I am testing a strategy on my Casino Verite software, playing one-on-one against the dealer using liberal Vegas rules. I have only played 7,500 hands, which I know does not represent a level reaching statistical significance. But, I don't know what level does.
My question, then, is how many hands does it take to make an online strategy test statistically significant - 20,000; 50,000; 100,000 or more?
I do not have the software necessary to have the computer play independently. And, even if I did have the software, I'm not sure it would accommodate what I am trying to do anyway.
Any feedback on this issue of "statistical significance" would be greatly appreciated.