Wheel Memory and Sample Size

Hello,

Great site and lots of good info. Thanks for having it!
Two questions:

1. If a roulette wheel does not have a memory, should it matter if I played X number of spins in a row, or X/10 spins per day for 10 days? Would those gaps in play (while the wheel is still spinning without me playing/tracking) skew the results gravitating towards the expected results over X spins (excepting of course a difference for normal distribution for the results in X spins)? Seems like X spins is X spins, no matter when or in what order they occur.

2. I think I've read that the bigger the sample size, the closer/tighter the results should reflected the expected results (you're more likely to see a 5.26 advantage in a roulette simulation in 1,000 spins vs 100 spins). That makes sense. Is there a point where the stability levels out? If you have more stability in the results at 1,000 than 100, does 10,000 get you even more accurate? Is there a "magic number" sample size where more spins won't make the results more accurate?

Thanks,
Trooper13

Quote: Trooper13

Hello,

Great site and lots of good info. Thanks for having it!
Two questions:

1. If a roulette wheel does not have a memory, should it matter if I played X number of spins in a row, or X/10 spins per day for 10 days? Would those gaps in play (while the wheel is still spinning without me playing/tracking) skew the results gravitating towards the expected results over X spins (excepting of course a difference for normal distribution for the results in X spins)? Seems like X spins is X spins, no matter when or in what order they occur.

There are so many versions of random, but i suspect people around here will simply answer "no" to this question, if the wheel is considered completely "random".

Quote: Trooper13

2. I think I've read that the bigger the sample size, the closer/tighter the results should reflected the expected results (you're more likely to see a 5.26 advantage in a roulette simulation in 1,000 spins vs 100 spins). That makes sense. Is there a point where the stability levels out? If you have more stability in the results at 1,000 than 100, does 10,000 get you even more accurate? Is there a "magic number" sample size where more spins won't make the results more accurate?

Sample size means that your first question, or assumption that roulette doesnt have a memory, would mean you answered your first question, because trying to use past results as a predictor for present results is not necessary.

Neither the wheel nor the little white ball know whether you are standing there or not, nor do either of them care if you win or lose. Neither will work on your behalf or work to even things out in the long run and restore the proper balance of the universe or of that particular wheel.

I hate on Roulette. But... it’s useful for one thing. If you want to turn $5 into $180 at a casino, there’s no cheaper way to do it.

>no cheaper way to do it.
Well all those math types may soon chime in with some quibbles to the nth power but it is good to know that it is atleast amongst the cheapest and you get some music, a drink, and some of that ambiance stuff that sure beats the local Quik Mart hands down. Now as for a comp for a Deli sandwich, maybe and maybe not but you sure are not going to get any comps at the Quik Mart.

(Note: I said 'are not'... ).

Welcome to WoV, Trooper13.

1. Gaps in play should result in different results (than continuing to play in a single session), but never should cause the data to be skewed one way or the other from random.

2. Larger sample sizes allow us greater confidence that our sample data correctly reflects the greater universe, here a random universe of 38 possible outcomes. More spins will increase our confidence in our data, but one never gets a guarantee. There is no measure for "more accurate" as you seem to use it, at least, none of which I am aware.

If it is convenient to collect samples of 100 roulette spins, why not collect data in that format? You may learn a lot by comparing the degree to which the smaller samples vary from their statistically expected values.

My newest and most trusted system is whenever i walk by a roulette table i put the min bet on the first column, and start to walk away and normally the table is like, he doesnt care so im going to let him win as the table thinks im not going to walk away with my winnings. little does that table know, i am.