Test for unfair Blackjack
Prelim results showed a major distortion between wins in lower bets and higher bets overall. Any suggestions/easy formulas for a novice to use to test these things. Would appreciate it.
I was thinking of testing bust percentage on both sides, for certain bet levels, overall 2 card total for all hands, likelihood of busting of dealer on 12-16 and seeing how many he busted on smaller and larger bets, etc etc.
Any other suggestions would truly be appreciated as if I find anything upsetting I'll pursue it basically.
Very common online.Quote: dfde50
Prelim results showed a major distortion between wins in lower bets and higher bets overall. Any suggestions/easy formulas for a novice to use to test these things. Would appreciate it.
I don't think anyone has been willing to invest the time or money and do it right.
I would love to see what would happen if 50 different people picked a casino and everyone bet table min for 5 hands and then suddenly bet table max. I think it would be very eye opening.
so I wanted to go hand to hand and just tabulate probabilities but in the end I'm not a mathmetician and/or statistician and don't know which test I can do that would give me conclusive results like a 1 out of xx probability of this happening for higher bets vs lower OR just overall.
my only guess is to do things like calculate overall bust rates and the odds of dealer hitting hands on certain first two cards on both lower bets and higher bets, then contrast and so forth.
You're betting $1500 a hand online for real money?Quote: dfde50just to give you an example. 1162 wins 1223 losses or so in the $5 to $99 area. 333 wins 368 losses in the $100-$199 area. then when I look at $500, $1000 and $1500 bets for example I get something like 130 wins and 200 losses. Its hard to do the analysis since I'm taking net wins or losses and I haven't went hand to hand yet, but the prelim results of net losses and wins show like a 1-2% win rate by the dealer for a lot of hands in the low levels, but when you just take all high bets as a sample, with granted less hands played, the win percentage is shockingly low.
so I wanted to go hand to hand and just tabulate probabilities but in the end I'm not a mathmetician and/or statistician and don't know which test I can do that would give me conclusive results like a 1 out of xx probability of this happening for higher bets vs lower OR just overall.
my only guess is to do things like calculate overall bust rates and the odds of dealer hitting hands on certain first two cards on both lower bets and higher bets, then contrast and so forth.
That's Brave.
Does the casino provide a hand log? That's going to be important.
I suggest you record your session with a desktop video recorder or something.
Mind naming the casino?
Quote: AxelWolfVery common online.
I don't think anyone has been willing to invest the time or money and do it right.
I would love to see what would happen if 50 different people picked a casino and everyone bet table min for 5 hands and then suddenly bet table max. I think it would be very eye opening.
I got heavy into online bonus-whoring about 2001. At that time they still were giving away the store. But I was having fits with the blackjack game on Microgaming software. And they were supposed to be THE reputable software. Fifteen years later and I can still hear the haunting sound of the dealer's voice, "Dealer has twenty, dealer has twenty-one, dealer has twenty". I swear, I didn't want to believe it because it was such a big company but I was convinced it was rigged. STILL, the promos were so generous I made plenty of money.
It wasn't long that the casinos started increasing wagering requirements and excluding blackjack from match-plays. I still thought it was rigged and insisted they should REQUIRE you to play their stinking blackjack. lol
Those were the good old days.
Quote: AxelWolfYou're betting $1500 a hand online for real money?
That's Brave.
Does the casino provide a hand log? That's going to be important.
I suggest you record your session with a desktop video recorder or something.
Mind naming the casino?
I'm trying to get a hand log. I'll post if I find anything out in the future. Rather not name the casino yet but its not some uncommon casino.
The patterns are just so unusual
Even in bovada back in the good bonus days I ran 500 deposit bonus to 20k plus on their Aztecs theme slots, then switched to blackjack after the bonus was done and lost hand after hand of large bet-
Good luck with that. Online casinos that don't have visible logs do that for a reason.Quote: dfde50I'm trying to get a hand log. I'll post if I find anything out in the future. Rather not name the casino yet but its not some uncommon casino.
Quote: AxelWolfGood luck with that. Online casinos that don't have visible logs do that for a reason.
Worst case I'll throw an action out there and force them into discovery if I have to. Doubt it will come to that, but I surely will if I have to, they can pay tens of thousands that way lol.
Either way, something doesn't feel kosher. Push levels feel elevated, all level bets felt harder to win then normal, but most of all all my high level bets are lost far far more than my lower level bets. Its shocking when I did the analysis, although maybe I'm wrong and I hope I am.
I'd like to be able to put a number on it like "1 out of 50,000" kind of thing, but since its dealing with actual odds of lower sized bets versus higher sized bets, Its not as simple as simply same betting 10,000 hands, know what I mean?
I want to go hand to hand, card to card and compile totals for dealer busts, player busts, how many dealer totals were reached (ie 18, 19, 20), blackjack count for player and dealer, totals of first two cards for dealer and player, and so on.. And basically I want to categorize THOSE totals additionally in totals when bets were high and low.
Of course, I have no idea how to find a "1 out of 50,000" type of determination but I feel like at least a few of those totals will be helpful and maybe can produce an exact odds as well.
Will update if I find out anything.
Quote: WizardofnothingI'm. It a conspiracy theorist but I play this one casino out of boredom on flight, I play onlily one slot machine , it's 21 cents a pull , in over 42k hands I have never hit 5 of a kind in any of the top four pay symbols, There are some other befits to the site which make anything besides the casino profitable and I'm only down about 900 on the casino over 3 years however I'm 1000000 percent sure it's rigged, one day I went into the casino with 5 dollars and ran it up on 21 cent pulls to over 2900 only to go to blackjack and lose 9 100 hands in a row, switched to pai gow and lost 6 250 hands in a row -
The patterns are just so unusual
Even in bovada back in the good bonus days I ran 500 deposit bonus to 20k plus on their Aztecs theme slots, then switched to blackjack after the bonus was done and lost hand after hand of large bet-
Believe me, I know the feeling but I also know extreme bad luck can happen, within reason of course. But not all bad luck imo is a result of bad luck. I think the very nature of online programs which can be altered secretly, subtly and so on keeps us all paranoid, warranted or not. And the allure and certain win nature of the casino tweaking odds even 2-3% more in their favor would technically be cheating and highly hard to prove, know what I mean? The very nature of online gaming is exciting but scary without rigorous and relentless regulation and surprise "checks", and even then, you cant watch things 24/7.
Not sure that's where he will start, but I'm just waiting on marching orders ;-)
Quote: AxelWolfGood luck with that. Online casinos that don't have visible logs do that for a reason.
So I thought you might appreciate my analysis after getting records. This is a "licensed" government casino we are talking about by the way as strict as it gets, lol.
For all bets $500 and over only as a random high bet sample (approximately 10-15% of total play excludes all bets $499 and below). Playing basic strategy 100% of the time.
February :
Total Deals (1 dealer hand and up to 5 player hands per deal): 215 Total Dealer Hands: 215 Total Player Hands: 434
Player blackjacks : 30
Dealer blackjacks: 11
# of player hands busted: 93
# of dealer hands busted: 43
$500 hand results: 196 wins 289 losses 49 pushes
$750 hand results: (from blackjacks): 30 wins
$1000 hand results: 5 wins 4 losses 0 pushes
March:
Total Deals (1 dealer hand and up to 5 player hands per deal): 145 Total Dealer Hands: 145 Total Player Hands: 232
Player blackjacks: 21
Dealer blackjacks: 10
# of player hands busted: 51
# of dealer hands busted: 28
$500 hand results: 121 wins 182 losses 29 pushes
$750 hand results: (from blackjacks): 20 wins
$1000 hand results: 5 wins 1 loss 0 pushes
April:
Total Deals (1 dealer hand and up to 5 player hands per deal): 98 Total Dealer Hands 98 Total Player Hands: 227
Player blackjacks: 11
Dealer blackjacks: 2
# of player hands busted: 45
# of dealer hands busted: 21
$500 hand results: 80 wins 122 losses 25 pushes
$750 hand results: (from blackjacks): 11 wins
$1000 hand results: 4 wins 3 losses 1 push
In the end, if you combine the blackjack hand wins plus the normal hand wins, roughly 1/4th or 1/3rd of all hands $500+ are lost more than won over a sample like this. Food for thought. While it would be super easy to test it statistically if I just flat bet $500 and 1 hand per deal etc, then compiled data, I did not play that way. I played 2 hands per deal usually but luckily I flat bet usually each hand.
Quote: WizardofnothingSo what did you record if you busted and the dealer didn't play out the hand?
Those types of hands were very small maybe a dozen or two since most were 2-5 hands played so very rarely did all hands bust out. I think just based on recollection maybe under a dozen. That said, I just recorded how many times player busted and dealer busted respectively. If all player hands busted out of a deal, I recorded those player busts but did not record a dealer bust obviously.
That said, its pretty striking in the sense that I did a full blown analysis of the net changes in value a few months ago as I did not have the records, and for 3x more bets under $50, the win/loss ratio was .493 wins and .507 losses for all net changes in value under $50.
The problem areas were bets $500 and above, that's why I focused on them, bets $400 and below mainly showed a reasonable win and loss ratio. Bets $500 and above showed huge skewed results that's why I dove deeper here.
Quote: WizardofnothingAnd this is a live or online casino?
This is a online casino
Quote: WizardofnothingIs this in Atlantic city?
can't go into name/location atm. (check ur mail).
Quote: dfde50
$1000 hand results: 5 wins 4 losses 0 pushes
$1000 hand results: 5 wins 1 loss 0 pushes
$1000 hand results: 4 wins 3 losses 1 push
So you say with $1000 hands the player tended to win 14 to 8
Now. Here's a question: Were they YOUR hands or the hands of another anonymous player? You do realise that the other players may not be real? Also were those hands resulting from doubles and splits? If so they had a higher inherant probability of winning.
Sadly your stats are way short of being statistically significant.
Remind me. Though this is online, is it Live dealer or RNG game. If it's live dealer with a shoe, I'd suggest that card counting each shoe as an observer would be a far far cheaper and more effective way of testing the game. If the place is cheating and it's live dealer, the likely way to cheat would be to take a few 10 value cards out of the shoe.
In truth, this is very unlikely to be casino cheating. They don't need to as they almost certainly have a rule along the lines of "If we consider that you are not playing in good faith in any way, then we reserve the right to close your account and keep all the balance that is in there" That is the easy and legal way that many of these places do 'cheat'
BULLSH*T.Quote: OnceDear
In truth, this is very unlikely to be casino cheating. They don't need to as they almost certainly have a rule along the lines of "If we consider that you are not playing in good faith in any way, then we reserve the right to close your account and keep all the balance that is in there" That is the easy and legal way that many of these places do 'cheat'
NVM it's regulated.
If it's some non regulated online casino I put the odds they have cheating software at 85%
That doesn't mean you can't obtain an Advantage or win, that just means you probably shouldn't expect to play straight up on something like Craps, roulette, BJ thinking you are giving up the standard HA.
Honestly if you're not trying to AP.... what's the difference? You're probably going to lose the same amount either way. It's just going to take longer.
I was recently at a location playing a good promo. Someone mentions they don't trust the machines because they have class II. My argument was.... that doesn't mean you cannot win AND it's still +EV. I actually did well on the class II machine, hell, it was better than the class III. The machines still have a minimum pay back, the outcome is just determined differently. It's only going to affect someone like Rob Singer who believes they have some wackjob system(SO IT REALLY WON'T AFFECT THEM EITHER). Obviously you can't play a class II VP machine or slot and use the standard math calculations to determine your advantage or disadvantage. However if you have a promotion that's a 35% edge on a normal class III VP machine I think you would be crazy not to play it on a class II machine if that was your only option. Same thing goes for online.
System bettors shouldn't treat unregulated or class II machines the same way as something random.
Quote: AxelWolfBULLSH*T.
NVM it's regulated.
If it's some non regulated online casino I put the odds they have cheating software at 85%
System bettors shouldn't treat unregulated or class II machines the same way as something random.
Thanks Axel.
No-one should be playing unregulated or loosely regulated online casinos and especially their slots or rng blackjack games. Betting thousands there would be the height of stupidity. I understand that regulated online games are easier to find in Europe because of your stringent money transfer and state laws. Did the OP ever identify the class of game he was looking at? There's a RNG roulette machine in one of my B&M casinos and it sucks bigtime. Anyone expecting a fair game from it would be bitterly disappointed.
I'd said
I confess that I hadn't really considered that the OP was playing an unregulated casino, where I'd put the probability of cheating at near 100%. If it's a regulated one and he's playing an RNG game of blackjack, I'd suspect maybe a 25%-50% probability of cheating simply because it's so easy for them to cheat.Quote:In truth, this is very unlikely to be casino cheating.
If it's online live dealer game at a regulated place, then I'd strongly doubt cheating because at all the ones I used, they spread the cards at the start of every shift and there's pretty much 24/7 visibility ( to someone) of the table. At the ones I use, the dealers don't have sleeves and there is a right palaver if there is any dealer error. Besides, for a regulated casino live dealer game, isn't it more likely that the OP was just unlucky. Didn't his stats show players winning with $1000 hands? ( I have to admit speed reading his stats and may have misinterpreted)
My assertion is still that for regulated live dealer online blackjack he would be more likely to have been unlucky than to have been cheated.
He'd have to be nuts to play thousands through on an unregulated or even regulated RNG game, especially if he already suspected cheating.
Observing the hands of others at those tables would also be pointless, because those other players could simply be simulated by the casino and simulated as big winners too for that matter, because there would be no money leaving the casino, just a back room bot or stooge.
The OP asserts that the outcomes look pretty normal for lower level bets but are apparently a bit all over the place with $400 and up wagers. But the sample distribution is such that there are more, and thus more representative, smaller bets and fewer and thus less representative higher value bets. We all know fewer bets analysed will be likely to be a bit 'all over the place' Thus his sample and his evidence is far, far too weak. His hypothesis is not supported by evidence and he his a self confessed novice whose evidence gathering process is not very scientific.
I do agree with you that someone should do a proper in depth analysis of the RNG games offered by the big name online casinos, especially as you are not alone in suspecting that they gaff the game for larger stake players. But how to do that without throwing lots of money in their direction? That's the question.
Quote: OnceDearSo you say with $1000 hands the player tended to win 14 to 8
Now. Here's a question: Were they YOUR hands or the hands of another anonymous player? You do realise that the other players may not be real? Also were those hands resulting from doubles and splits? If so they had a higher inherant probability of winning.
Sadly your stats are way short of being statistically significant.
Remind me. Though this is online, is it Live dealer or RNG game. If it's live dealer with a shoe, I'd suggest that card counting each shoe as an observer would be a far far cheaper and more effective way of testing the game. If the place is cheating and it's live dealer, the likely way to cheat would be to take a few 10 value cards out of the shoe.
In truth, this is very unlikely to be casino cheating. They don't need to as they almost certainly have a rule along the lines of "If we consider that you are not playing in good faith in any way, then we reserve the right to close your account and keep all the balance that is in there" That is the easy and legal way that many of these places do 'cheat'
The max limit per hand is $500, so the $1000 wins (or losses) are double downs. The Split hands are included in the "$500" win/loss as wins or losses respectively, as new hands. It is a online casino as I already stated.
Personally, I think any argument that states that its highly unlikely a casino would cheat because they put terms in that lend credibility is a kind of ridiculous argument. If you check CasinoBar, they probably had terms in which you describe as well.
As for statistically insignificant, just on face value, all of my hands of $25-$300 or so had very small win loss spreads as I attested in May. But the $500+ hands were wildly deviated. I find that alone remarkable that high hands could literally deviate so wildly and nearly all low hands just didn't lol
Quote: OnceDearThanks Axel.
No-one should be playing unregulated or loosely regulated online casinos and especially their slots or rng blackjack games. Betting thousands there would be the height of stupidity. I understand that regulated online games are easier to find in Europe because of your stringent money transfer and state laws. Did the OP ever identify the class of game he was looking at? There's a RNG roulette machine in one of my B&M casinos and it sucks bigtime. Anyone expecting a fair game from it would be bitterly disappointed.
I'd saidI confess that I hadn't really considered that the OP was playing an unregulated casino, where I'd put the probability of cheating at near 100%. If it's a regulated one and he's playing an RNG game of blackjack, I'd suspect maybe a 25%-50% probability of cheating simply because it's so easy for them to cheat.Quote:In truth, this is very unlikely to be casino cheating.
If it's online live dealer game at a regulated place, then I'd strongly doubt cheating because at all the ones I used, they spread the cards at the start of every shift and there's pretty much 24/7 visibility ( to someone) of the table. At the ones I use, the dealers don't have sleeves and there is a right palaver if there is any dealer error. Besides, for a regulated casino live dealer game, isn't it more likely that the OP was just unlucky. Didn't his stats show players winning with $1000 hands? ( I have to admit speed reading his stats and may have misinterpreted)
My assertion is still that for regulated live dealer online blackjack he would be more likely to have been unlucky than to have been cheated.
He'd have to be nuts to play thousands through on an unregulated or even regulated RNG game, especially if he already suspected cheating.
Observing the hands of others at those tables would also be pointless, because those other players could simply be simulated by the casino and simulated as big winners too for that matter, because there would be no money leaving the casino, just a back room bot or stooge.
The OP asserts that the outcomes look pretty normal for lower level bets but are apparently a bit all over the place with $400 and up wagers. But the sample distribution is such that there are more, and thus more representative, smaller bets and fewer and thus less representative higher value bets. We all know fewer bets analysed will be likely to be a bit 'all over the place' Thus his sample and is evidence is far, far too weak. Hi hypothesis is not supported by evidence and he his a self confessed novice whose evidence gathering process is not very scientific.
I do agree with you that someone should do a proper in depth analysis of the RNG games offered by the big name online casinos, especially as you are not alone in suspecting that they gaff the game for larger stake players. But how to do that without throwing lots of money in their direction? That's the question.
Initially I had to measure my "net value changes" since I didn't have the actual hands. When I did that in May as kind of a very casual substitute, I found very close wins/losses under like $300 or $400 something like that.
Going hand to hand and looking at the data doesn't change much for bets $500+. I still see a huge amount of losses over wins. During my play much of this was consecutive, in huge streaks, where the dealer would miraculously hit 19, 20 or 21 on huge table values. I have line charts which show the streaks in 8 scenarios.
This is not an unregulated casino, I wouldn't play this level of money on some "shark casino" or something like that. This is supposed to be a government regulated casino online.
Quote: dfde50Initially I had to measure my "net value changes" since I didn't have the actual hands. When I did that in May as kind of a very casual substitute, I found very close wins/losses under like $300 or $400 something like that.
Going hand to hand and looking at the data doesn't change much for bets $500+. I still see a huge amount of losses over wins. During my play much of this was consecutive, in huge streaks, where the dealer would miraculously hit 19, 20 or 21 on huge table values. I have line charts which show the streaks in 8 scenarios.
This is not an unregulated casino, I wouldn't play this level of money on some "shark casino" or something like that. This is supposed to be a government regulated casino online.
You still never said: Live dealer or Random Number Generator games? You mention other players at the table which implies live dealer?
Bring up their rules page and see what house edge they claim. If it's a rng game, you could reasonably divulge to us the software supplier. You could pm me and I'd keep it private. You could also trust Axel, who I understand has his own concerns about online BJ games.
If, in spite of suspecting them of cheating you, you decide to throw money their way while gathering evidence, you could get some free software to video capture your play. You need LOTS of totally detailed evidence.
What you've collected so far is not laid out in an obvious form for analysis and raises more questions than answers. What for instance is a $750 game?
Do you confirm that your stats show $1000 hands as predominantly winning for the player? ( Just as they should)
Are you really counting the hands of other players in your analysis?
What I was emphasising was that they don't need the games to be gaffed in order to rob you. Their own terms make that easy enough for them to do without resorting to gaffed games. Do you have experience of getting paid out promptly when you won?Quote: dfde50Personally, I think any argument that states that its highly unlikely a casino would cheat because they put terms in that lend credibility is a kind of ridiculous argument.
I find that you are willing to throw $500 per hand at a game which you believe to be gaffed pretty remarkable and LOL too.Quote: dfde50But the $500+ hands were wildly deviated. I find that alone remarkable that high hands could literally deviate so wildly and nearly all low hands just didn't lol
Quote: OnceDearYou still never said: Live dealer or Random Number Generator games? You mention other players at the table which implies live dealer?
Bring up their rules page and see what house edge they claim. If it's a rng game, you could reasonably divulge to us the software supplier. You could pm me and I'd keep it private. You could also trust Axel, who I understand has his own concerns about online BJ games.
If, in spite of suspecting them of cheating you, you decide to throw money their way while gathering evidence, you could get some free software to video capture your play. You need LOTS of totally detailed evidence.
What you've collected so far is not laid out in an obvious form for analysis and raises more questions than answers. What for instance is a $750 game?
Do you confirm that your stats show $1000 hands as predominantly winning for the player? ( Just as they should)
Are you really counting the hands of other players in your analysis?
the house edge is .46%. I have not played since I had suspicions (obviously). it is not live dealer, why would I analyze hands of a live dealer, that would make no sense realistically.
The $750 results are blackjacks wins entirely. The $500 is the table max, thus, $1000's are double downs, $750 is the win from a blackjack hands, and $500 are splits and normal hands won or lost. While I am aware that maybe splits have better odds, I treated each split hand (of which there weren't that many) as another player hand for simplicity.
As stated, since $1000 hands are double downs, are less than two dozen hands total or so, I didn't feel a need to pluck that out and discuss it.
There are no hands from "other players", as I've stated, these are all hands I've played personally and have the records.
I'm not going to discuss the rng or casino due to several concerns, least of which are I see no benefit in doing so If either I'm wrong or some people will just not believe even if there was evidence in front of them.
Quote: OnceDearWhat I was emphasising was that they don't need the games to be gaffed in order to rob you. Their own terms make that easy enough for them to do without resorting to gaffed games. Do you have experience of getting paid out promptly when you won?
I find that you are willing to throw $500 per hand at a game which you believe to be gaffed pretty remarkable and LOL too.
That's another silly statement. Yes, by nature the casino can win, but this isn't about absolutes, its about degree and values. Mcdonald's might charge $5 for a burger and maybe make $2, but if they charge $6 they can make $3. Obviously the incentive is to make MORE. Considering the struggling industry, conflicts of interest, I'm surprised you'd think by nature casinos aren't tweaked at all, since there's every incentive to do so, and probably extremely lax "once a year" style audits in a pre-arranged account, which can't even measure small lots of hands, since programming can presumably be programmed any way you like, such as programming the outcomes of lots of hands and/or small sets of hands different ways.
As for throwing $500 per hand at a game I believe to be gaffed, Obviously one cannot play 3 hands then think its gaffed, one must come to that conclusion over the course of play, and act accordingly. I still did not say it's gaffed, I just said that there's a ton of red flags which I think given immense conflicts of interests are highly suspicious to say the least. After all the software creator alone just doubled revenue, is tied to "royalties" from results, somewhat manages or controls the software overseas AND on top of all that the entire industry is wildly struggling. Call me crazy to think there's a huge incentive to tweak odds somewhat, maybe from 49.5% return to 40% for higher level bets. Does degree/magnitutde invalidate cheating if its a small tweak instead of a large one? lol
Quote: dfde50I want to run tests on 10,000 hands of blackjack to test the fairness.
Do you have records of those 10,000 hands already available? If so, to what level of detail?
I suggest if you do have some detail over that size of sample, then present it as a raw spreadsheet. there are plenty here who would surely be happy to analyse it for fun.
If you don't have detail, then there's not much analysis to be done.
You apparently suspect the game to be gaffed at least to some extent because you had a few bad hands where the wager was significant and because you see cheating as a viable and likely business model, because they can. That's not making a strong case.
There was a time when bricks and mortar casinos very probably cheated vastly because they could and because their mobster owners thought they could get away with it. What's to stop such mischief being prevalent today? Rumours of cheating would possibly be enough to get your online casino shut down or reputationally damaged. They'd be somewhat stupid to get so greedy as to get CAUGHT cheating. Far easier to just run crappy games with high house edge, rubbish paytables in small print rules openly published.
By the way, if you were throwing down two or three simultaneous hands at $500, then your $500 hand probabilities would not be representative because of the co-variance which they would have.
Quote: OnceDearDo you have records of those 10,000 hands already available? If so, to what level of detail?
I suggest if you do have some detail over that size of sample, then present it as a raw spreadsheet. there are plenty here who would surely be happy to analyse it for fun.
If you don't have detail, then there's not much analysis to be done.
It's in the form of screenshots but it contains all details, and the number of hands is hard to ascertain. There are 4,616 "deals" of which there may be 2-5 hands on each deal, but from recollection 2 hands were played the majority (70%+), so I estimate 10,000 to 12,000 hands.
Going hand to hand and inputting the details in a spreadsheet would likely take a really really long time. Just doing the math, assuming I could do 30 deals a day out of 4,616 deals total, that would mean it would take around 153 days.
Even then, I would have no idea what to input as far as what a statistician would need, which is why I made this thread in May, to ask what I could pluck out of the entire results to "measure" and come up with a "1 out of xx" scenario.
For simplicity, and considering mostly only the higher bets had these problems in general, I just went hand to hand on $500 and above hands and just calculated the results. It was completely in line with my first analysis of the net changes in value for all deals, which showed the lower sized bets really didn't have any large spreads.
Besides, even if I did take 150 days and put it into a spreadsheet in a measurable fashion, what would it accomplish anyway? I doubt anything would change even if it does show a statistically improbable outcome.
Quote: OnceDearBy the way, if you were throwing down two or three simultaneous hands at $500, then your $500 hand probabilities would not be representative because of the co-variance which they would have.
Well barring an exact method for measuring "multihands" I had to compile them this way. Most of the hands were simultaneous, as in 2-5 hands versus a single dealer hand. I reflected accurate totals in the dealer hands played and player hands played but I understand there are factors within statistics that make this as a not entirely scientific representation.
Although, this isn't a court, and I don't need to really present absolute scientific statistic evidence. I already analyzed net value changes for everything, and I'm using the absolute same exact basis for everything and it still shows deviation in higher values. In other words, if you're using the basis of measuring the net value change for all deals $50 or under and you win .493 and lose .507, then you use the same exact basis and measure the net value changes $500+ and you win .320 and lose .680, then you go hand to hand and show basically the same fundamental structure, ie, smaller bets show break even results and larger bets do not, then I think you can allow a more casual interpretation in general.
In any event, the problem would be turning screenshots into spreadsheets, the time it would take, and what data to compile for what values and so on. That's the problem.
Quote: dfde50It's in the form of screenshots but it contains all details, and the number of hands is hard to ascertain. There are 4,616 "deals" .
I have the tools to turn your 4616 screenshots into a spreadsheet. Would take far less than 150 days. So long as the screen grabs are of good consistent quality.
Quote: OnceDearI have the tools to turn your 4616 screenshots into a spreadsheet. Would take far less than 150 days. So long as the screen grabs are of good consistent quality.
All of the screenshots are of generally perfect quality. They are .jpeg images with a size between 250-400KB each I'd say. Basically, 4 files (months) of 1.4GB collectively. Something like that. The screenshots are basically like a summary screenshot, not a screenshot of the actual table.
As to how I'd input something like that into a spreadsheet Is beyond me. .
Quote: dfde50All of the screenshots are of generally perfect quality. They are .jpeg images with a size between 250-400KB each I'd say. Basically, 4 files (months) of 1.4GB collectively. Something like that. The screenshots are basically like a summary screenshot, not a screenshot of the actual table.
As to how I'd input something like that into a spreadsheet Is beyond me. .
Make that my problem. Pop them into google drive and share them with me. I'll see what could be done. It's somewhere between crushingly easy and somewhat tricky.
Pointless of course unless it shows a readable display of what cards were laid and where and at what stake. Maybe just make two of those images available to me to assess what's involved.
Like you, I do question whether the results yielded would be worth the effort, but I'm reaching out to offer to help because it could be an interesting distraction project. Evidence trail does need to start somewhere or else it's going nowhere and we should forget it right now.
Unfortunately usually nothing is put up and no one shuts up.
Quote: dfde50Understood. Sent 2 check your mail. Or if you could tell me realistically how to convert these images to a spreadsheet, and If I'm capable of doing so, I would, but manually would be impossible mainly because of the inputs involved and the multihand aspect. Also keep in mind that the images on drive are smaller than if you open them in windows viewer they are 2x larger etc, if that makes a diff. Also keep in mind the 2 photos show hands where no splits/double downs occurred, but if they did occur, those are also marked on the bottom as "split" or "doubledown" with separated values next to the hand it was done to or what not.
Thanks Files received.
The amount of detail there is good and the structure of those two seems consistent enough. The font used and resolution is a bit crappy, but most likely machine readable. I'd initially assess the exercise of machine reading those files as quite do-able but needing a bit of time ( say one or two days ) of coding and 'sperimenting. I've done similar exercises before: This could be fun on my next day off work.
Here's the mechanism I'd probably employ:-
Install a command line OCR capable package I have one already that reads this kind of random image stuff.
Create an Excel VBA script that would cycle through the images and for each image, invoke the OCR library by shelling through a command line. THE OCR package would strip out the text into a simple but messy text file. Excel would then clean up and interpret the text file and stash the contents. Clean up would depend on dealing with multiple hand, splits etc.
That would rely on the OCR library being OK with that font. If it isn't, I have another way.
PM to tell me what process you adopted to get those screengrabs and describe any post processing you did to save them. Was it a laborious manual trawl through after the event from your game logs? If so, are those game logs still available to you online?
Quote: OnceDearThanks Files received.
The amount of detail there is good and the structure of those two seems consistent enough. The font used and resolution is a bit crappy, but most likely machine readable. I'd initially assess the exercise of machine reading those files as quite do-able but needing a bit of time ( say one or two days ) of coding and 'sperimenting. I've done similar exercises before: This could be fun on my next day off work.
Here's the mechanism I'd probably employ:-
Install a command line OCR capable package I have one already that reads this kind of random image stuff.
Create an Excel VBA script that would cycle through the images and for each image, invoke the OCR library by shelling through a command line. THE OCR package would strip out the text into a simple but messy text file. Excel would then clean up and interpret the text file and stash the contents. Clean up would depend on dealing with multiple hand, splits etc.
That would rely on the OCR library being OK with that font. If it isn't, I have another way.
PM to tell me what process you adopted to get those screengrabs and describe any post processing you did to save them. Was it a laborious manual trawl through after the event from your game logs? If so, are those game logs still available to you online?
The screengrabs (screenshots) were delivered from the software company itself (through a casino request that encompassed a lot of effort on my part), as generally the only way they could deliver such records they stated. As far as saving them like that, they just came in rar files segregated by the months, and inside had all the jpeg images as shown.
If you want all the records I can put them in google drive I don't mind.
Quote: dfde50If you want all the records I can put them in google drive I don't mind.
It will be a fun exercise. I can't promise when I'll look at this, or even whether my enthusiasm will persist. But yes. please share. Rar files will be fine. I won't share them or any info about them.
Quote: OnceDearIt will be a fun exercise. I can't promise when I'll look at this, or even whether my enthusiasm will persist. But yes. please share. Rar files will be fine. I won't share them or any info about them.
Okay check your mail
I understand that anyone can have bad luck, but mine seems to be pushing the limits of standard deviation.
I played 102 discrete sessions over the course of about 2 months and found the results to be ... less than optimal.
Not being a high roller, my bets were penny ante in comparison $2-$6, but still ... real money.
To get an idea,
Standard Multi Hand (3) blackjack (RNG) ... I always played 3 hands
(no gimmicks or side bets)
6 Decks, shuffle after each hand
Dealer hits soft 17
Double after split
Split on any first two cards
resplit to 3 hands
No hit on or resplit of aces
Late Surrender
BJ Pays 3-2
According to the wizards house edge calculator, playing basic strategy as published by the wizard (99.9% allowing for the occasional mis-click) the house edge should be 0.53904% or $5.39 for ever $1000 I put on the table.
Over the course of 33,835 hands I bet a total of $120,872 for an expected loss of $651.55 ... my actual loss was $2,409 ... this seems like some really really bad luck (or maybe I'm missing something)
I don't have screen shots, but do have a spread sheet which tracks some of what I thought were key items...
For each session I tracked:
* Total Hands Played
* Wins
* Losses
* Pushes
* Doubled Hands Won
* Doubled Hands Lost
* Player Blackjacks
* Surrenders
* Dealer Hands Dealt
* Dealer Hands Busted
* Dealer Hands Showing an Upcard of 4
* Dealer Hands Showing an Upcard of 4 Which Busted
* Dealer Hands Showing an Upcard of 5
* Dealer Hands Showing an Upcard of 5 Which Busted
* Dealer Hands Showing an Upcard of 6
* Dealer Hands Showing an Upcard of 6 Which Busted
After the 19th session (when I began to suspect something was amiss) I also tracked
* Dealer hands Drawn to 21
* Dealer Blackjacks
Any Interest? .. Thanks
PJ.
Quote: WizardofnothingSo you are down 4 x why you should be- I could be wrong but I do not think that it's that much of a statistical abnormality
With a house edge of .53904% the expected loss for 33835 hands is 182 betting units, the actual loss was 743 leaving the loss of 561 units to "bad luck". A standard deviation of 1.16135 puts the probability of that happening at approximately 0.34% or almost 300 to 1 ... certainly not out of the realm of possibility, but damned suspicious givin the hands were played during a hundred different sessions over a period of two months.
Quote: pjt36With a house edge of .53904% the expected loss for 33835 hands is 182 betting units, the actual loss was 743 leaving the loss of 561 units to "bad luck". A standard deviation of 1.16135 puts the probability of that happening at approximately 0.34% or almost 300 to 1 ... certainly not out of the realm of possibility, but damned suspicious givin the hands were played during a hundred different sessions over a period of two months.
I did like 4,700 hands or so, I remember that I "lost" like 350 when counting "net value losses" per deal, ie, if throwing 2 hands out there I ended up with a net loss that deal, I counted it as a loss. I know that dear eluded to multiple hands having a coefficient factor which skews probabilities, I think that's the hard part about checking these statistics is because when you play multihand it's terribly hard to figure out deviation if your a novice with regard to statistics.
The odd thing for mine wasn't so much the total losses no matter how you calculate the multihand aspect (although I don't really know my total "losses" since most were 2 hands at a time), but was the fact that I saw so many more losses for higher hands and so little losses over wins for lower hands. I mean, when I checked I found a 1% spread between wins and losses on bets in the $25-$50 area, but when I bet $500 or $1500 or $2500 over the deal, my losses were staggering over my wins. That alone is making me say like, well if the deals are all blind why would just higher level wagers lose so much. etc.
I don't want to be a conspiracy theorist or anything, but lets be realistic, online casinos are not rigorously audited, there is plenty of room for tweaking or other aspects, we've seen it before (ala Casinobar/air), the audits are presumably done preannounced, in a forewarned account, based on tons of hands and not just a few. I mean they could do a lot of things online that nobody could know about. I just don't know. And when you don't know firmly, you don't give money to people and "trust" them with your money lol. If my results are not deviated as my unscientific analysis seemed to show, maybe I'd play online again, I don't know.
For one I wouldn't gaff small bets to lose. I would even let people win small bets more often for a short period of time. Give them a taste of winning.
Quote: AxelWolfIf you had an online casino and wanted to gaff it how would you do it???
For one I wouldn't gaff small bets to lose. I would even let people win small bets more often for a short period of time. Give them a taste of winning.
Just off the top of my head?
Change payback when/before audited, then change it back.
Set limits one can win per day, forcing losses afterwards. Or manually adjust players in "loss" mode where he's already hooked on betting big. Most people bet bigger after losses to "win back" so that's an obvious thing.
Set limits on bets or outcomes for other eventualities.
Change payback percentage overall slightly so nobody picks up on it.
Make changes to high roller accounts utilizing the above.
Change payback depending on certain level bets.
Program "overall" winnings limit like a scratch off system, ie, only all collective people in a given time interval may win 99.5% back, ie, if someone wins big, you are forced to lose.
I mean, honestly, that's just off the top of my head. Software can presumably be tweaked, changed, to and from instantly. Heck, there are screenshots of softwares which show administration layouts which effectually allow for all of the above. I find it hard to believe billion dollar (or multi million dollar) casinos can't hire 3rd parties or in house engineers to design or tweak programs to accomplish any or all of the above, lol.
The question is, will they. Is there effective oversight. Is there something to lose. Frankly, as an unbiased person I have to say what's the worst that could happen? Fine of 5% of what they netted? Loss of trust maybe for a little while? IDK.
I ran 1000 trials of 33835 hands (using UK blackjack as I already had that set up and shuffling up after every hand). The average loss was -£161.16 and the SD was £216.40. Only 2 runs had a loss more than 743 (-822 and -767.50).Quote: pjt36......expected loss for 33835 hands is 182 betting units, the actual loss was 743....standard deviation of 1.16135....
Thus yes it is bad luck to have lost so much but not outside the realms of possibility,
... Also, There should also have been two runs where there were gains of more than 561 ...
Funny we so very very rarely hear about those kinds of wins in online blackjack :-)
