Let's look at the logic of playing the unhit numbers, as in Turbo's example. Suppose you wait until there are only 12 numbers left unhit and then begin playing those numbers for a full cycle (37 spins). If there was any advantage in doing so, you would expect there to be some correlation between how long it takes until there are only 12 numbers left, and either (a) the number of wins you would get in the cycle, or (b) the number of spins before you got the FIRST win in the cycle (or both). eg, suppose it takes 50 spins before there are 12 numbers left. Wouldn't this be better than the case when it only took 30 spins before there were only 12 numbers left? If you created a chart with the "sleep" length on the horizontal axis (ie number of spins before 12 unhit #s) against (a) number of wins in the following cycle and (b) number of spins before the first win, you would expect there to be at least some hint of a trend or correlation between these variables. But there is none. Here's a chart showing the results of 1000 sessions with SLEEP on the x-axis and WINS on the y-axis : As you can see, there's no correlation or trend at all. It's just a random "cloud" which shows that it doesn't matter how long the sleep is; the number of wins could be anywhere between 5 and 19, regardless of how many spins you had to wait until there were only 12 numbers left. Again, here's another chart showing SLEEP on the x-axis and "Spins Until First Win" on the y-axis (SUFW) : Except for the outcomes being clustered more towards the bottom of the chart (which just reflects the fact that on average the first hit will occur sooner rather than later), there is again no correlation or trend, which means that the first win isn't likely to occur sooner if the waiting time for 12 unhits is longer.