How to detect "unfair decks" in Holdem

In many phone-based Holdem apps, the decks are made unfair so the game is more exciting. The unfairness is "attractive" because there are more high hands, more ridiculous bets that pay off, making the game more addictive. That's just a fact of life...and it may apply to online decks as well.

Obviously, you can detect unfair decks by playing 10,000 hands and tracking the winning hands...but that's a bit time consuming.

A faster way would be along these lines:
* play 100 hands
* track how often you "could have won" if you had stayed in the hand all the way through
* compare (1) your total number of coulda-won hands and (2) the longest sequence of losing hands to a statistically valid norm

That last bullet is the tough one -- but it would be possible to generate a table for 5-handed and 10-handed games, showing the average, one-sigma, two-sigma norms for a truly fair deck. I have no idea *how* to generate it (other than tallying a few million hands), but such a table would be super useful in detecting unfair decks.

Does anything like this currently exist?

Why not just log the individual cards you receive and do an analysis on how often each appears?

Quote: FCBLComish

Why not just log the individual cards you receive and do an analysis on how often each appears?

Would take thousands of hands to be statistically significant, IMHO.

Quote: davodavo

In many phone-based Holdem apps, the decks are made unfair so the game is more exciting. The unfairness is "attractive" because there are more high hands, more ridiculous bets that pay off, making the game more addictive. That's just a fact of life...and it may apply to online decks as well.

Obviously, you can detect unfair decks by playing 10,000 hands and tracking the winning hands...but that's a bit time consuming.

A faster way would be along these lines:
* play 100 hands
* track how often you "could have won" if you had stayed in the hand all the way through
* compare (1) your total number of coulda-won hands and (2) the longest sequence of losing hands to a statistically valid norm

That last bullet is the tough one -- but it would be possible to generate a table for 5-handed and 10-handed games, showing the average, one-sigma, two-sigma norms for a truly fair deck. I have no idea *how* to generate it (other than tallying a few million hands), but such a table would be super useful in detecting unfair decks.

Does anything like this currently exist?

I don’t follow what (1) and (2) are doing. If I understand the “bias” you postulate, (1) should not be “unfair”. Each player is winning or losing the average number of times, it’s just that there are more “bad beats” than expected. More second nut vs nuts or more miracle river cards. Am I understanding the hypothesis correctly?

Quote: unJon

I don’t follow what (1) and (2) are doing. If I understand the “bias” you postulate, (1) should not be “unfair”. Each player is winning or losing the average number of times, it’s just that there are more “bad beats” than expected. More second nut vs nuts or more miracle river cards. Am I understanding the hypothesis correctly?

Hi unJon,
Sorry for the ambiguity in my original post. There are several things going on at once with these "juiced" decks, so there are a couple of flavors of the bias that I detect in some of the games that run on smartphones (e.g., PokerKing):
A. Good or even great hands happen too frequently.
B. Two great hands happen at the same time, encouraging huge bets (e.g., two full houses, or a full house and a flush, or even four of a kind and three of a kind)
C. Miracle river cards happen every couple of minutes.
D. Crazy opening bets and all-in bets appear to improve the odds for that better (lots of bad beats)
E. Folding often at the outset seems to yield more crummy cards.
F. If you win one hand, the starting cards for the next hand seem above average.
With that in mind, my original item 1 is to try to detect symptoms A, C, and E. Item 2 is try detecting all of them.

Let me try an example suspicion: if I bet very conservatively, I seem to get crap cards for 20 hands or more (I will almost always fold them unless the blinds have me in the game). Is that a statistically reasonable series? As I typically fold anything below a J7, I'm in at the flop maybe 20% of the time...I'm certainly no expert in the Poisson distribution, but to have 20 hands in sequence be below average opening cards doesn't seem reasonable. But maybe running a million hands in a Monte Carlo simulation would show how probable that event really is.

About unfair gameplay on Gambling Apps, I remember a guy saying he was getting constant really good hands on a Poker game in Practice free gambling app. He took a good look at what the practice Dealer was doing and he realized the practice Dealer wasn't dealing the cards the right way and he most likely would have lost his money had he went to a real world Casino using the practice app as a guide. On a hilarious side note, a gambling app had this disclaimer, "This Gambling app gives better payouts than a real world Casino. We are not legally responsible if players lose wagers in a real world Casino using this gambling app as a guide!" LMAO!

Quote: Nathan

About unfair gameplay on Gambling Apps, I remember a guy saying he was getting constant really good hands on a Poker game in Practice free gambling app. He took a good look at what the practice Dealer was doing and he realized the practice Dealer wasn't dealing the cards the right way and he most likely would have lost his money had he went to a real world Casino using the practice app as a guide. On a hilarious side note, a gambling app had this disclaimer, "This Gambling app gives better payouts than a real world Casino. We are not legally responsible if players lose wagers in a real world Casino using this gambling app as a guide!" LMAO!

Yessir.
People want to be fooled by randomness (read: Taleb) and in general have a very poor grasp of statistics and odds (read: innumeracy and Kahneman).
WHY people want to be fooled is beyond me, but they do.