100% winnings on Playtech's roulette, just another scam ?

[tristan727]

You gotta give them credit for cleverly conning people anyways...my guess is that it's their own software posing as, or hijacking this fun mode? Nahhh...actually more likely it's getting the result back from the casino, and somehow delaying the graphical result in time for the punter to be told what it is. Only today I'd noticed my balance being updated at a Wagerworks casino, before the wheel had stopped spinning.

Please Alexandre...try to get this in your head - no-one who ever had really truly discovered the secrets to stock market, property or roulette etc. success would actually decide to tell the world and sell the secrets, as a)they'd be shooting themselves in the foot & b) who needs to hawk, peddle & spam when you've become a master of the universe? If anyone really knew how to predict roulette to the % you say they can in realplay - they would make a dozen fortunes in no time, before a single casino manager had even begun to take note of their account. And if they even had eventually found it difficult to wager - they'd just go visit their friends and relatives - not tell the likes of you or I.

 

[Kaynos72]

I just noticed how funny the 100% claim from that thread, go down to 97% only in fun mode or 60% in real mode, 80% if you pay ? lol come on.

All that to give red or black that have an almost 50/50 chances anyway ?

 

[gerilege]

Only in fun mode - a very likely explanation

I believe in the following explanation, because there was a similar issue in the good old days with a poker room.

(See this link for a more detailed technical explanation, probably you can start with chapter Shuffling a Virtual Deck of Cards, and read chapter Breaking the System on page 3:

You do not have permission to view link Log in or register now.

)

The fun mode uses some algorithm to generate the next number, which is different to the algorithm used in real mode. The fun mode algorithm is PC local, the real mode goes to the server.

The fun mode algorithm has a period (period is limited by the seed of the local PRNG), let's say 2^32. So it can only produce 2^32 different outcomes, which is then mapped to the 0..36 interval, and probably even not all 2^32 outcomes are used. With some "cracking" it might be possible to produce the complete list of numbers within the period. After that it is possible to do a pattern matching algorithm using the 10-element sample you provide on the whole sequence of numbers within the period. After finding the correct prefix via the sample it is simple to tell the next number in the period.

Once again, this was only a very brief description, see the article for more details.
And it only works for fun mode, because fun mode PRNG doesn't have so high security standards and period for practical purposes I guess. So I can imagine that they can predict fun mode, but I don't think that they will ever be able to predict real mode, because it should be a different algorithm.

The hard "cracking" step might have been done with or without insider help.