Counting Virtual Blackjack
In this situation with out doing any other tracking, how would you know when to raise your bets? Are you saying, since you would be right part of the time and wrong part of the time, you could have a +ev situation over all, if you waited long enough for a good count and had huge spreads?Quote: MangoJIf you don't know the shuffle point, a rough edge estimate would be similar to a game with penetration halfed, i.e. 1/3 in your case.
LOLQuote: BleedingChipsSlowlyIf you are playing a ShuffleMaster machine, I have heard the theory that the shuffle is done when they change the dealer pictured on the video screen.
My intuition is that it gives additional information that can be used (how much difference it makes is a different question)
The shuffle card could be anywhere in the 6 decks, each time you see a card there is 1/312 probability the the shuffle card just appeared
Say you observe and count 52 cards (1 deck)
Then 5/6 (52/312) of the time you did not encounter the shuffle point. So 5/6 of the time you have the correct count and know the correct penetration.
1/6 of the time you have the wrong count and wrong penetration (which can vary depending where the shuffle card appeared in the previous 52 cards)
It seems to me that the above give additional information that can be used.
An extreme exqample, you count this 52 cards and get a running count of +20 and a True count of +4 (20/5)
5/6 of the time the TC of +4 is correct
1/6 of the time the TC of +4 is incorrect . BUT still many of those times the TC wouls still be positive (ie shuffle card apepeard 50 cards before, so you only overstimated the first 2 cards)
So the question is how to you use this information and how much difference in HE does it make?
Quote: AceTwoI also would like to know (at least on a theoretical level) from someone who thought about this at a mathecmatical level, whether knowing that there is a shuffle at 4/6 gives any additional information that can be used.
My intuition is that it gives additional information that can be used (how much difference it makes is a different question)
The shuffle card could be anywhere in the 6 decks, each time you see a card there is 1/312 probability the the shuffle card just appeared
Say you observe and count 52 cards (1 deck)
Then 5/6 (52/312) of the time you did not encounter the shuffle point. So 5/6 of the time you have the correct count and know the correct penetration.
1/6 of the time you have the wrong count and wrong penetration (which can vary depending where the shuffle card appeared in the previous 52 cards)
It seems to me that the above give additional information that can be used.
An extreme exqample, you count this 52 cards and get a running count of +20 and a True count of +4 (20/5)
5/6 of the time the TC of +4 is correct
1/6 of the time the TC of +4 is incorrect . BUT still many of those times the TC wouls still be positive (ie shuffle card apepeard 50 cards before, so you only overstimated the first 2 cards)
So the question is how to you use this information and how much difference in HE does it make?
There is not a 1/312 chance you just encountered the shuffle card. Implicitly, you are stating the shuffle card could be behind the 312th card of the shoe, when the machine states that the card always follows the 208th card. You also do not know whether the computer immediately shuffles, or completes the hand first.
No one is going to invent or buy a blackjack machine that will be undersupervised by casino personnel and that can be beaten by counting. The difference in house edge is minimal. To play a game like this, perfect composition-dependent strategy (which I rebuke) is the proper way to go, not counting.
The only reason the game would offer a shuffle point is to attract people who think this is a good thing; they know that for the typical player, shuffling after every hand is either favorable or has no effect because of the cut card effect. You can't guess where you are at in the shuffle. You won't be wrong 1/6 the time, you will be wrong 99.9% of the time. 50% of the time, you will be way off the mark.
In your example, there is a 75% chance that the count is +4 at 1/6 penetration. Counting this way would slightly decrease the house edge, so long as you adjust the count to the probability you are wrong. I don't usually bother, but it might be helpful here to remember that high counts at the beginning of a shoe are less advantageous than the same counts at the end of a shoe. So your +4 that I would convert to +3 should be bet as +2, and how often do you see a +20 RC after 1 deck?
I estimate the impact on house edge thru counting would be less than than using composition dependent strategy, probably somewhere around .01%. You can only appropriately increase your bet at the "beginning" of a shoe at rare counts.
So the figures in my example should be adjusted accordingly.
Obviously the computer shuffles after the hand is played. So sometimes you just encounter the shuffle point and the count remains. But it is a minor issue.
You seem though to agree with the main theoretical point I was making that the knoweldge that there is a shuffle point (even if you do not know where it is) provides additional information that can be used to reduce the HE. It is extra information. It is not neutral.
I agree that the difference that it makes in HE is probably small.
You say that Composition Dependent strategy would impact more. Well you can use both Composition Dependent strategy + 'Counting Strategy'.
So this strategy will impact more, so there is value in it.
You say only 0,01%. I agree that the impact would be small but not that small.
Obviously the impact depends on spread . But in a machine you would have zero (minimal) heat, and wong in and wong out at any time. Especially if there are many machines you could wong out and start another machine whenever it is negative.
So the impact should be calculated at assumptions not usually possible at real BJ: ie spread 1:100.
The question is though what is this correct 'Counting' Strategy.
When do you start to count, do you assume after a point taht the shuffle has come.
Lets say under optimal conditions to start with. Like yiu can have pen and paper and write things down.
My intution tells me that the following would be the 'counting' strategy.
You sit down and start counting upto a certain optimal point (say X) which is less than the 208 cards (4 decks).
Upto counting to this optimal point you assume that the shuffle card has not appeared and you calclulate TC based on the cards counted.
SAYotimal point is at 156 cards (3 decks).
After you reach this optimal point, you take into account as Count the last 156 cards only.
Ie, when you encounter the 157 card you count that one BUT we adjust for the first card you observed (ie take it out of the count).
Then for the 158 card you adjust for the 2nd card etc. (Obviously you need pen and paper if not a laptop to do this)
To find the optimal point you need to run sims to find what it is. And also to find the corresponding change in advantahe for each TC.
The change in advantage would obviously be less than the around 0,5% change associated with normal counting.
The optimal point could be somewhere around 2 decks for 4/6 pen.
And with a huge spread the game could turn positive. ie the impact on HE could be in the 0,30%-0,50% range
Quote:You sit down and start counting upto a certain optimal point (say X) which is less than the 208 cards (4 decks).
Upto counting to this optimal point you assume that the shuffle card has not appeared and you calclulate TC based on the cards counted.
SAYotimal point is at 156 cards (3 decks).
After you reach this optimal point, you take into account as Count the last 156 cards only.
Ie, when you encounter the 157 card you count that one BUT we adjust for the first card you observed (ie take it out of the count).
Then for the 158 card you adjust for the 2nd card etc. (Obviously you need pen and paper if not a laptop to do this)
Say our window is 3 decks. First we have to remember that we will be full betting our counts through the whole shoe. This will be painful for the first deck of a shoe (only known to the machine), and not particularly good though the second (occasionally it will be). However, somewhere in the third deck we can expect our count to almost match the true. And for the final deck it will be correct. Problematically, we don't know how our running count converts to true count, so there is that issue also, but lets not worry about that.
Lets say that for 2 out of the 4 decks our count is worthless and for 2 decks it is useful.
Since we don't know where those shuffle points are, I would play basic strategy all the time and not do any of the play deviations (this assumption may be called into question, but I think it fine for this back of the envelope argument).
Now the standard HE depends on the rules. Lets say for the half the time you have the right count your average edge +.5% (remembering that you have to play through the negative counts too). This only cancels the HE for when you don't have the right count (approx -.5%). Remembering that our average bet size will be the same for right counts and wrong counts, we get that this a break even proposition. Better than losing, but no true advantage.
Now maybe you could try card eating whenever you have bad counts, and with a huge spread (to push up your Adv in plus counts when you have the right count) the game may be -just barely- beatable.
ON THE OTHER HAND...
Just because it is a machine doesn't mean there won't be heat. In, fact it would be VERY easy to program in a counting subroutine which can then look for correlation in player's betting and the true count or the procedure outlined above. If I owned a casino, I might have machines like these sitting out as traps for counters. The subroutine pings, we review the data (already digital!), and make a decision.
Sorry, I should have separated my original post in two so that I made the distinction better (idealized math versus reality).
Quote: endermikeOf course not, we had already said we are bringing in a pen and paper/laptop to track it. This was a purely mathematical question. To me optimal window size is the interesting part (given machinery or Rainman abilities).
Sorry, I should have separated my original post in two so that I made the distinction better (idealized math versus reality).
Ok, but in that case, you can do better than a single-parameter count (why not perfect betting and playing?).
While you are at it, why bother to fix a sliding window size at all? Simply take the weighted average of all possible states.
Quote: AxiomOfChoiceOk, but in that case, you can do better than a single-parameter count (why not perfect betting and playing?).
While you are at it, why bother to fix a sliding window size at all? Simply take the weighted average of all possible states.
I agree, that is a more sensible conclusion.
One better: we could really get after it and sic our laptop on trying to discover the underlying RNG in the game and then beat the shit out of it.
Quote: AceTwoYes you are right. The probability that I just encountered the shuffle is 1/208 (and not 1/312 as I state).
So the figures in my example should be adjusted accordingly.
Obviously the computer shuffles after the hand is played. So sometimes you just encounter the shuffle point and the count remains. But it is a minor issue.
You seem though to agree with the main theoretical point I was making that the knoweldge that there is a shuffle point (even if you do not know where it is) provides additional information that can be used to reduce the HE. It is extra information. It is not neutral.
I agree that the difference that it makes in HE is probably small.
You say that Composition Dependent strategy would impact more. Well you can use both Composition Dependent strategy + 'Counting Strategy'.
So this strategy will impact more, so there is value in it.
You say only 0,01%. I agree that the impact would be small but not that small.
Obviously the impact depends on spread . But in a machine you would have zero (minimal) heat, and wong in and wong out at any time. Especially if there are many machines you could wong out and start another machine whenever it is negative.
So the impact should be calculated at assumptions not usually possible at real BJ: ie spread 1:100.
The question is though what is this correct 'Counting' Strategy.
When do you start to count, do you assume after a point taht the shuffle has come.
Lets say under optimal conditions to start with. Like yiu can have pen and paper and write things down.
My intution tells me that the following would be the 'counting' strategy.
You sit down and start counting upto a certain optimal point (say X) which is less than the 208 cards (4 decks).
Upto counting to this optimal point you assume that the shuffle card has not appeared and you calclulate TC based on the cards counted.
SAYotimal point is at 156 cards (3 decks).
After you reach this optimal point, you take into account as Count the last 156 cards only.
Ie, when you encounter the 157 card you count that one BUT we adjust for the first card you observed (ie take it out of the count).
Then for the 158 card you adjust for the 2nd card etc. (Obviously you need pen and paper if not a laptop to do this)
To find the optimal point you need to run sims to find what it is. And also to find the corresponding change in advantahe for each TC.
The change in advantage would obviously be less than the around 0,5% change associated with normal counting.
The optimal point could be somewhere around 2 decks for 4/6 pen.
And with a huge spread the game could turn positive. ie the impact on HE could be in the 0,30%-0,50% range
I do not see how composition dependent strategy and counting are compatible. When the count above zero, you stand on 16 v. 10. When you have 10. 6 v. 10, you hit. You are going to often get two contradictory decisions from using two different strategies. Composition-dependent strategy enhances basic strategy, not counting. You gain about a 1% advantage from typical counting. 99% of your advantage comes from counts arising after the first deck, and from using play deviations. I do not think .01 is a low estimation. Perhaps a range is more suitable: I would estimate between .005-.03. I imagine you are not fully incorporating the facts that you are not even close to 100% certain that your count is accurate, and that you cannot even count beyond extremely poor penetration levels.
I could not speak to an optimal point. But if you count more than 104 cards, you are less than 50% likely to be accurate at only 1/3 penetration and would only increase your risk of ruin without changing the house edge. 103 cards would be the maximum countable.
If you used a computer to map the cards in realtime, you could probably find a likely shuffle point. I think this would take quite a long time, you would not be able to leave the machine without re-doing the entire analysis, and I do not think this is legal. Since there are plenty of table games with better counting opportunities, this seems at first glance, to be a terrible idea.
At 1/3 penetration in atable game assuming a play-all situation, you would barely dent the house edge with counting. Decrease it by almost 0.1% if you were lucky. And in a table game, you are 100% sure where the cut card is. You would have to wong out and play only 'probable' positive counts. If you are spreading from $1 to $500--then your numbers are probably closer because I have no knowledge or experience with anything so outlandish. I have been assuming typical spreads.
If we wanted to step back inside the realm of reality, if I were trying to count this game in my head, I think I would have a chance of tracking the count round by round. (There would probably with long pauses in my play, but it is a machine, so lets go with it). Lets assume I play all the spots (6 of them). This means that each round will pull around 20 cards per round (would definetly need to find the true number, but lets go with it). So to track the count of about 3 decks, I would need to keep 7-8 rounds of hands count in my head
Example:
-Round 1: 6 low and 7 high ->-1
What I need to remember m1
-Round 2: 4 low and 6 high ->-2
What I need to remember m1, m2
-Round 3: 8 low and 3 high ->+5
What I need to remember m1, m2, p5
...
-Round 8: 6 low and 6 high ->0
What I need to remember m1, m2, p5, ..., e
-Round 9, bet using sum of the running count, lets say +2: 4 low and 5 high ->+1
What I need to remember m2, p5, ..., e, p1
-Round 10, bet using sum of the running count, now it is +3
and so on
Again this gives up some amount of the edge from counting, but it is something which could be done in a real casino environment.
With some tracking you might be able to get a feel for when your count is "correct-ish" and when it is bad. I'm guessing this would take a significant number of shoes, but it is not beyond the skills of anyone is able to write some things down in a subtle manner. However, I would guess that after a while we might be able to get a good sense of "where-ish" the shuffle point is then we might be able to get closer to a standard counting process.
In summary, if:
#1 Huge (maybe even just medium to large depending on #3) spreads are available
#2 I was allowed even a pad of paper like in a baccarat table
#3 It has even decent rules
#4 It actually deals 4 decks plus or minus a hand each time
#5 I (or surrogates) can play all the spots, slowly if needed
I would wager I could break this game.
Quote: endermikeif:
#1 Huge (maybe even just medium to large depending on #3) spreads are available
#2 I was allowed even a pad of paper like in a baccarat table
#3 It has even decent rules
#4 It actually deals 4 decks plus or minus a hand each time
#5 I (or surrogates) can play all the spots, slowly if needed
I would be happy to wager a small amount of money ($5-$20) on the this proposition. The key would be finding an impartial judgement on this issue. That I believe this will be difficult. I am not the OP, and do not know where said machine exists. Hence confirming said assumptions seems difficult, most particularly #4.
Is there a particular flaw in my analysis you see? Or were you simply saying that you believe at least one of my assumptions will be impossible to fulfill (because I think that seems totally possible).
However, as of yet I stand by my (limited) analysis that this game could be beaten under the assumptions listed so far.
Quote: endermikeI was using "wager" in the cliche sense ("I think it is likely"). Obviously not a safe choice on this site. However, I do fully believe that given the 5 assumptions outlined the game is beatable.
I would be happy to wager a small amount of money ($5-$20) on the this proposition. The key would be finding an impartial judgement on this issue. That I believe this will be difficult. I am not the OP, and do not know where said machine exists. Hence confirming said assumptions seems difficult, most particularly #4.
Is there a particular flaw in my analysis you see? Or were you simply saying that you believe at least one of my assumptions will be impossible to fulfill (because I think that seems totally possible).
However, as of yet I stand by my (limited) analysis that this game could be beaten under the assumptions listed so far.
No serious counter would even consider doing this for real. So I would say that's the main flaw. Anyone willing to do this for profit would not know what they are doing.
How would you even know you were beating it? Anyone can win a series of hands.
Theoretically, a huge spread might be able to get an edge. But very few would have the BR for that, and it wouldn't make much sense to waste it on such an extremely poor game. In addition, there are variables unaccounted for that could make the game unbeatable no matter what. For instance, it is not blackjack it is a slot machine. There's no reason I can see that it isn't 'rigged' in a sense that isn't necessarily cheating, but tends to alter the payout.
I'd give you a pen and paper, but no computer and no team of players/spotters. Use whatever spread you want. No wonging. An expert has to conclude you played with an advantage. You cannot bust your bankroll. You must double it, or reach a set time limit. $1000 minimum wager. I'm gonna bet $5 dollars? No. It's fair to say I'm more confident than you are. Someone should make a poll...can VBJ be beaten?
Quote: BizzyBNo serious counter would even consider doing this for real. So I would say that's the main flaw. Anyone willing to do this for profit would not know what they are doing.
How would you even know you were beating it? Anyone can win a series of hands.
Theoretically, a huge spread might be able to get an edge. But very few would have the BR for that, and it wouldn't make much sense to waste it on such an extremely poor game. In addition, there are variables unaccounted for that could make the game unbeatable no matter what. For instance, it is not blackjack it is a slot machine. There's no reason I can see that it isn't 'rigged' in a sense that isn't necessarily cheating, but tends to alter the payout.
I'd give you a pen and paper, but no computer and no team of players/spotters. Use whatever spread you want. No wonging. An expert has to conclude you played with an advantage. You cannot bust your bankroll. You must double it, or reach a set time limit. $1000 minimum wager. I'm gonna bet $5 dollars? No. It's fair to say I'm more confident than you are. Someone should make a poll...can VBJ be beaten?
In fairness, lots of people have the bankroll for that.
Also, if the game is in Nevada, the probabilities for the cards have to be the same as from a normal shoe of cards. It's not like a slot machine where the reels can be weighted. Anything that represents cards or dice has to be fair.
The problem of analyzing results to determine whether there's an edge just comes down to how sure you want to be. The variance of a hand is easily determined (it changes based on the rules -- limiting splits and doubles lowers it since you have less money out there on average). From that, you can determine what the requirements would be to have 95% confidence or 98% confidence or whatever confidence you want that the game is being beaten (of course, you can never get the confidence level all the way to 100% -- no matter how long you beat a game for, there is always some tiny possibility that the results are due to luck)
I was, and am, asserting mathematical analysis would show that under certain parameters this game could be beaten. Real world play is known to be glacially slow at revealing true edge.
We don't know min bet or max bet so the bankroll requirements may be less than standard table counting levels. However, that would probably destroy average value per hour, hence why it is unlikely to be a serious threat to a casino even if the game is beatable. The question is beat-ability, not real profitability.
It is impossible for you to have such certainty about a game where you do not know the rules. (Do you have info not listed in the thread?) The machine could be offering early surrender (good) or 6:5 BJ (bad). Right now, operating under the assumption of standard strip rules and the caveats I have listed, this game could be beaten.
As I stated, I am not the OP, so I am in no position to actually carry this through in the real world. His question was "could there be advantage?", and the answer is yes. We extended the debate to "could it be beaten?" and the answer there is maybe. Depends on info we don't have. Some video BJ machines are impossible to beat even with insane counting assumptions (which don't hold since most machines shuffle after each round). However, some are much closer to liberal table BJ, and if such rules were present in this case, and the game plays as OP stated, the game can be played at +EV (but probably not +EV over opportunity cost).
Quote: AxiomOfChoiceIn fairness, lots of people have the bankroll for that.
Also, if the game is in Nevada, the probabilities for the cards have to be the same as from a normal shoe of cards. It's not like a slot machine where the reels can be weighted. Anything that represents cards or dice has to be fair.
The problem of analyzing results to determine whether there's an edge just comes down to how sure you want to be. The variance of a hand is easily determined (it changes based on the rules -- limiting splits and doubles lowers it since you have less money out there on average). From that, you can determine what the requirements would be to have 95% confidence or 98% confidence or whatever confidence you want that the game is being beaten (of course, you can never get the confidence level all the way to 100% -- no matter how long you beat a game for, there is always some tiny possibility that the results are due to luck)
I don't think I agree with what you are saying here, with all due respect. A normal huge spread is not bigger than 1-16. You cannot beat this with 1-16. So say your unit size is $25 and this game requires $1. You can bet 1/25th of a unit. And then bet 16 units. Very few people having that probably is not the correct thing to say, but I think I covered it when I said no one is gonna waste their BR on that. This is something a newbie would do, and very few newbie would have a BR capable of high spread on this game...that's my opinion. I wasn't thinking or wording things very clearly. But that is my meaning for that.
I would like to read the Nevada statute or regulation because I doubt any language goes so far to protect the card counter as you suggest. The casino is obliged to offer a fair game, but has no obligation to make sure it can be beaten by a professional counter. An RNG can determine win, loss, or condition precedent the moment the button is pressed. It can give then give you cards to fulfill its determination. The dealer does not have to pull cards from the top of the deck. It can take them from anywhere before the shuffle point. A card is no more likely to be in the middle than it is on the top. The dealer is looking at all the cards face-up, and they are invisible to you. There are no probability violations. That is only speculation, but it is known that RNGs increase variance and decreases bias. Seems a fools errand to attempt to detect bias when it is not there.
I would settle for 95% confidence the game is being beaten. Your method seems to require keeping track of every single hand played, but I guess that is probably the only way since it cannot be simulated. I'm not sure I would trust the person to report accurately.
But endermike said about playing all 6 spots. I googled for these machines and I see that they are multiplayer machines where many players play together.
If that is the case (ie many players play on the same 6 decks) then you can have have infinite spread !!!
I mean, you can only make a bet only when you have the advantage.
I assume that the machine gives you some time to press the buton to put a bet and if you do not then the game for that round continues with the rest of the players. Of course it will piss off many players if you start/stop playing hands but we are talking optimum conditions.
So by definition if you play only hands that you have advantage you can beat this game.
Of course most serious players definition of beating a game does not mean that. It means to beat the game by enough % or more correctly by having at least $x win rate per hour.
I think most of the posters agree with the theoretical point that the knoweledge of a shuffle point gives extra information that can be used by the player to his advantage. The disagreement is by how much. And also what should be the strategy.
These can only be answered by sims. Existing BJ sim programs cannot do such a sim as far as I know. So someone needs to design/alter a BJ sim software to get the answer.
I am not so sure that even by using a computer to analysis thousands of cards observed will tell you where the shuffle point is or even give you probabilities of the shuffle point being at a specific point.
The Computer will analyse the following cards observed (lets disregard for simplicity that after the shuffle card, more cards are played to finish the hand, assume shuffle begins after the 208th card) :
Cards 1-208, 209- 416, 417-624 etc AND
Cards 2-209, 210-417, 418-625 etc and so on for all possible 208 positions the shuffle card can be.
I cannot think of what kind of parameter(s) or patterns will say than one position is more probable than the other
Quote: BizzyBI don't think I agree with what you are saying here, with all due respect. A normal huge spread is not bigger than 1-16. You cannot beat this with 1-16. So say your unit size is $25 and this game requires $1. You can bet 1/25th of a unit. And then bet 16 units. Very few people having that probably is not the correct thing to say, but I think I covered it when I said no one is gonna waste their BR on that. This is something a newbie would do, and very few newbie would have a BR capable of high spread on this game...that's my opinion. I wasn't thinking or wording things very clearly. But that is my meaning for that.
It's not really the size of the spread that's relevant, it's the size of the max bet. If you have the BR to spread $10-$100 then you have the BR to spread $1-$100.
Quote:I would like to read the Nevada statute or regulation because I doubt any language goes so far to protect the card counter as you suggest. The casino is obliged to offer a fair game, but has no obligation to make sure it can be beaten by a professional counter. An RNG can determine win, loss, or condition precedent the moment the button is pressed. It can give then give you cards to fulfill its determination. The dealer does not have to pull cards from the top of the deck. It can take them from anywhere before the shuffle point. A card is no more likely to be in the middle than it is on the top. The dealer is looking at all the cards face-up, and they are invisible to you. There are no probability violations. That is only speculation, but it is known that RNGs increase variance and decreases bias. Seems a fools errand to attempt to detect bias when it is not there.
I'm not sure what you are saying here.
If the card is being picked randomly (with uniform distribution), it doesn't matter whether it is being picked from the top, second from top, middle, or bottom of the deck. It doesn't affect the expectation at all -- there is nothing special about the top card. Random is random.
If it is not being picked randomly, with uniform distribution (eg, if they are weighting the cards that would make you lose higher than the cards that would make you win) then it is not legal in the state of Nevada. This is no different from the rules that govern video poker, or any slot machine bonus game that depicts cards or dice (eg, see the WoO's recent write-up about deconstructing the Hot Roll slot machine -- they can weight the slot reels however they want, but the dice rolls must be independent with each face having a 1/6 probability of showing up)
Quote:I would settle for 95% confidence the game is being beaten. Your method seems to require keeping track of every single hand played, but I guess that is probably the only way since it cannot be simulated. I'm not sure I would trust the person to report accurately.
I'd just be curious to know the rules of the game to see whether the game is even theoretically beatable, and by how much. I'm guessing that it's not actually worth playing, but it's still interesting IMO.
Quote: AceTwoI have never seen these Virtual BJ machnines (I am not from the US). I assumed it was a single player, single spot machine.
But endermike said about playing all 6 spots. I googled for these machines and I see that they are multiplayer machines where many players play together.
If that is the case (ie many players play on the same 6 decks) then you can have have infinite spread !!!
I mean, you can only make a bet only when you have the advantage.
I assume that the machine gives you some time to press the buton to put a bet and if you do not then the game for that round continues with the rest of the players. Of course it will piss off many players if you start/stop playing hands but we are talking optimum conditions.
So by definition if you play only hands that you have advantage you can beat this game.
Of course most serious players definition of beating a game does not mean that. It means to beat the game by enough % or more correctly by having at least $x win rate per hour.
I think most of the posters agree with the theoretical point that the knoweledge of a shuffle point gives extra information that can be used by the player to his advantage. The disagreement is by how much. And also what should be the strategy.
These can only be answered by sims. Existing BJ sim programs cannot do such a sim as far as I know. So someone needs to design/alter a BJ sim software to get the answer.
I am not so sure that even by using a computer to analysis thousands of cards observed will tell you where the shuffle point is or even give you probabilities of the shuffle point being at a specific point.
The Computer will analyse the following cards observed (lets disregard for simplicity that after the shuffle card, more cards are played to finish the hand, assume shuffle begins after the 208th card) :
Cards 1-208, 209- 416, 417-624 etc AND
Cards 2-209, 210-417, 418-625 etc and so on for all possible 208 positions the shuffle card can be.
I cannot think of what kind of parameter(s) or patterns will say than one position is more probable than the other
You can beat any shoe game with an 'infinite' spread if you wong in only on positive counts, all of which have better penetration, making for a better opporunity. Not sure why you are so excited, as here the potentially more limited opportunity might be completely illusory. So you are going to look to see if the shuffle card is in deck #12, even tho there are only 6 decks? It's pretty clear what kind of people are going to be trying to beat VBJ. The casino execs know what they are doing.
And seriously, I keep hearing computer. What computer? Are you guys gonna write down every hand, go home, and try to figure out the shuffle point with your computer? How are you going to know where it is once you get back to the casino? Your only hope is that it is indeed when they switch dealers. When no one is playing, they change dealers. I have seen it. There goes that theory.
Quote: BizzyBAnd seriously, I keep hearing computer. What computer? Are you guys gonna write down every hand, go home, and try to figure out the shuffle point with your computer? How are you going to know where it is once you get back to the casino? Your only hope is that it is indeed when they switch dealers. When no one is playing, they change dealers. I have seen it. There goes that theory.
If you read what I said is that I think that theoretically even with a computer you can not know when the shuffle point is. Not even attribute higher probabilities for specific points for the shuffle point to be. All points have the same probability of being the shuffle point.
There are two opposite factors that determine the Optimum Point.
The less cards you take into account (say only 1 deck) the bigger the probability that the shuffle point did not come up so the higher the probability that the Count you calculate is accurate. But the less usefull the information as you reduce the penetration (ie you play with pen of 1/6).
By increasing the cards to take into account you increase the pen but reduce its accuracy (more probable the shuffle point came up)
There is a point (ie number of cards to take into account) that maximises the potential impact of these 2 opposite factors.
Quote: BizzyBNot sure why you are so excited, as here the potentially more limited opportunity might be completely illusory.
The reason for the initial excitement was that the game being on screen means no personnel watching each move in theory. Hence you could make obscene bet spread decisions. I did point out earlier that I would not feel comfortable doing that because it would very easy to write a subroutine to catch counting.
Quote: BizzyBwhat kind of people are going to be trying to beat VBJ. The casino execs know what they are doing.
That's what they said about table blackjack...until card counters showed them they were wrong.
A hand is the individual set of cards (2-6) for the player or dealer. A round is a set of hands including the dealer's.
If given and paper, one could track the count by round as well as the number of cards dealt. Using that information we could tracking the moving window quite easily.
Quote: BizzyBAre you guys gonna write down every hand, go home, and try to figure out the shuffle point with your computer? How are you going to know where it is once you get back to the casino? Your only hope is that it is indeed when they switch dealers. When no one is playing, they change dealers. I have seen it. There goes that theory.
Further, as I noted over time based on the cards which come out and the estimate of the count before each round we could find which rounds our count seems effective and which it is no help. Through simple periodic auto-regression we could estimate the shuffle point by pen, paper, and brain.
Read the 5 caveats I listed. I did not throw them together haphazardly. Can you point to particular break in my analysis (If you want I am willing to re-summarize, but I think it should be understandable). It is possible some of my thought process is a bit more sophisticated than you are expecting.
Quote: AceTwoIf you read what I said is that I think that theoretically even with a computer you can not know when the shuffle point is. Not even attribute higher probabilities for specific points for the shuffle point to be. All points have the same probability of being the shuffle point.
If, as OP said they deal 4 decks then shuffle, not all points have the same probability of being the shuffle point. If all points had equal prob of being the shuffle point, sometimes it would shuffle after 1 round, sometime after the entire shoe. All cards have equal chance since they are assumed to be randomized each shuffle, but that is different that each point being the same.
However, when you walk up to the machine you no idea how close the state of the game is to the shuffle point. The most important problem for beat-ability (besides rules of the BJ game) is being able to estimate the shuffle point. It is not simple, but it can be found by pen and paper given time.
However, I think it is most likely they shuffle after every round and so no counting is possible (that's what most machines do).
Quote: endermikeIf, as OP said they deal 4 decks then shuffle, not all points have the same probability of being the shuffle point. If all points had equal prob of being the shuffle point, sometimes it would shuffle after 1 round, sometime after the entire shoe. All cards have equal chance since they are assumed to be randomized each shuffle, but that is different that each point being the same.
However, when you walk up to the machine you no idea how close the state of the game is to the shuffle point. The most important problem for beat-ability (besides rules of the BJ game) is being able to estimate the shuffle point. It is not simple, but it can be found by pen and paper given time.
You know that they shuffle after 4 decks. You do not know where that point is though.
Say you observe the last 10.000 cards that have been played (around 50 shoes).
What method, number, pattern etc would tell you where the shuffle point is?
My point is that the knoweledge that there is a shuffle after 4 decks (even though you can not know where it is) gives information that help.
It helps little (or very liitle) but it does help.
Like in roulette, each spin is independent from the previous.
In BJ, with shuffle after each hand (as in internet casino) each hand is independent from the past.
In BJ with a shoe, each hand is NOT independent of past cards appearing.
The same with the Virtual BJ (with no knoweledge of where the shuffle is), each hand is NOT independent of past cards appearing.
Quote: AceTwoYou know that they shuffle after 4 decks. You do not know where that point is though.
Say you observe the last 10.000 cards that have been played (around 50 shoes).
What method, number, pattern etc would tell you where the shuffle point is?
You don't need to know where the shuffle point is. You can treat the shuffle point as a random variable, just like you treat the next card out of the deck as a random variable until you see it.
Quote: endermikeIf, as OP said they deal 4 decks then shuffle, not all points have the same probability of being the shuffle point. If all points had equal prob of being the shuffle point, sometimes it would shuffle after 1 round, sometime after the entire shoe. All cards have equal chance since they are assumed to be randomized each shuffle, but that is different that each point being the same.
However, when you walk up to the machine you no idea how close the state of the game is to the shuffle point. The most important problem for beat-ability (besides rules of the BJ game) is being able to estimate the shuffle point. It is not simple, but it can be found by pen and paper given time.
However, I think it is most likely they shuffle after every round and so no counting is possible (that's what most machines do).
Wrong. All cards have an equal probability of being the shuffle point. The OP wrongly stated 1/312, this was revised to 1/208.
Wrong. You cannot find the shuffle point with a pen and paper, not in any reasonable time frame anyway.
Wrong. The game cannot lie about its procedure.
Please explain how you will determine the shuffle point in detail for those of us who are less sophisticated than you. If it is a secret, perhaps I can list 312 cards and you can tell me the shuffle point in my shoe. Or 312 cards times two.
I'm curious to know how long this would take, on average.
Finding the shuffle point will not be fast, but it will not take forever either (less than a day). All card counting is computed based on the long run. This is an overly demanding version of the long run but, once the point is found and track-able, the long run will take over and overcome the cost of finding it.
The biggest pitfall is how we assume the machine does the shuffle. Simply saying every 208 cards makes the math easiest, but that is probably not true. What is likely is it shuffles after the round where the shuffle point appears. Using that assumption is a double edged sword. On one hand it means the shuffle point is always moving. On the other it means more than 4 decks are dealt from the shoe. I think I will do the analysis on the 208 assumption and then the "start of the first round after the shuffle point."
Quote: AxiomOfChoiceYou could eliminate all windows outside of which you see 7 of the same card (rank & suit). For example if they shuffling after 4 decks (208 cards) and you have seen 7 aces of spades in the last 100 cards, you know that the shuffle point was within the last 100 cards (and not in the 108 before or after those 100). After going through several shuffles you may have narrowed it down to a small window.
I'm curious to know how long this would take, on average.
An extremely long time. You could go oh that's the 25th deuce I have seen, it has shuffled between X and Y points. I know how it could be done, no one giving me any earth shattering information. The only practical way is for a computer to calculate logarithms and other junk to find an increasingly likely shuffle point, which is not only pointless, it is cheating. I am also fairly sure there is a predetermined payout range that cannot be altered with counting; I have faith this is an amateur-counter trap.
Nevada Gaming Regulation 14.040(2) "Gaming devices that use a software random number generator (RNG) as part of a random number selection process to produce a predetermined set of outcomes must....2. Prevent the use of this information for the purposes of tracking deck composition and 'count' that would otherwise result in a violation of NRS 465.075."
After some time searching, it appears contrary to your belief, it is actually illegal to manufacter a beatable game. I just thought they would not do such a stupid thing, when in fact, it is not even allowed. In some other states, where there is no regulation, the house edge will reflect your average slot machine...20%.
Quote: endermikeThe biggest pitfall is how we assume the machine does the shuffle. Simply saying every 208 cards makes the math easiest, but that is probably not true. What is likely is it shuffles after the round where the shuffle point appears. Using that assumption is a double edged sword. On one hand it means the shuffle point is always moving. On the other it means more than 4 decks are dealt from the shoe. I think I will do the analysis on the 208 assumption and then the "start of the first round after the shuffle point."
I was thinking this too, but doesn't that actually make it easier to find? There are fewer choices for the shuffle point (since we are talking about rounds instead of cards) and once we identify the first round of a shuffle we know it forever (we simply keep a running tally of the number of cards played, and when that hits 208 we finish the round and the next hand is dealt from a newly shuffled shoe)
Quote: BizzyBI am also fairly sure there is a predetermined payout range that cannot be altered with counting; I have faith this is an amateur-counter trap.
As was pointed out before this is illegal in Las Vegas. If the machine shows cards being dealt or dice being rolled it has to follow the exact same probability distribution as those things would naturally. A machine can't capriciously deal a bust card to a player.
Quote: BizzyBThe only practical way is for a computer to calculate logarithms and other junk to find an increasingly likely shuffle point
Huh? What are you trying to calculate the logarithm of?
This would easily be done with a pencil and paper. You just keep 52 short lists of numbers, one for each card (rank & suit). The only question is how long it would take.
I seriously doubt that this is worth the time (the gains would probably be too small) but you never know. As for a practical matter, heat at machines tends to be VERY low, but a pencil and paper might be the thing that gets you noticed. Also you would have to play pretty slowly while finding the shuffle point, and this might be a problem if other people want to play as well (they are muti-seat machines, after all)
I still wonder how much of an advantage you could get by treating the shuffle point as a random variable.
Quote: BizzyBAn extremely long time. You could go oh that's the 25th deuce I have seen, it has shuffled between X and Y points.
Read the post and understand. We are fragmenting the count down. Not just 13 denominations, but 52 distinctive cards.
Quote: BizzyBThe only practical way is for a computer to calculate logarithms and other junk to find an increasingly likely shuffle point
False, just false. the phrase "calculate logarithms and other junk" demonstrates a lack of understanding of what is being proposed. A firm grasp of mathematics is crucial to any card counting disscussion.
Quote: BizzyBI am also fairly sure there is a predetermined payout range that cannot be altered with counting; I have faith this is an amateur-counter trap.
Hard to say with full confidence that this quote is nonsense, but it would violate everything we know about of video blackjack and video poker are designed. Regardless of how you feel about the validity of our methods, this type of argument undercuts the credibility of anything else you say.
Quote: endermikeHard to say with full confidence that this quote is nonsense, but it would violate everything we know about of video blackjack and video poker are designed. Regardless of how you feel about the validity of our methods, this type of argument undercuts the credibility of anything else you say.
In Nevada this (having set payouts rather than cards drawn randomly) would be illegal.
I'm not sure if some states have VLT-style video blackjack machines, though. But if they did, the phrase "shuffles after 4 decks" would be meaningless.
Quote: TwirdmanAs was pointed out before this is illegal in Las Vegas. If the machine shows cards being dealt or dice being rolled it has to follow the exact same probability distribution as those things would naturally. A machine can't capriciously deal a bust card to a player.
LOL ok you go count then. I give up. You are obviously more skilled and knowledgeable than I am.
Quote: AxiomOfChoiceIn Nevada this (having set payouts rather than cards drawn randomly) would be illegal.
I'm not sure if some states have VLT-style video blackjack machines, though. But if they did, the phrase "shuffles after 4 decks" would be meaningless.
Some states do. Nevada does not, but they might as well.
Quote: endermikeRead the post and understand. We are fragmenting the count down. Not just 13 denominations, but 52 distinctive cards.
False, just false. the phrase "calculate logarithms and other junk" demonstrates a lack of understanding of what is being proposed. A firm grasp of mathematics is crucial to any card counting disscussion.
Hard to say with full confidence that this quote is nonsense, but it would violate everything we know about of video blackjack and video poker are designed. Regardless of how you feel about the validity of our methods, this type of argument undercuts the credibility of anything else you say.
LOL I've had enough. Useless amateur with ploppy theories, talking about vague advanced mathematics that he can't describe and how he will beat an unbeatable game, telling me I'm talking nonsense. Go lose your money, I hope you do. It would violate everything you know because you know so little.
Nevada Gaming Regulation 14.040(2) "Gaming devices that use a software random number generator (RNG) as part of a random number selection process to produce a predetermined set of outcomes must....2. Prevent the use of this information for the purposes of tracking deck composition and 'count' that would otherwise result in a violation of NRS 465.075."
Quote: AxiomOfChoiceHuh? What are you trying to calculate the logarithm of?
This would easily be done with a pencil and paper. You just keep 52 short lists of numbers, one for each card (rank & suit). The only question is how long it would take.
I seriously doubt that this is worth the time (the gains would probably be too small) but you never know. As for a practical matter, heat at machines tends to be VERY low, but a pencil and paper might be the thing that gets you noticed. Also you would have to play pretty slowly while finding the shuffle point, and this might be a problem if other people want to play as well (they are muti-seat machines, after all)
I still wonder how much of an advantage you could get by treating the shuffle point as a random variable.
I'm saying a computer would use math operations to find a likely shuffle point based on input you feed it without needing as much info. It would take a long time to find it on paper. It is NOT worth the time, it is ridiculous proposition. And yeah, it would be extremely bizarre for someone to writing stuff down at a bj machine for hours on end. It WOULD get noticed, what they would do...who knows. You know other players wanting to play would berate you and complain about the escaped mental patient counting cards on a machine. Someone at a local place thought he was counting cards at one of these things...two people came to my table talkin about it, and he was a 10 minute joke.
