That was just a particular roulette wheel. If the wheel is well balanced and isn’t rigged it will distribute the outcomes evenly. If not, that can be exploited.
Good casinos run statistical checks on their wheels often to make sure they meet a randomness criteria. The expected payout of the wheel, which of course is favirable to the house, is calculated assuming the wheel is an RNG.
See the show, he did it in every casino he tried. They had to to take the wheels apart and shuffle the parts every night to defeat him.
I thought the show was about a famous episode of which I have a vague recolection. Anyway I’m still very skeptical of such a sweeping statement. I stand by what I said before but without knowing more about this exploit I can’t comment any further.
Here is a link to the show. Scroll down, the episode is entitled The Roulette Assault. Be careful, there are other episodes about different ways of beating roulette. And don’t misunderstand, I share your skepticism. If you can debunk this somehow, I’d be grateful. But understand that the claim was not about an exceptional wheel somewhere. The claim is that he succeeded in many casinos.
I don’t see how that’s possible. For it to even be theoretically possible to break even at roulette, you don’t need just a slightly biased wheel, you need a wheel that’s biased by at least 3% (on european tables) or 5% (on american tables).
Such a bias would be glaringly obvious in even the most elementry statistical tests that casinos routinely perform. Forget roulette wheels, if this were true, the mathematical insight alone would be enough to have this guy living comfortable for the rest of his life if he published it.
You need to see the show. I don’t remember all of the details. His edge was slight, I seem to remember 8%. He wasn’t happy when he had to come to the USA because of the double-zero which ate into he edge. If the show was a fraud, it was an elaborate one. They interviewed a number of witnesses. shrug
It is incredible, but so is your claim. Do you have any proof for these statistical tests that you claim the casinos do? It’s a lot of work. He recorded thousands of spins on each wheel to find the bias. How often do the casinos spin the wheel, say, 10,000 times and run a chi-square test of the results? You say “routinely”…cite?
One needs to be careful about how one applies the definition of ‘random’. For example, the next spin of a wobbly badly-tuned roulette wheel will still have a random discrete result (i.e. you can’t predict with certainty that ‘12’ will result), even if the overall system has a preferential bias toward certain numbers.
From Here
If he did indeed have an 8% edge, added onto the 3% house take means that roulette wheels need to be routinely over 10% off random. Thats an absolutely huge bias that even a human observer would be able to detect. Furthermore, it’s stated that he was well known by casino managers for exploiting this flaw which means that there must be a 10% bias that cannot be overcome by better engineering. I find it absolutely staggering to believe that you couldn’t engineer a roulette wheel with less than a 10% bias and, frankly, an elaborate ruse is a far more likely explaination that a 10% bias.
Oh, and when it comes to gambling, you can get a hundred “eye witnesses” to fevrently believe any whacky scheme that you come up with that claims to make them money. It’s probably second only to high end speaker cables in it’s ability to fool gullible folk.
Every single time. The casinos watch every spin of every roulette wheel in the house, and do the statistical tests on all of them. This is not in the least bit difficult, with electronic surveillance. This is a much larger data set than this guy could possibly have collected, since he can only watch one wheel at a time, and he doesn’t spend every waking hour in the casino, and he doesn’t know how the wheels are swapped around each day. So if he could detect the pattern, then the casinos could, too.
On the other hand, consider the following scenario: A guy goes around claiming to be able to clock the wheels. The casinos very publicly make a show of kicking him out. He uses this as evidence that his scheme works, and loiters around outside the casino telling folks his system, for a fee. These schmucks think they can win now, so they gamble more. He gets his sure money from selling his system, and the casinos get there sure money from all the extra business he’s sending them. It’s a win-win situation, and casinos like win-win situations. And it explains all of the observed facts, without the awkward assumption that the house is stupid.
Well ok but that’s not very helpful to debunk the show. I’m just gonna go with it’s TV and it’s what sells.
I am careful.
Anyway my point is, walk into a casino to play roulette, virtually every number has an equal probabily to come up. The “virtually” is so that you don’t nitpick me to death.
By the way this is the famous episode I mentioned earlier. I had assumed the TV show was about Joseph Jaggers. He made some serious money exploiting roulette wheel bias in Monte Carlo but since then (my opinion) casinos have wisened up and you can’t beat the game like this anymore.
Interestingly a roulette wheel is in some ways similar to a software RNG. It has a seed (dealer casting the ball) and a mechanical algorithm. Both elements can induce bias but like it has been said casinos have very strong checks and balances for that.
And I just noticed that the same Wikipedia article that mentions Joseph Jaggers also mentions Gonzalo Garcia-Pelayo:
I understand that date/time(milliseconds) as a seed can’t truly produce random numbers.
If it’s generated by a machine. A machine could recognize the pattern. But a human won’t be able to.
But for things like poker, online gaming, certainly that must be good enough?
Combine that against the thousands that may be playing at a time and I can’t imagine that a pattern would emerge. One person that decides to go to the bathroom should throw the whole thing into new randomness.
It seems to me, that the most random thing we have is the attempt to predict world wide weather. Couldn’t they toss that into the mix? A seed for the seed for the seed.
It seems that to get truly random, something that can’t be recognized as a pattern, we need to include psychology and movement.
I just scratched my nose. And coughed.
I guess nothing, in the end, can be truly random because it’s predicated on something else.
I think the best seed could be gleaned from the movement of our new puppy. Ya never know what she is going to do next.
There are several types of ‘randomness’ involved in a roulette wheel. The first is the speed at which the wheel spins. The second is the speed at which the ball spins. The third is the position of the wheel when the ball was put into play. Finally, ‘randomizers’ on the wheel bounce the ball around as it comes off the track.
When all of these are added together, the distribution of the outcomes will be pretty random. No doubt about it. HOWEVER, for any combination of wheel velocity, ball velocity, and starting position, the results are decidedly biased.
First of all, a ball spinning on the track of a roulette wheel has a ‘capture velocity’, the speed above which it will stay in the track. Drag gradually slows the ball down, and once it reaches its capture velocity it will fall out of the track. Since the drag is relatively consistent from spin to spin, this implies that if you know the instantaneous velocity of the ball, you can easily predict the location on the track where the ball will drop into the playing area.
Second, the rotation of the wheel behaves in a predictable manner. For a particular instantaneous velocity and position, you can predict exactly where the wheel will be when the ball drops.
Therefore, all that’s left to ‘chance’ are the randomizers on the table, and the dynamics of the ball’s collision with the wheel and the frets between numbers. This is enough that you cannot predict which number the ball will land on, but the distribution of results from this point is not even. For example, if you hold a ball stationary on the track of a roulette wheel, and release it as the 00 passes by, and plot the results of 100 trials of this, you will find a definite bias of some numbers over others. There will be sections of the wheel where the ball is less likely to stop than other sections.
A number of people have built computers that can be used to exploit a wheel in this way. The wheel is not biased or broken in any way, but can still be beat. It works like this:
You have a computer with a touch switch in each shoe. One side is a ‘mode’ switch which resets the computer for the next run and can also be used to put it in ‘train’ mode. With the computer in ‘train mode’, you tap a button when the ball passes a fixed point, and tap it again when it hits that point a second time. This gives you the instantaneous velocity of the ball. Then you tap a third time when the 00 passes a fixed position, and a fourth time when the 00 passes again. Now you have the velocity of the wheel and its position. Finally, when the ball lands on a number, you record the number with a simple code of taps. Repeat for a while, and you will build a decision table mapping the wheel. Then you switch the computer into ‘play’ mode. Now you tap out the ball and wheel positions and velocity again, and the computer uses a piezo device to tap out the number to bet.
People who have used such computers have gained as much as a 30% advantage over the house.
There are two problems with this scheme - first, it’s a felony to use a device to beat the game. And the casinos are very good at catching you. So don’t do it. Second, the casino can easily thwart you by simply preventing the placement of bets after the ball is in play. Unlike bias exploits, you can’t know which numbers to bet until the ball is in motion.
But here’s a more interesting question: Given that it’s possible to do this, is it possible for a human to do it without a computer? Then it would be legal. I actually spent a little time doing some rough experiments around this, trying to figure out if I could ‘calibrate’ myself well enough to time a ball with a certain amount of precision without the use of a watch. It turns out I could get reasonable accuracy by just setting up a ‘rhythm’ in my head like a metronome and counting beats. Clearly not as accurate as a computer, but reasonably accurate. Accurate enough that I think I could maybe get a slight edge. However, the system you would have to come up with to be able to do all the calculations would be hideously difficult to use, and I wasn’t willing to go to any more effort (I did this exercise as background for an article I was writing). But based on that, I think it might be possible for a highly dedicated individual with a very quick mind and a lot of free time to *maybe be able to do better than chance at the roulette table.
The thing is, the casino can still shut you down by preventing post-betting, and since there are other games in the casino that can be beaten with much less effort, it seems rather pointless to me. It’s an interesting speculation, though.
This actually makes a worse random number generator. It’s actually a quite common source of flaws when an ignorant designed decides to seed the seed a few times and ends up with a RNG thats ridiculously weak.
I believe the best RNG source is the decay of single atoms in a radioactive source. From what we know of physics, this is one thing which seems to be truely random. However, you have to be careful massaging the exponential distribution in time into a flat distribution. There have been several schemes which do this which have had serious flaws found in them.
BTW: I did a google on this Gonzalo fellow and the only two english sites on the first 4 google pages was the TV guide linked above and the Wikipedia article. If he did indeed do what he claimed, surely there would be something written up about him.
This gets off on the philisophical point of free will, but do you really think your coughing and scratching are entirely random. They are deterministic results of electrochemical events in your body, in response to deterministic events in your environment (unless you’re reacting to the decay of individual atoms). No, our simulations are not sophisticated enough to track all the individual dust particles, predict when they will irritate your nose, and predict your reaction, but the fact that we can’t simulate it (yet) doesn’t mean it’s not deterministic.
The fact that it’s not random doesn’t mean it doesn’t make a useful seed for a RNG, but still, in a discussion like this, we should be careful about what we label random.
I have read articles about generating random numbers based on the seek time for hard drive accesses, the time between key presses, pictures from a web cam pointed at a blank wall. Just about any time the computer has to talk with something out side of its own clock regime is a good source of random numbers even more so if the process somehow involves measuring an external event. The main trick is estimating how random things are and not over using the randomness.