Statistics advice to assess new strategy results

I'm testing a new strategy. I've been using S-A-G-E BJ simulator program and have generated and recorded results for 200 sessions with strict entry and exit conditions. Eachs ession represents anywhere from 8 to thousands of hands. What I don't know is what sample size N of session results I should be collecting until I can reach a Confidence Interval (C.I.) of X% that the results are representative of all sessions. I'd like to determine N for a C.I. of 90% and 95%. I know how to determine the standard deviation in my sample set. What I don't know is what distribution would be appropriate to use to approximate this experiment and what equation to use to solve for N. Any insight would be appreciated.