For all you math guys out there

[Clayman]

I sent them out to a former associate who has a program that analyzes statistical data for poker players but is easily apapted to other games of chance so I can't speak to the accuracy of these figures at this time.

Keep us posted. I hope his program works better than True Gambler's did.

And, I gather, people actually pay for OCA now. Amazing.

If you think I can help in any way, feel free to PM me.

[GrandMaster]

I read it as 1109 over ~35000 hands. Which isn't so bad. If it's over 1156 hands, as you say, they are cheating. No doubt at all. Do you know how many BJ's, 20's 21's over the 35000 hands? That is, I assume you got some on non-max-bets.

1109 over 35000 would be way too low, that's even less likely, if that were true then the game would be rigged in the player's favour. The only way these numbers make sense if 35000 is off by a factor of 10, and the number of hands is only 3500.

The probability of the dealer having 20, 21 or BJ 1109 times out of 3500 hands is tricky to calculate, becuase it is depends very strongly on the exact probability of 20, 21 or BJ, which in turn depends on the number of decks and whether the dealer hist soft 17 or not. Taking the probability of dealer's 20, 21 or BJ to be 0.3, the probability of the dealer having 20, 21 or BJ at least 1109 times out of 3500 is about 0.0144 or 1 in 69. If you change 0.3 to 0.305, the probability increases to 0.0618 or 1 in 16.

Bruno712, can you check the numbers again because they don't seem right to me.

[Clayman]

1109 over 35000 would be way too low,

Sorry - of course you're right. I wasn't too good at multiplying 35000 by 30%

Basically not very much makes very much sense.

I'm sure we'll hear more from Bruno when he gets back to the office next week. Hope so anyway.

Edited to ask

OK lets say there were 10,500 BJ's, 20's & 21's in 35000 hands (30%). Let's say he made 1156 max bets in those 35000 hands and that 1109 of those max bets were against a BJ, 20 or 21. Wouldn't that alone be strong evidence that bet size effects outcome? Even though the traditional units won/lost analysis for the 35000 hands might still appear to be OK? Would you think it likely a casino over a long period of time could get away with rigging only max bets in this fashion?

[Bruno712]

Clay,

You are using the 1109 figure as the total number of 20, 21 and black jacks the dealer received out of the 35,000 hands. That is not the total number of those hands. It is the number of those hands the dealer received when I maxed my bet ONLY. I do not have the total amount of ALL 20, 21 and BJs as of yet. I maxed my bet 1156 times. Out of those 1156 hands only, the dealer caught 20, 21 or BJ 1109 times. He, of course, caught those hands on my other wagers, but I haven't received that figure back yet. Another thing I may not have said (or I may have, I'm getting older by the minute here). I did not lose the entire 1109 hands where the dealer caught those hands. I pushed some and won very few.

I hope that clarifys things and gives you more insight about my concern.

Dead Tired In Massachusetts,

Bruno

[gamblinboi]

wow Bruno, at least tell us what software this was on?

[Bruno712]

Gambolini,

In 1973, to my shock, my name appeared in my hometown (a Boston suburb)newspaper as having been arrested for assualt with a dangerous weapon. Not a real good thing to happen to a pre-law student. The problem was that on the day in question, a Saturday in October, I was running for 115 yards against the U of Miami IN MIAMI!!!!!! Some idiot punk who I grew up with got arrested, presented no ID and used my name and the ignorant cops booked him without verifying his identity. Thus, the erroneous police blotter. Even after the newspapers retraction, I was questioned about and looked down upon for a crime I never commited. So you won't catch me casting aspersion at anyone until all the evidence is in and verified.

I was probably wrong to broach the subject before I had all the answers, but I did. Look at it like a serial, a cliff hanger, LOL. So tune in same Bat time, same Bat channel.

[GrandMaster]

Clay I maxed my bet 1156 times. Out of those 1156 hands only, the dealer caught 20, 21 or BJ 1109 times.

Thank you for confirming it. This is conclusive evidence that the software is rigged. You must have lost quite a lot of money. You should have stopped earlier, just 24 out of 25 times (probability 1 in 200000000000) would already make me very suspicious.

[Bruno712]

Grandmaster,

Indeed i did! From up $3k to down $3K at one casino. But I had just hit a Royal so I wasn't as concerned as I should have been. A real bad habit I have. I habitually break the rule that says "Any money sitting in my account is My money". I continue to justify my playing with "Their money". But I'm aware of it and I'm working on it. My analyst would be proud.

[Clayman]

Thank you for confirming it. This is conclusive evidence that the software is rigged. You must have lost quite a lot of money. You should have stopped earlier, just 24 out of 25 times (probability 1 in 200000000000) would already make me very suspicious.

Bruno - thanks for the clarification - I think I finally understood it right (see after my "edited to ask" above.)

GM - You are saying that if one randomly picked any 1156 hands from a 35000 hand sample, one would expect about 1156*.3=347 hands that would be a BJ 20 or 21 versus the 1109 Bruno is reporting. This still would be conclusive evidence of being rigged even if the entire 35000 sampled contained the expected number of BJ's 20's or 21's, say 9391 non-max-bet hands. Right? (9391+1109=35000*.3)

What I find interesting, if this is true, is that True Gambler's OCA software might never have found a rigging like this since he never took bet size into account, assuming it to be irrelevant. Correct? Apparently Bruno had a max bet about 3% of the time but I wonder what % of all bets made at a casino are max bets. I'd guess alot less than 1%. And I'd guess it might be a while before anyone would suspect this since typically I doubt if most bankrolls would survive the loss of much more than a few max bets and therefore it would be difficult to get a sample size. I doubt it is something PWC looks at when issuing those silly monthly payback reports. Is it? If they did, the W/L %age on max bets would stick out like an elephant at a dog show. What is your opinion on the feasibility of a casino rigging only max bets in this fashion? A little too obvious?

This whole question of "losing more with bigger bets" is something that has often been complained about over the years and something I have found difficult to disprove conclusively. If this is occurring with one of the major software providers, would it be an exaggeration to say something like this could change the entire landscape of internet gambling?

Bruno - you have the right attitude in saying you may have been wrong in even broaching this before you had all the facts. I think you are aware that you are dealing with TNT here. You also said earlier "I can't speak to the accuracy of these figures at this time". Has that now changed? Is your friend certain of this max-bet analysis? Rest assured, if this is confirmed, many questions will be asked as to his methodology. Will you be ultimately willing to have these logs subjected to other 3rd party analysis? Have you even verified them for completeness yet? I'm curious - were your logs presented to him in a text or database format or as picture files?

I certainly am staying tuned to this Bat-channel!

[case]

Bruno, would you mind providing some more details about the bets?

You mention that the preliminary analysis looked at the $100 and $400 bets. Is the 1109 of 1156 figure $400 bets only (and all such bets), or does it include $100 bets and others?

Is 1156 determined by a specific bet size (e.g. all bets of $100 or more), or by some other method?

What is the table limit?

Do you use a progression to determine your bet size?

Is the software provider one of the big ones, or a small operation?

TIA.