Charlie Springer wrote:
> On Sun, 14 May 2006 15:38:33 -0700, Mike Hore wrote
> (in article <C08DE680.6F793%mike.hore@[EMAIL PROTECTED]
>):
>
>> Did you mean Poisson, or was that deliberate? Ummm... I'd think that
>> any system that aims to predict random numbers would be "poison" -- our
>> universe just doesn't seem to work that way ;-
>
> Yes, but I thought the misspelling looked appropriate :-)
>
> There are 38 numbers on a roulette wheel and the payout is 35 to one.
OK then, right off the top of my head this looks like a random walk
problem in 2 dimensions, with a probability of a step up of 35 units
being 1/38, and the probablility of a step down of 1 unit being
37/38.
I forget how the math goes after that, but it looks like after N
steps your average position is going to be down by 3/38 * N ???
Of course your might get lucky on a particular day, but on other days
you're bank is going to run out. In the long run the house's bank wins.
It always does ;-)
>
> My calculations (and simulations) showed a 50% chance of a given number
being
> a winner within the first 26 spins. The break-even point is 35 spins.
Something's up. This means if each day I decide on 30 spins, it's going
to cost me say $30 at a dollar a spin, but I've got a better than 50%
chance
of winning $35? So over a year I'm very likely to be ahead? I'm not sure
where the error is, but that just can't be right.
>
> The problem is twofold. First, there is an infinitly long tail to the
curve
> (according to the distribution you are most likely to hit on the first
spin.
> Does this mean you should pick a new number each time? No, but it can
get
> confusing). Second and rather insignificant, I think the Poisson in this
case
> is an approximation to a binomial distribution where with large numbers
of
> trials the approximation is very close.
>
> I tried it out Friday and made $200 into $6,000 in about 5 hours and
reduced
> it back to $400 on Saturday! Darn! I could have had a quad G5 and Grid
> Mathematica with a couple thousand left over!
Hmmm... I think even if you decide on a "quit when ahead" strategy, some
days you'll never be ahead...
>
> As a friend from Stanford said, "If you are teaching physics and playing
> roulette you should be hospitalized!"
Spot on, mate! :-)
Cheers, Mike.
----------------------------------------------------------------
Mike Hore mike.horeREM@[EMAIL PROTECTED]
|
