting are) are ignored for index generation purposes.

thoughts?

thanks,
rukus
Hey rukus,
If you discover the secret betting strategy, please send it over to me.Ty.
Im pretty sure carlsons indexes were derived from three decks? And ya, the ace is left neutral.
 

rukus

Well-Known Member
#6
will send it over as soon as i have everything run. i left this running over night and its still generating... in hindsight i should not have told it to find indices for splitting aces or 8s as those are taking hours for each dealer upcard (which i dont understand since carslon's book has #s for Ace vs * so apparently aces are not splittable all the time. 8s on the other hand was just stupid move on my part). anyone know if there is something wrong that it is taking that long on my machine for decisions that make no sense, ie splitting 8s vs 5 or 6? i know it should take "a while", but hours per upcard decision??
 

nightspirit

Well-Known Member
#7
And I thought it's because of my slowly computer...

welcome to the club :) It takes so long because for RA indices it will be simmed a few million additional rounds. Last week I simmed complete indices for the 2D UBZ, using the Greater Accuracy function. It took more than 12 hours. I realized that for the more uncommon indices (those except catch 22) it took notably longer, because they were either at very high or at very low counts. The indices against 5 and 6 took not that long. Curious, but maybe because of the RA function.
 

rukus

Well-Known Member
#8
thanks for your input. been running for well over 12 hours now, probably closer to 15-18 and just about the whole splitting table left. im not even doing complete indices. i think its the splitting table that is taking forever, ie 7 out of those 15 hours or so for the aces vs *... didnt think it should take THAT long! ill give it another 12 hours and see whats up. thanks again.
 

nightspirit

Well-Known Member
#9
Your welcome. Maybe you got a few results at the end of the day, or tomorrow, or the day after tomorrow or...;)

You can also abort it and proceed another time, the results won't get lost.
 

rukus

Well-Known Member
#10
just about finishing the index run up (should be done in a few hours) after several days of constant running (i did the very accurate mode with multi-pass option).

however, one very weird thing im seeing (and which is making me think i did something wrong) is that CVdata spit out NEVER split 8's vs a dealer 10. can someone explain to me how that makes any sense at all? if there was ever a dealer upcard to split 8s, wouldnt it be a 10 (unless the count was very high obviously)??? not sure what i did wrong but something doesnt make sense here. all other indices seem to make sense from what i can tell. but can anyone make sense of what i said above about it showing to never split 8,8 v 10?

thanks,
rukus
 

nightspirit

Well-Known Member
#11
rukus said:
just about finishing the index run up (should be done in a few hours) after several days of constant running (i did the very accurate mode with multi-pass option).

however, one very weird thing im seeing (and which is making me think i did something wrong) is that CVdata spit out NEVER split 8's vs a dealer 10. can someone explain to me how that makes any sense at all? if there was ever a dealer upcard to split 8s, wouldnt it be a 10 (unless the count was very high obviously)??? not sure what i did wrong but something doesnt make sense here. all other indices seem to make sense from what i can tell. but can anyone make sense of what i said above about it showing to never split 8,8 v 10?

thanks,
rukus
Normally you would receive an index for splitting 8's. Take a look at the surrender chart, if you set the option to sim an index for surrender 8's, it could be that CVData provides an index for this play, if the rules for the index generation are LS. For Hi-lo you would stand on a pair of 8's v. 10 at +5 and surrender at +1.
 

jack.jackson

Well-Known Member
#12
Make sure you dont lose additional wager due to dealers natural. Note: There is no index for this hand on this rule.I think this is european no peek!
 

jack.jackson

Well-Known Member
#14
rukus said:
im using AC rules, with checking under both the ace and 10, no surrender.
Possibly, it has something to do with the aces being counted as neutral. Since the aces play a big part on a pair of 8s. For example, 8+11=19)8+2+11=21)beating a dealers 20.

You could run a few tests to prove this theory.

1. Run a sim on another ace neutral count, to see if the same thing happens.

2. Run a sim on a ace reckoned count, to see if it gives you an index number.

Good luck and keep us posted!

______________________________
Peaceful waters: John Prine
 

jack.jackson

Well-Known Member
#15
rukus said:
just about finishing the index run up (should be done in a few hours) after several days of constant running (i did the very accurate mode with multi-pass option).

however, one very weird thing im seeing (and which is making me think i did something wrong) is that CVdata spit out NEVER split 8's vs a dealer 10. can someone explain to me how that makes any sense at all? if there was ever a dealer upcard to split 8s, wouldnt it be a 10 (unless the count was very high obviously)??? not sure what i did wrong but something doesnt make sense here. all other indices seem to make sense from what i can tell. but can anyone make sense of what i said above about it showing to never split 8,8 v 10?

thanks,
rukus
Did you ever figure why this was happening, because I'm having the same problem?

It doesnt make any sense, because the count has to be ptetty high, not to split. It would at least make a little more sense if it spit-out ALWAYS split, but it keeps saying NEVER split.

Its like ENHC is enabled, but Im getting the correct indices for 11,10 vs Ten, so I dont think this is the problem.:confused:

EV maximum!
 

bjcount

Well-Known Member
#16
jack said:
Did you ever figure why this was happening, because I'm having the same problem?

It doesnt make any sense, because the count has to be ptetty high, not to split. It would at least make a little more sense if it spit-out ALWAYS split, but it keeps saying NEVER split.

Its like ENHC is enabled, but Im getting the correct indices for 11,10 vs Ten, so I dont think this is the problem.:confused:

EV maximum!
JJ

I have always split using Zen RA. In my RPC RA I use split < 2. The hands a big loser either way and the difference between splitting always and splitting <2 is about 0.04 cents. The difference in splitting always and splitting at >2 is a loss of 0.10 cents. (per the sims I ran). The higher TC just causes you to lose more money.

I always surrender it at tc=0 when LS is available. I look at it this way, if I can't surrender, then split when the shoes is neg. or slightly pos.. 18 is still a loser against a 10. Why put more money out on a high count in a bad hand when the high cards are not going to help you and chances are better that the dealer has 20. (other than if the dealer has a low card in the hole and you know it.) In a low count the dealer just may have a low card in the hole and you can pull two cards on a split to make a pat hand. Just IMO.



The EV prob chart is:
Hand Standing Hitting Doubling Splitting Probability
8,8 vs T -0.53685 -0.53536 -1.07072 -0.4748 0.00176
T,8 vs T -0.17455 -0.64375 -1.28749 -------- 0.014553

Maybe with EV max indices it says never split since its a losing hand so why put out twice the money into a losing situation? Anyone else have a logical explanation?


BJC
 
Last edited:

jack.jackson

Well-Known Member
#17
Thanks Bj

bjcount said:
JJ

I have always split using Zen RA. In my RPC RA I use split < 2. The hands a big loser either way and the difference between splitting always and splitting <2 is about 0.04 cents. The difference in splitting always and splitting at >2 is a loss of 0.10 cents. (per the sims I ran). The higher TC just causes you to lose more money.

I always surrender it at tc=0 when LS is available. I look at it this way, if I can't surrender, then split when the shoes is neg. or slightly pos.. 18 is still a loser against a 10. Why put more money out on a high count in a bad hand when the high cards are not going to help you and chances are better that the dealer has 20. (other than if the dealer has a low card in the hole and you know it.) In a low count the dealer just may have a low card in the hole and you can pull two cards on a split to make a pat hand. Just IMO.



The EV prob chart is:
Hand Standing Hitting Doubling Splitting Probability
8,8 vs T -0.53685 -0.53536 -1.07072 -0.4748 0.00176
T,8 vs T -0.17455 -0.64375 -1.28749 -------- 0.014553

Maybe with EV max indices it says never split since its a losing hand so why put out twice the money into a losing situation? Anyone else have a logical explanation?


BJC
Well maybe if I was generating RA indices, it would make a little more sense, but im not. CVdata should still spit-out an index, for this hand. Ive also got DAS enabled.

To make things easy, im just generating the split indexes, w/"quick and dirty" selected, so it doesnt, take very long(20min) before it gets to 88vX. No matter what I do, it just says Never to split them. Even if the difference is small, it should still give me an index. Thanks again!
 
Last edited:

bjcount

Well-Known Member
#18
jack said:
Well maybe if I was generating RA indices, it would make a little more sense, but im not. CVdata should still spit-out an index, for this hand. Ive also got DAS enabled.

To make things easy, im just generating the split indexes, w/"quick and dirty" selected, so it doesnt, take very long(20min) before it gets to 88vX. No matter what I do, it just says Never to split them. Even if the difference is small, it should still give me an index. Thanks again!
JJ

Run a sim just for the 88vX by clicking the split index box and uncheck all. Then just check 88 v X and set it to BEAT TO DEATH. It will probably take half an hour to run the one index but then you will have an exact result. Make sure you did not check any of the boxes such as big 63, I18, etc, you want full indices to run so it doesnt block out the 8,8 v X index which may be what is happening. Also make sure your playing strategy is using full indices, not I18, fab4, etc. See if that helps.


BJC
 
Last edited:

jack.jackson

Well-Known Member
#19
bjcount said:
JJ

Run a sim just for the 88vX by clicking the split index box and uncheck all. Then just check 88 v X and set it to BEAT TO DEATH. It will probably take half an hour to run the one index but then you will have an exact result. Make sure you did not check any of the boxes such as big 63, I18, etc, you want full indices to run so it doesnt block out the 8,8 v X index which may be what is happening. Also make sure your playing strategy is using full indices, not I18, fab4, etc. See if that helps.


BJC
The Beat to deaths, didnt make any difference.

The problem was,for some reason, you must genereate, the H/S indexes, first, otherwise you'll get NEVER split on 88vX. As far as the other split indexes go, I believe you can Generate these(accurately) without having to Generate the H/S indexes first. Thats what made it so hard to track down.

One of the reasons, I was doing it this way, was because I was Generating seperate indexes, for 2 different counts. Also, make sure your playing strategy name and index generation name are the same when running seperate indexes.

On a side issue, could I get your opinion on which of these counts you think has better efficiency for 88vX? I cant decide!

a)12223210-1-3(A/X)

b)2223210-1-2-3(2/A)
 
Last edited:

ace157

Well-Known Member
#20
has ne1 ever tried to gen indexes and got a "runtime error 6: overflow" error notice?

what does it mean and how do i fix it?
 
Top