Doing a statistically-based analysis (i.e. calculating means, labelling outliers) on such a small sample size is of almost no use and is likely more misleading than anything.
With this small amount of data, there is no basis whatsoever to label the first day as an "outlier", since we don't have nearly enough data to reliably characterize the statistics of the daily returns.
Frosty, it has been said before, but bears repeating, if your backtesting results show a longterm profit and you trust them (both of which seem to be the case) and you've done the proper money management/risk calculations to ensure that you can handle the worst times you project, you just need to hang in there.
One suggestion, if it hasn't been said before, it seems that much of your anxiety comes from the nature of your algorithm's returns, you said you expect it to have a few big winners, spread between many smaller losers, correct? It's natural that such a "herky-jerky" equity curve would cause nervousness. If you can't handle the ups and downs, try to modify the algorithm to achieve a smoother curve (a lot to ask, I know), but otherwise just have faith that IN THE LONG RUN (and I emphasize that because you are nowhere near approaching the long run, yet), things will balance out.
Good luck and thanks for your candidness, I'm watching with much interest.
