palestis,

The theory is not wrong. Skipping spins or "virtual" betting has no effect on the distribution of outcomes, and this can easily be shown.

The problem is that most gamblers have no understanding of scientific methodology or statistics. They have what seems to them to be a good idea, it may appear to be just common sense, but common sense is often wrong where probability is concerned, that's why it took thousands of years and some very smart people to develop the theory you are dismissing. Basic methodology says that when you're doing an experiment, you need a "control group", so instead of just assuming that when see 3 or 4 reds in a row, it will be to your advantage if you start betting black, you should run a parallel test without waiting for the 3 or 4 reds before betting, in other words just bet randomly.

If your theory is correct, the variance (dispersion or number of losses) will be fewer in the first "experiment" than in the second, but if there is no difference (the variance is the same) it means your theory is incorrect, and you will find this to be the case.

I'm not saying you're not winning, but the reason why is not because of any virtual bets. You can choose not to believe me, of course, but numbers don't lie.

Edit:

Sometimes you may not be sure whether the data shows a difference in variance or not. There are some statistical tests you can use for this, the most common being the F-test, which is available in Excel.

The test is sensitive to Normality (that is, it is assumed that the population the data comes from is Normal, but this will be the case for roulette outcomes, unless you have a biased wheel. ;-))

http://www.statisticshowto.com/how-to-conduct-a-statistical-f-test-to-compare-two-variances/