(Roulette continued)

Strategy, Clocking, and such

     

Forget about strategy for roulette. The results produced by a roulette wheel are random and each spin is totally independent of all others. Most roulette tables have a display that tracks previous results, but this means nothing. It is just another way of pandering to the gambler's fallacy. Just because red has been the result the last 15 trials does not mean red is less likely to be the result on the 16th trial.

There are those who believe that a roulette wheel can be timed and the results predicted with enough accuracy to overcome the house edge. Early on in his career Thorpe worked on this problem, but as far as I know, was never able to predict the outcome in a lab, much less inside a casino.

The basic idea of predicting the outcome of roulette is attractive because it would only require predicting the range of the outcome a little more than 50% of the time to overcome the house edge. It is also attractive because casual observation would seem to suggest that it is possible. However, taking a closer looks reveals more complexity.

If the roulette wheel was stationary and a ball was spun around the wheel, it would be reasonable to assume that once the ball reached a certain rpm (x), that the location of the ball on the wheel at that exact moment would predict where the ball would first fall off the inner wall of the wheel. If you knew this, you should be able to predict within a certain percentage where the ball eventually stops.

However, this is not the case. The wheel itself is spun in the opposite direction of the ball. Therefore, I would think that you would also need to know when the wheel reached a certain rpm (y) and where the ball is at that exact moment. These two variables would have to be used in conjunction to make any predictions.

This method would eliminate the problem of different spins having different characteristics, but adds requirements that can not be fulfilled in a casino setting. Furthermore, a large number of trials would have to be conducted for each wheel to discover the range of outcomes predicted by the variables (x and y), each wheel being at least slightly different.