Sample size

[Guest scozz2003]

I was wondering if anyone has any ideas how to determine mathematically the confidence you can have in the results from your sample. If I have 500 bets is that enough to be "fairly confident", or do I need 1000, or 5000 or what? Obviously the more the better, but at what point can you become confident that your results are representative of the "population"? I am sure there is a statistical/mathematical way of doing this but I slept through the Introduction to Statistics class at Uni :| Presumably any such thing would also depend on the average probability of the trials in the sample i.e. you would need more trials to work out the odds of winning the lottery than you would need to work out the odds of tossing a coin and getting a tail...if you see what I mean. Any advice or pointers to such theory would be appreciated. :)

[Guest madmick]

confidence testing scozz you want to look up confidence testing and/or hypothesis testing. In particular things like the f-test, t-est and chi-squared test will be relevant to you here M.