Cheating Online Casino? (Data Inside)

Quote: Wizard

I am open to it. In what format is the data? Can you copy and paste about ten hands here so we know what we're dealing with.

It came as a txt file, but I've imported it all into excel. I tried pasting it here, but the formatting was ugly and wouldn't be helpful. If you send me a PM of your email address, I can share the file w/ you.

John

Are we still keeping the name of this casino secret?

Quote: AxelWolf

Are we still keeping the name of this casino secret?

Yes. I think until/unless someone besides myself can corroborate that the game is gaffed, there's no sense naming anyone or getting egg on my face.

Also, if the game IS gaffed, step #1 for me would be to privately negotiate a refund/settlement.

Quote: Johnboy85

Also, if the game IS gaffed, step #1 for me would be to privately negotiate a refund/settlement.

Good thinking, some good old blackmail might do the trick.

Quote: AxelWolf

Good thinking, some good old blackmail might do the trick.

Not sure if sarcasm? What would you do?

Here is some additional information on the rules:

1. Is double after a split allowed? - Yes.
2. Can the player re-split? If so, to how many hands? - Yes to five hands
3. Same question as above but on re-splitting aces. - No resplit on Aces.
4. What happens if the player doubles or splits and the dealer gets a blackjack? In other words, does the player lose his original bet only or everything? - Early Surrender and Dealer Peeks, so N/A to this game.

Six decks and dealer hits a soft 17 were already stated.

Using my house edge calculator, before considering the early surrender rule, I get a house edge of 0.61873%.

From my list of rule variations, the value of early surrender against an ace is 0.39% and against a 10 is 0.24%. So, the overall house edge is 0.62% - 0.39% - 0.24% = -0.01%.

Hello!

I estimated that the possible range might be:



And if we had a history of cards that were released, then it's possible to establish the honesty of random events with a high probability ...

Quote: Dobrij

Hello!

I estimated that the possible range might be:



And if we had a history of cards that were released, then it's possible to establish the honesty of random events with a high probability ...

Hi, thanks for doing this!

Questions:
Which rules and variables did you use to sim this?
The lines represent 1,2, and 3 SD positive and negative, and blue is EV?
Y axis is dollars, and X axis is # of hands?

Finally, Wizard has the hand history in his possession and I assume will be doing a basic analysis. I've asked him to keep the results of that analysis private for now, but I will happy to update this thread once we know more.

My calculations are average and may not be accurate ...

I think if you analyze the concentration of hands and dealer cards, you can find the answer, or maybe not:)

Here are some initial rough results. First, let me explain the way the log files are organized. The amount bet column is the amount after any double, if any. Also, if the player split, there is a separate line entry for each hand. So, the ratio of the win to the initial bet can only be:

0, for a loss
0.5, for a surrender
1, for a win (including after doubling)
1.5, for a blackjack

For this reason, the usual standard deviation for blackjack of around 1.14 cannot be used. The standard deviation based on the way the log file is formatted would be much less. For now, I'm going to make an educated guess at a variance of 0.8, which equates to a standard deviation of 0.8944. The reason it is so low is there are no 2x wins and lots of surrendering, due to the early surrender rule.

Total bet	$8,629,233.50
Total win	$(44,847.30)
Expected win	$862.92
Win compared to expectations	$(45,710.22)
Variance	$538,878,465.80
Standard deviation	$23,213.76
Number std. dev	(1.97)
Probability	0.0244707637
Inverse	40.87

So, we see the player's loss due to bad luck is $45,710 compared to his small expected win, based on a player edge of 0.01%. A standard deviation on all this play (based on my rough estimate) is $23,214. So, the player is down 1.97 standard deviations. The probability of luck that bad or worse in a fair game is 2.45%, or 1 in 41.

Unless there is something I have wrong about the rules or am reading the file wrong, this looks like just moderate bad luck to me.

Quote: Wizard

Unless there is something I have wrong about the rules or am reading the file wrong, this looks like just moderate bad luck to me.

It seems that it's oftentimes just moderate bad luck online, yet it's rarely ever moderate good luck.

A smart online casino would gaff their games just enough to leave doubt.
Rarely is anyone really going to play long enough and at the levels it takes to find out.

Quote: Johnboy85

Not sure if sarcasm? What would you do?

Seems like it's a moot point for now.

It seems strange they are offering a game like this. If it's not gaffed and it's a mistake they should be noticing an overall lower than normal hold on the game. They don't need to offer a game like this to attract players, ploppys will play anything just the same. The people who are attracted to this game because of the generous rules are probably playing halfway decent. So why have this game available unless they are making an unusual amount of money on it?

It was not really sarcasm. If I had proof an online casino was gaffed. First I would look at some other games and see if there was any possible way to use that in my favor. For example, I would see if they had multiplayer games like craps, baccarat, roulette.

If there was no way to take advantage of the situation, I would try to negotiate something with them.

It seems to me that they have plenty of scapegoats so they can just blame whoever and move on.

The Betsoft incident comes to mind.

Quote: Wizard

Here are some initial rough results. First, let me explain the way the log files are organized. The amount bet column is the amount after any double, if any. Also, if the player split, there is a separate line entry for each hand. So, the ratio of the win to the initial bet can only be:

0, for a loss
0.5, for a surrender
1, for a win (including after doubling)
1.5, for a blackjack

For this reason, the usual standard deviation for blackjack of around 1.14 cannot be used. The standard deviation based on the way the log file is formatted would be much less. For now, I'm going to make an educated guess at a variance of 0.8, which equates to a standard deviation of 0.8944. The reason it is so low is there are no 2x wins and lots of surrendering, due to the early surrender rule.

Total bet	$8,629,233.50
Total win	$(44,847.30)
Expected win	$862.92
Win compared to expectations	$(45,710.22)
Variance	$538,878,465.80
Standard deviation	$23,213.76
Number std. dev	(1.97)
Probability	0.0244707637
Inverse	40.87

So, we see the player's loss due to bad luck is $45,710 compared to his small expected win, based on a player edge of 0.01%. A standard deviation on all this play (based on my rough estimate) is $23,214. So, the player is down 1.97 standard deviations. The probability of luck that bad or worse in a fair game is 2.45%, or 1 in 41.

Unless there is something I have wrong about the rules or am reading the file wrong, this looks like just moderate bad luck to me.

First, I want to publicly thank the Wizard for taking the time to answer my questions and evaluate the file I sent him.

I have PM'd him separately as I do have some questions about his interpretation of the log file. Of course, I submit he is way more intelligent on these matters than I, and therefore I certainly may be wrong. I'm not sure how much it will change the overall assessment of game fairness - we shall see.

Quote: AxelWolf

Seems like it's a moot point for now.

It seems strange they are offering a game like this. If it's not gaffed and it's a mistake they should be noticing an overall lower than normal hold on the game. They don't need to offer a game like this to attract players, ploppys will play anything just the same. The people who are attracted to this game because of the generous rules are probably playing halfway decent. So why have this game available unless they are making an unusual amount of money on it?

It was not really sarcasm. If I had proof an online casino was gaffed. First I would look at some other games and see if there was any possible way to use that in my favor. For example, I would see if they had multiplayer games like craps, baccarat, roulette.

If there was no way to take advantage of the situation, I would try to negotiate something with them.

It seems to me that they have plenty of scapegoats so they can just blame whoever and move on.

The Betsoft incident comes to mind.

I'm not familiar with Betsoft - I'll have to look it up.

But I agree with you. I still think the game is gaffed because the Wizard has confirmed a player edge given the effective rules in place. To that end, I really have no reason to quit playing the game since it's in my favor longterm. That being said, I could also take the route of alerting them to their error and trying to negotiate some kind of settlement for my silence, but then they may say thanks and not pay me anything...kind of a weird situation to be in. And it's not like I can cry foul because the game is gaffed against the players, it's actually in our favor, so the casino could probably turn it into a positive if anything ever became publicly of it and not have to worry about paying out upset customers. Weird situation to be in for sure...

John

I misunderstood who the file was organized and based on a PM Johnboy sent, am amending my table. This may not be the only amendment. I previously misunderstood how splitting and playing two hands at once was handled. I am also adjusting for the fact that Johnboy said he played two hands at once most of the time, something I previously didn't factor into the standard deviation. Playing two hands at once increases standard deviation by 45.82%.

Total bet	$5,096,758.00
Total win	$(44,847.30)
Expected win	$509.68
Win compared to expectations	$(45,356.98)
Variance	$264,539,581.65
Standard deviation	$16,264.67
Number std. dev	(2.79)
Probability	0.0026461624
Inverse	377.91

So I'm now showing 1 in 378 bad luck, but I would still consider this a rough guess, at best. I still may not get how this file is organized completely.

Quote: Wizard

Here is some additional information on the rules:

1. Is double after a split allowed? - Yes.
2. Can the player re-split? If so, to how many hands? - Yes to five hands
3. Same question as above but on re-splitting aces. - No resplit on Aces.
4. What happens if the player doubles or splits and the dealer gets a blackjack? In other words, does the player lose his original bet only or everything? - Early Surrender and Dealer Peeks, so N/A to this game.

Six decks and dealer hits a soft 17 were already stated.

Using my house edge calculator, before considering the early surrender rule, I get a house edge of 0.61873%.

From my list of rule variations, the value of early surrender against an ace is 0.39% and against a 10 is 0.24%. So, the overall house edge is 0.62% - 0.39% - 0.24% = -0.01%.

Wizard,

Was hoping you could maybe shed some light on why your house edge calculation results in player edge of 0.01%, whereas other calculators, including the one found on the qfit website returns 0.1% player edge for the same rules input. The difference would be fairly substantial in a case like this...

John