No you aren't, roulette is not an RNG, it is a PRNG. If you take into account the initial marble velocity and rate of rotation of the wheel, you can seriously reduce the probability space of where the ball will land. Further, this can be modelled statistically pretty simply.
To win at betting with roulette you only need to bet on a set of numbers, where the probability of the marble landing on that set is pretty good, and the payout of winning over time covers the losses from being wrong.
So say you can accurately guess which 1/4 of the wheel will capture the marble. (and there are no greens... the real math is the same but uglier a bit). You put down equal bets on the 9 numbers represented by that. Each beting round will net you a 4:1 payout.
Now say you can only guess that 1/4 of the wheel, 50% of the time. You will still average a 2:1 payout.
Guessing the 1/4 of the board 50% of the time is not that hard, you just need to calculate the linear drift of the time the marble passes a fixed point, and know the frequency of "loops" that is too slow for the marble to stay on the track. Accounting for the rotation of the board is straight-forward, it is almost constant for a given play. The 50% covers the wierd bounces. Just calculate board position for the time that the marble will get below the threshold, and the marble position as well, and you've won. A person can't do it well, but an arduino is overkill for it.
Edit: further notes on this- casinos will be able to detect this easily. All they have to do is look for people on hot streaks that are "too hot", then compare betting placement with wheel positions. And they generally kick people out who are winning too much anyway, even if they can't prove 'chaeting'.
And this is why shoe computers are banned in casinos!
Google Glass will make this hack so simple, the casino's will have to ban them from being anywhere close to a roulette table to reduce team cheating possibilities.
This seems highly unlikely to be practical, given chaos. If this system shows high sensitivity to initial conditions, then we probably can't know the various inputs to enough precision to get any results.
Consider that a 2 link pendulum is chaotic. The tiniest difference in your estimate of the initial kinematics results in divergent predictions.
Some folks published a paper last year or the year before about their roulette monitoring system. You can improve your odds because you're allowed to place bets while the ball is bouncing. Some physics prof claims to have successfully used a device to count the bounces and decide which half of the wheel to bet on. I'd find the citations for you, but I'm sure you could do that just as fast as I.
One thing to keep in mind, always in such situations[1], you don't have to be right, you just have to be probabilistically right -- that is the goals is to just increase the chances of being right in such a way as to have an edge.
[1] Where you are betting, or otherwise working on odds and long term averages.
1) In the age of the smart phones, could one not develop an app to do this calculation for you? It might be tricky to film the wheel, but I'm sure that can be done stealthily somehow.
2) Would this work on online roulette games who claim to simulate the real-world physics of a roulette table?
To win at betting with roulette you only need to bet on a set of numbers, where the probability of the marble landing on that set is pretty good, and the payout of winning over time covers the losses from being wrong.
So say you can accurately guess which 1/4 of the wheel will capture the marble. (and there are no greens... the real math is the same but uglier a bit). You put down equal bets on the 9 numbers represented by that. Each beting round will net you a 4:1 payout.
Now say you can only guess that 1/4 of the wheel, 50% of the time. You will still average a 2:1 payout.
Guessing the 1/4 of the board 50% of the time is not that hard, you just need to calculate the linear drift of the time the marble passes a fixed point, and know the frequency of "loops" that is too slow for the marble to stay on the track. Accounting for the rotation of the board is straight-forward, it is almost constant for a given play. The 50% covers the wierd bounces. Just calculate board position for the time that the marble will get below the threshold, and the marble position as well, and you've won. A person can't do it well, but an arduino is overkill for it.
Edit: further notes on this- casinos will be able to detect this easily. All they have to do is look for people on hot streaks that are "too hot", then compare betting placement with wheel positions. And they generally kick people out who are winning too much anyway, even if they can't prove 'chaeting'.