Wow grats! Be warned though, this is a raw system designed by someone else who only put in a surface analysis. There is still some tweaking to be done to "tighten up" that win rate. According to Vitalij there is a 1 in 20 chance of busting out completely with the raw system.

How could the win rate be tightened? Instead of random number generation after a win, selecting a random number from say the 5 worst performing numbers in the past 37 spins?

Well first of all his simulation was based on 455 spins in the progression which is not the true loss point as the system doesn't quit on spin 455 and not to mention there won't be any spins past 455 with our number not hitting. We aren't likely to see that in our lifetime ever actually, anyway.

But the weakness will be in successive draw-downs without recovery because the progression is too weak to recover from the larger losses. What we need to do is discover the flags that indicate this stagnation has occurred and beef up the progression to recover better. Just my theory as I haven't seen the losing sequences yet.

Like I mean 200 spins a win, 90 spins a win, 300 spins a win, 75 spins a win prolonged out where the progression never recovers or earns enough to get back to profit before it gets drawn down further and further until it busts at -3200 units.

I am starting to suspect his simulation is flawed... If the definition of a loss is hitting 456 spins, we have nothing to worry about as we will simply not hit that point in our lifetimes and it is quite certainly IMPOSSIBLE that there is a 5% chance of hitting spin 456 without our number hitting -- simply IMPOSSIBLE.

I am willing to maintain some minor variance on this issue because the potential variance on a single number is just so huge BUT the variance MUST occur in the area of 1 in 10M+ odds which is NOWHERE near 5%, like in an another universe that is on a different plane of existence even. In other words, compare the number 20 to the number 10,000,000 and tell me they are even close??