Re: Roulette.

Tudor Jones writes:
So the puzzle is, why does a random number generator
in a computer come up with different results from a
roulette wheel? After all, a roulette wheel should
be a perfect random number generator.

Five obvious possible reasons:
1. The roulette wheel is not a sufficiently good random number
generator.
2. Your program did not use a sufficiently good pseudo-random
number generator.
3. You didn't run enough trials to be sensibly able to draw any
conclusions about them.
4. You did run enough trials, but drew the wrong conclusions.
5. You're lying.

There is also:
6. Someone was stealing chips from you without your noticing.

This actually splits into 7 reasons, since reasons 3 and 4 apply
(A) to the computer test and (B) to the casino test. I'll rule
out reasons 5 and 6 for purposes of discussion since there is no
evidence for either one.

Okay, first, reason 3B obviously applies (which rules out 4B).
Without a statistical analysis that I'm not equipped to perform,
I don't know how many trials would be needed to sensibly form
conclusions, but I would think it would be on the order of 20
to 100, and they would need to have a prespecified terminating
condition.

My guess is that reason 2 or 3A also applies, and more likely
both. If 3A doesn't apply then 4A likely does, given the mindset
of a person who would even think that trying such a system could
improve their chances of winning at roulette.

Finally, I'm inclined to discard reason 1 on the grounds that if
the wheel had an obvious pattern on 50-50 bets then both players
and the casino management would notice. This would not apply if
the player was varying which way he bet (e.g. sometimes red,
sometimes black) *and* the casino was not only cheating somehow,
but choosing to cheat in such a way so as to beat *him* at the
expense of enriching players betting the other way; but there is
no reason to believe that scenario either.
--
Mark Brader, Toronto, msb@xxxxxxx
"But even though they probably certainly know that you probably
wouldn't, they don't certainly know that although you probably
wouldn't there's no probability that you certainly would."
-- Sir Humphrey Appleby ("Yes, Prime Minister") on nuclear deterrence

My text in this article is in the public domain.
.

Relevant Pages

  • Re: "hat" container class [C++]
    ... > replacement, with non-uniform probabilities. ... I would leave the choice of a random number generator to the client, ... There are good reasons for not making this choice for the user. ... RNGs) have some troubles with the low order bytes. ...
    (comp.lang.cpp)
  • Re: "hat" container class [C++]
    ... Kai-Uwe Bux wrote: ... > I would leave the choice of a random number generator to the client, ... There are good reasons for not making this choice for the user. ... > the first place for which the quality of low order bits is established. ...
    (comp.lang.cpp)
  • Re: rand() / (RAND_MAX / N + 1)
    ... > For reasons that have already been endlessly discussed on the faq, ... > generator. ... depending on the application it may not even be noticeable. ... finally he warns against using rand() % 6 due to the fact that the low ...
    (alt.comp.lang.learn.c-cpp)
  • random number generators
    ... Are there any reasons, other than reproducability of the series for ... What really stops a generator from using a ... What is the state of the art in PRNGs? ...
    (comp.theory)
  • Re: Fibonacci and the roullette wheel
    ... Alexander Donis wrote: ... Other than a long losing sequence, can anyone offer any other reasons why this system would not work? ... A roulette wheel does not have the numbers 37, ...
    (sci.math)