Question on Shuffle quality and impact on Card Counting Strategies

[bev005]

You suggest writing a simulator to test your ideas and I encourage you to do just that. That's exactly what I did when I started on this journey and needed a way to separate the wheat from the chaff when it comes to published information. This is not a new idea and there are hundreds of simulators out there, some commercial but most written by individuals for their own needs. All of us who have performed this exercise will tell you in advance that the true count does indeed rise and fall enough to result in significant advantage and disadvantage even when using a uniformly distributed RNG to shuffle the cards. We can also confirm with you what the expected true count frequencies and advantages are among other things.

You don't need to believe everything you are told on faith but please, do not simply dismiss the advice and opinions of those of us who have been on this journey for years and even decades. The math and the software needed to do this are not rocket science and, while you are encouraged to embark on this journey, please understand that most of us have already asked the same questions and are already well ahead of you.

Thanks it is a very wise advice. And I assure you I do respect all opinions and past work been done. It wasn't intent of my message. I just wanted to save some time if someone already made such analysis , not to repeat it.

[bev005]

LOL, have fun !

Well it is actually fun I never thought in my life that simple 21 can in fact be so reach of theory. So I do have it. Reading these books by Grandmasters also fun. Well combinatorial analysis makes me bring back old math books as well. Long time forgot staff need to refresh in order to understand ideas.

So thank you.

Still it would be interesting to train the Model similar to DeepMind's Alpha Zero or Alpha go and train to play BJ. Wondering what strategy it will tune herself to and what would her win rate. In Chess and Go her strategies were way different from all known calculated by generations of humans.

[G Man]

But still the question remains how often such clusters occur and for how long in order to tc rise and stay reasonably high so someone could really get advantage of that.

You don't get any advantage when a TC stays high. You get an advantage has the TC goes down. In fact, when the TC is high, you make a bet on the fact that it will fall.
Note: I reread my post... I should say RC because the TC has a tendency to stay the same, but you want the RC to go down in order to materialize your advantage.

[bev005]

You don't get any advantage when a TC stays high. You get an advantage has the TC goes down. In fact, when the TC is high, you make a bet on the fact that it will fall.
Note: I reread my post... I should say RC because the TC has a tendency to stay the same, but you want the RC to go down in order to materialize your advantage.

Makes sense in respect of RC. Yeah you want to get those cards to win.

Here is one of example of game against 6 deck prng 1 player s17 das (private table so no rush) CAC2 system. penetration about 4 and 1/2 decks.

The only thing- hard to asses discard tray on app -so i approximated that 1 player and dealer usually use 1/2 deck in 5 rounds (23-33 cards). imho good enough approximation compared to errors you do estimating tray.

So divider 12 - rc from 0 went -17 tc -1
divider 11 - rc from -17 went to -7 tc -0
divider 10 - rc from -7 went -10 tc -1
divider 9 - rc from -10 went to +8 tc +1
divider 8 - rc from +8 went to +17 tc +2 lost all 3 x2 bets
divider 7 - rc from +17 went to +6 tc +1
divider 6 - rc from +6 went to +1 tc =+0
divider 5 - rc from +1 went to -7 tc = -1
divider 4 - rc from -1 went to -7 tc = -2 3 rounds shuffle card

total result = -8 units

sometimes result positive but not because of a spread. it would be positive even with flat bet. variance as you call it.

And this picture is quite typical for majority of the shoes i played on this app. Rarely i had tc rising to +3 +4 for hand or two.

Yeah question if they programmatically tweak something to compensate against card counters. This is also question to be answered.

Thought wanted to write pure simulator to see if such case is typical for all prng implementations.

[Gronbog]

Then tried to play on a casino app with PNRG shoes. Same result. As a matter of fact I could easily just use flat betting Basic Strategy with almost same if not better result.

You might not know this but, unless stated otherwise, casino apps generally shuffle after every round. Online casino apps in particular.

Hence I started to think into has anyone done any statistical analysis on this.

Blackjack could possibly be the most researched game in existence.

I just don't see it so far working in a real life tests.

You mention playing around 60 shoes at home. This is far too few to form a meaningful statistical sample. That's why we write simulators. The variance inherent in the game is such that even professional counters playing every day will occasionally experience a losing month.

[bev005]

You might not know this but, unless stated otherwise, casino apps generally shuffle after every round. Online casino apps in particular.

No I didn't Really!!!!! But I verified the shoe is formed correctly. It is not an infinite shoe kind of thing. They don't return dealt cards. If they shuffle only remained shoe it is imho shouldn't affect probabilities.

Blackjack could possibly be the most researched game in existence.

Yeah I have figured that out recently Amazing! Never could imagine that 21 will be as interesting for me as chess.

You mention playing around 60 shoes at home. This is far too few to form a meaningful statistical sample. That's why we write simulators.

100% agree

The variance inherent in the game is such that even professional counters playing every day will occasionally experience a losing month.

Variance yes and I completely understand that 50/50 works only on large numbers. That doesn't eliminate possibility 15 heads or 15 tails in a row.

But IMHO everytime someone uses the "Variance" argument they forget that there is still valid the possibility of variance to occur.

If variance happens way too often it is no longer variance it is pattern.

Can you flip 15 heads once upon a time yes. But if you flip 15 heads streaks 2-3 times in front of me you are cheating. It is impossible IMHO.

[bev005]

>>>>That's why we write simulators.

Forgot to ask how much confidence you have that simulators you write folks via PRNG and use to get all these amazing tables represent the physical shoe shuffled by hand and even not perfect machine.

If to summarize that's why I asked myself if shuffle algorithms we use have an effect. And when you get numbers received via simulator to live shoes the situation can dramatically defer.

This is a question I ask not statement This is a question I seek answers to in a nutshell.

What if you change shuffle algorithms in your simulators? BTW Ihave no idea which ones are being used.

What if we create riffle algorithm which varies the number of cards you grab mimic the human shuffler.

Will it change the results you get?

You are using billions of hands to be statistically correct but within the 1000 hands realm that say weekend player plays the World is different so to speak.

[DSchles]

You're pursuing a dead-end. All of your ideas have been researched 1,000 times. Time to move on.

Don

[Gramazeka]

You're pursuing a dead-end. All of your ideas have been researched 1,000 times. Time to move on.

Don

Don, please note that he downvoted my post where I sent him a link to software by Norm. LOL !

[Gronbog]

What if you change shuffle algorithms in your simulators? BTW Ihave no idea which ones are being used.

What if we create riffle algorithm which varies the number of cards you grab mimic the human shuffler.

Will it change the results you get?

As Don says, this has all been researched to death. The answer is no statistically significant difference to the basic strategy or index strategy player. I know that Norm has data on this. Perhaps he will chime in.

You are using billions of hands to be statistically correct but within the 1000 hands realm that say weekend player plays the World is different so to speak.

For shorter term analysis you can use the standard deviation to determine the probability that your results will be +/- a given amount from expectation.

May I suggest that you read through the information here

https://www.blackjackincolor.com/

before you ask any more basic questions. You'll get your answers faster and gain a more solid foundation of knowledge at the same time.

[bev005]

I already have bought the software but it is a tool without having access to the code it is not helpful sorry, hence downvoted. No offense.

[bev005]

Maybe .

I need to read more probably.

[Gronbog]

It is the premier gold-standard software in our community and was written by our host, Norm. I have vetted my own software against it many times (i.e. achieved independently identical results).