Thanks for the nice words. Basically, even a normal distribution that looks symmetrical will have a slight bias one way or the other - just an artifact of random sampling (if you had an infinite number of iterations, it would disappear - but for any finite number there will be some skew). So the 50th percentile might be slightly higher or lower than the mean. You can see the bias using the SimulationSkewness function - it will be small, but not zero.
Try comparing it to the simulation mean (or if you're really worried about it, I can tell you how to pull out all the sample values and use Excel to calculate averages and percentiles).
Alex Edwards
alex@riskamp.com