Could Real-Time Card Tracking and Adjusted Basic Strategy Beat Online Blackjack?
I believe that since traditional high-low card counting strategy already gives the player an advantage when the count is high, having precise information about every card dealt could allow for even greater benefits. Could such a system be effectively used against online casinos, where players aren’t kicked out for not betting regularly? Would this approach be viable in practice, considering the potential limitations online casinos might have?
Furthermore, with precise knowledge of the remaining cards, could the expected value (EV) of side bets become positive? Would this open additional opportunities to exploit the game, or are there practical limitations that make this unfeasible?
Quote: FatihhIf we had a program or an Excel spreadsheet that tracks every card dealt in a blackjack game, calculates the house edge in real-time based on the remaining cards, and suggests when to play, could this strategy be profitable? Additionally, the program would adjust the "basic strategy" based on the exact composition of the remaining deck.
I believe that since traditional high-low card counting strategy already gives the player an advantage when the count is high, having precise information about every card dealt could allow for even greater benefits. Could such a system be effectively used against online casinos, where players aren’t kicked out for not betting regularly? Would this approach be viable in practice, considering the potential limitations online casinos might have?
Furthermore, with precise knowledge of the remaining cards, could the expected value (EV) of side bets become positive? Would this open additional opportunities to exploit the game, or are there practical limitations that make this unfeasible?
link to original post
I did it. exactly as described. except I also made it auto play unsupervised. Kelly betting when I had edge >0.5% Min betting when no edge and wonging out when house had edge. It was 6 deck.
Program spent most of its time switching between tables or sat out. Had to join table and place the odd min bet to stop getting booted.
It was a fun exercise, but not viable. After about 3 months, I got singled out and rebuked for card counting, which they told me was not allowed. They didnt seem to notice the programmatic play.
There are easier ways to make money.
Out of curiosity, what was the highest edge you ever observed during your experiment? Do you remember?
I don't have much experience with state regulated online blackjack so I haven't any clue what their procedures are, and if it's viable in those locations.
Well known offshore places have been caught cheating(search Google, and its been talked about here)
I wouldn't automatically give up on the idea, there's probably money to be made via offshore casinos.
Not all the casinos cheat all the time, but some of the casinos cheat some of the time. You would need to factor that in somehow along with getting no paid.
I believe you would have to bounce around and keep your wagering to moderate levels. And don't forget , oftentimes the same companies offer their services to multiple online casinos thus limiting your bounce around ability.
The two decisions that provide the most equity are (obviously) the Insurance bet and '16 v T" because they are relatively close decisions and occur relatively frequently. You don't really need a computer to play those very close to perfectly.
Most decisions that are close will only occur rarely and even more rarely with a card composition that changes the decision, and then the average advantages gained by changing the decisions are usually quite small.
At any reasonable level of penetration set by the casino on 6-8 decks, many decisions essentially never change and provide no equity at all.
My general impression is that flat betting with computer-perfect strategy on shoes with 6-8 decks with 75% penetration is not profitable.
There are some combinations of penetration and number of players where flat betting computer-perfect strategy on single deck (and possibly double deck) games might be profitable, but good luck finding such games.
Quote: FatihhI was imagining a manual program where the user inputs the cards, but your approach is far more advanced. You must have excellent programming skills to create a program that can read cards, automate gameplay, and even switch tables.
Out of curiosity, what was the highest edge you ever observed during your experiment? Do you remember?
link to original post
It started as you envisage with me entering card values as seen and being prompted when to bet. Typically just a couple of bets per hour at up to about 2% edge. Incredibly tedious. So then I devised a sort of OCR reader that read the card values and flagged when to bet. Next step was to automate betting. But that was hard because I kept getting booted for non playing. spent 90% of time wonged out or playing the odd cover bet.
Best advantage was about 2%, but once in a blue moon..
Made less than £10 over a couple of months before they closed me down. It was fun, but not worth the effort.
Quote: Fatihh<snip>Furthermore, with precise knowledge of the remaining cards, could the expected value (EV) of side bets become positive? Would this open additional opportunities to exploit the game, or are there practical limitations that make this unfeasible?
link to original post
Fatihh,
Welcome to the forum!
As others have already posted, with the weak penetration typically available in online games, computer-perfect play will only rarely overcome the house edge in BJ.
However, targeting side bets seems a lot more feasible. The EV of side bets varies quickly with varying deck composition, and most SB EVs are very easy to compute for depleted shoes almost instantaneously using, say, Excel.
I have not tried this myself, so I cannot state with certainty that the frequency and value of this approach would be financially worthwhile, but I suspect it might be.
Dog Hand
Quote: FatihhIf we had a program or an Excel spreadsheet that tracks every card dealt in a blackjack game, calculates the house edge in real-time based on the remaining cards, and suggests when to play, could this strategy be profitable? Additionally, the program would adjust the "basic strategy" based on the exact composition of the remaining deck.
I believe that since traditional high-low card counting strategy already gives the player an advantage when the count is high, having precise information about every card dealt could allow for even greater benefits. Could such a system be effectively used against online casinos, where players aren’t kicked out for not betting regularly? Would this approach be viable in practice, considering the potential limitations online casinos might have?
Furthermore, with precise knowledge of the remaining cards, could the expected value (EV) of side bets become positive? Would this open additional opportunities to exploit the game, or are there practical limitations that make this unfeasible?
link to original post
Count using hi-lo and you will see your computer optimized betting is almost identical. Mostly the bet recommendation will be negative.
Computers were only useful for counting is very deeply dealt sd games.
Quote: DogHandQuote: Fatihh<snip>Furthermore, with precise knowledge of the remaining cards, could the expected value (EV) of side bets become positive? Would this open additional opportunities to exploit the game, or are there practical limitations that make this unfeasible?
link to original post
Fatihh,
Welcome to the forum!
As others have already posted, with the weak penetration typically available in online games, computer-perfect play will only rarely overcome the house edge in BJ.
However, targeting side bets seems a lot more feasible. The EV of side bets varies quickly with varying deck composition, and most SB EVs are very easy to compute for depleted shoes almost instantaneously using, say, Excel.
I have not tried this myself, so I cannot state with certainty that the frequency and value of this approach would be financially worthwhile, but I suspect it might be.
Dog Hand
link to original post
I have done exactly that, and have written an awful lot of code to do exactly these things, and very efficiently.
The pen in online BJ is awful and while sidebets can often be played very efficiently, the ones they tend to use online are the poker hand ones, where the EOR are relative rather than absolute and those are dreadful without really good pen.
Now there is another game with a lot of sidebets dealt from a shoe, and that is where all the money used to be. Incredible advantages, I once found. But obviously I am not the only one, and they really came down hard on that and while some advantage plays are still there, they're typically not worth it for the time and work required.
Something that stuck with me from a Grosjean book was a description of a beatable game, and his observation that if you have the sophistication to beat this, you have the sophistication to beat things much better. So this code I have written has been used in other ways, that are probably not welcome to be advanced on this site. Suffice it to say it involved electrodes on a sensitive area. (My own. I would never ask someone else to take that role.) But even that required a very special situation to be profitable enough to pay the other people involved and still be better than what I can do with my own mind, no other-than-monetary risk, and no wincing before 12% of hands. So it'll probably never happen again- I can do better just walking into a casino, after appropriate preparation, and counting.
Quote: AutomaticMonkey
Something that stuck with me from a Grosjean book was a description of a beatable game, and his observation that if you have the sophistication to beat this, you have the sophistication to beat things much better.
To be honest I was never terribly impressed with his game selection. There seemed to be an overt focus on next-card/hole-card opportunities in carny games especially which require lots of scouting time, are often very slow, have high variance and often require cover unless you want to spend a long time in court. You can have a 10% edge in games like that but when you break it down they are often less profitable than garden variety counting.
Conversely you make a lot of money online with tiny edges if you can scale up automation. Or even if you can't-there's a reason remote working has become so popular. I just mention this because ap literature in general is full of people trying to wipe their arse with their feet being applauded by post-graduate math nerds.
Wizard has done an online blackjack calculator that gives a correct decision for every deck composition at every decision point. So, I hope someone can combine this online calculator with their own simulation skills to find a reasonable EV number.
Quote: acesideGrosjean book has been cited a couple of times above. I’d like to have a copy of it, but I’m not sure if Grosjean has researched this subject of computer-perfect play for blackjack or not.
Wizard has done an online blackjack calculator that gives a correct decision for every deck composition at every decision point. So, I hope someone can combine this online calculator with their own simulation skills to find a reasonable EV number.
link to original post
Isn't Grosjean a member of this forum? I am pretty sure I remember reading some posts from him in the past.
Quote: acesideSo, I hope someone can combine this online calculator with their own simulation skills to find a reasonable EV number.
link to original post
There is no practical difference between optimized computer-perfect play and the use of a level 1 count system in normal online blackjack games.
FYI I'm selling Beyond Counting: Exploiting Casino Blackjack to Video Poker - James Grosjean if anyone is interested. Offer me a fair discounted price.Quote: DRichQuote: acesideGrosjean book has been cited a couple of times above. I’d like to have a copy of it, but I’m not sure if Grosjean has researched this subject of computer-perfect play for blackjack or not.
Wizard has done an online blackjack calculator that gives a correct decision for every deck composition at every decision point. So, I hope someone can combine this online calculator with their own simulation skills to find a reasonable EV number.
link to original post
Isn't Grosjean a member of this forum? I am pretty sure I remember reading some posts from him in the past.
link to original post
Quote: acesideI did this comparison myself and found a few differences between HiLo and Computer for the hand,16vsT. I put it in a note somewhere but can’t find it now.
link to original post
Yeah but that's a very marginal decision. The composition of the hand can actually change it to stand. The gains aren't worth much.
Even counting a conventional real-world shoe game you are looking at maybe gaining 0.1% over a level 1 system using a computer.
I haven't calculated the gains when the extremely shallow penetration is taken into consideration but I doubt it is more than a few hundredths of a %.
The main difficulties I had were:
1) That the resolution from the live casino cameras sometimes deteriorated for a minute or so, which meant that even with a large screen I was squinting and guessing what cards I'd seen dealt to the other players.
2) Online live casino blackjack is dealt so slowly I was losing the will to live, and to get any value out of the tool I'd built I had to pay close attention for hours at a time.
I would be tempted to revisit building a shuffle tracker now that I have developed the relevant skills for making such things on sheets/excel, but I have so many other things on which definitely make me money that I doubt I'll find the time or energy - unless we're all locked down again.
Quote: blackjacklad
2) Online live casino blackjack is dealt so slowly I was losing the will to live, and to get any value out of the tool I'd built I had to pay close attention for hours at a time.
There is nothing to prevent you from playing multiple tables. They don't even have to be at the same casino. If speed isn't the issue you would definitely want to to be doing this.
Playing an extra table should increase your win rate by 100% all things being equal. The gains from computer-perfect strategy on a single table are a tiny fraction of playing two tables with hi-lo or other system.
Quote: Archvaldor1Quote: acesideI did this comparison myself and found a few differences between HiLo and Computer for the hand,16vsT. I put it in a note somewhere but can’t find it now.
link to original post
Yeah but that's a very marginal decision. The composition of the hand can actually change it to stand. The gains aren't worth much.
Even counting a conventional real-world shoe game you are looking at maybe gaining 0.1% over a level 1 system using a computer.
I haven't calculated the gains when the extremely shallow penetration is taken into consideration but I doubt it is more than a few hundredths of a %.
link to original post
But we would like to see a numerical value of computer-perfect strategy on the hand of 16 vs. Ten.
Is it possible you simulate a 6-deck shoe with a 75% penetration? First, using Hi-Lo for both betting and playing; second, using HiLo for betting but computer-perfect for playing every incidence of 16 vs. Ten.
Quote: blackjackladIf I had automated software keeping the count on each table I could multi-table. Counting manually more than 1 table would be tricky.
link to original post
Practice, practice, practice! I regularly count 3 at a time (in a real casino not online). It helps to modify your counts, to give up a percent or two of EV but at the same time facilitate multiple sidecounts and multiple tables. You'll catch it, and enjoy it.
It only gets really difficult when a bettable count strikes at multiple tables at the same time. Then it becomes a matter of physical dexterity to get all your bets down in time and on the right bets. Makes you look like a nut too. (Fortunately I look like a nut all the time so it's not a tell.)
Feel free to reach out if you have such skills.
Quote: acesideQuote: Archvaldor1Quote: acesideI did this comparison myself and found a few differences between HiLo and Computer for the hand,16vsT. I put it in a note somewhere but can’t find it now.
link to original post
Yeah but that's a very marginal decision. The composition of the hand can actually change it to stand. The gains aren't worth much.
Even counting a conventional real-world shoe game you are looking at maybe gaining 0.1% over a level 1 system using a computer.
I haven't calculated the gains when the extremely shallow penetration is taken into consideration but I doubt it is more than a few hundredths of a %.
link to original post
But we would like to see a numerical value of computer-perfect strategy on the hand of 16 vs. Ten.
Is it possible you simulate a 6-deck shoe with a 75% penetration? First, using Hi-Lo for both betting and playing; second, using HiLo for betting but computer-perfect for playing every incidence of 16 vs. Ten.
link to original post
16 vs. 10 is not a useful play for a card counting AP, in most modern real-world games.
The reason why counts like High-Low perform poorly on that play is the 5 and the 6 are counted the same. One gives you a bust and one gives you a very strong winner on 16 vs. 10.
But if you are counting, especially an online game, or a backcount game where you are not betting anything at all without an advantage, you are nowhere near the point where you would consider hitting 16 vs. 10 when you have any money on the table. The relative value of plays for flat bettors, like those that can be derived from Schlesinger's BJA3, are not the same as they are for a counter. For a counter 15 vs. 10 is a much more valuable play, and he can always stand on 16 vs. 10 and it won't make a significant difference.
Quote: blackjackladIf I had automated software keeping the count on each table I could multi-table. Counting manually more than 1 table would be tricky.
link to original post
Most of the time you are simply closing the table if the count doesn't go positive immediately.
Counting a slow game especially with whatever digital aids you need (notepad, clickers whatever) is very easy. I don't know how someone with the intellect to do what you described you did could somehow struggle with counting multiple tables.
Note mult-tabling is very common at online poker. I'm not sure why card counters never consider it.
Quote: AutomaticMonkey
But if you are counting, especially an online game, or a backcount game where you are not betting anything at all without an advantage, you are nowhere near the point where you would consider hitting 16 vs. 10 when you have any money on the table. The relative value of plays for flat bettors, like those that can be derived from Schlesinger's BJA3, are not the same as they are for a counter. For a counter 15 vs. 10 is a much more valuable play, and he can always stand on 16 vs. 10 and it won't make a significant difference.
link to original post
It looks like you are saying “wonging in” is an important part of online blackjack. I doubt that.
For math purposes, as well as for all practical purposes, AP players must play all. By all means, the 16 vs. T play is the most important strategy of blackjack games.
Quote: aceside
For math purposes, as well as for all practical purposes, AP players must play all.
The more you play when the count does not indicate an advantage the more you will lose.
There is no mathematical purpose to playing all.
Quote: acesideSo, you are saying mid-shoe entry is allowed in online blackjack games. But, is the shuffle done by a dealer or by a machine. That should make a difference too.
link to original post
Many of the games I've seen have a second dealer shuffle.
A new deck is brought to the table.
The discards are taken to a table in the background, but still in view of the camera.
The cards are shuffled reasonably thoroughly, usually.
Quote: blackjackladIt was too arduous for me to watch more than one table because I needed to record not just which cards had been dealt, but also the exact order they had gone into the discard pile. If I didn't know what order they had been put into the pile then there was no way for me to find out whether their shuffle was imperfect enough to make use of. Combine that with the video feeds occasionally switching to low resolution and it was a tedious ball-ache just to monitor one table accurately.
link to original post
Ah I see yes I can imagine with shuffle analysis that would be problematic if not completely insurmountable.
I did some work on this a while back and generally speaking the initial card order is preserved to some extent through the shuffle - there are various reasons as to why which I'm guessing you probably know.
Quote: acesideQuote: AutomaticMonkey
But if you are counting, especially an online game, or a backcount game where you are not betting anything at all without an advantage, you are nowhere near the point where you would consider hitting 16 vs. 10 when you have any money on the table. The relative value of plays for flat bettors, like those that can be derived from Schlesinger's BJA3, are not the same as they are for a counter. For a counter 15 vs. 10 is a much more valuable play, and he can always stand on 16 vs. 10 and it won't make a significant difference.
link to original post
It looks like you are saying “wonging in” is an important part of online blackjack. I doubt that.
For math purposes, as well as for all practical purposes, AP players must play all. By all means, the 16 vs. T play is the most important strategy of blackjack games.
link to original post
Exactly the opposite. The whole point of playing online is that you are not playing -EV hands, something that would be too much of a tell if you were doing it at a physical table. It's related to the reason why sidebets can be profitable; you never have to put anything on a sidebet.
Now it's true that on some software you have to occasionally place a bet to keep the video feed going, but that's a minimum bet once every 5-10 hands and that won't break the bank. Just with the weak rules and the awful pen of those games you will rarely see an advantage over 1%, and for the same reasons the play deviations are also rare. So a Counter's Basic Strategy optimized for a small +EV count will do just fine. Your leverage from this kind of play comes from being able to play any time, watching multiple tables, and not having any travel expenses or time overhead. If I was cripple I'd probably be doing stuff like this.
That's probably going to get noticed. You're better off setting up some shill accounts for viewing purposes. I think you would want multiple viewing shill accounts you could switch around and mix things up a bit so they can't detect the same accounts at the tables all the time. I would go as far as playing only one account at a time occasionally % of making it harder for them to connect the dots. I haven't any clue what kind of sophisticated detection systems they have, but I wouldn't want to make it simple for them.Quote: AutomaticMonkey
Now it's true that on some software you have to occasionally place a bet to keep the video feed going, but that's a minimum bet once every 5-10 hands and that won't break the bank.
