As far as statistics go you can test the sample size in several ways. It is important to have a good sample size so that the calculated mean number of pips you can expect is accurate. Purely statistically speaking you could:
1. Crunch your numbers and come up with a mean and do just as you suggested and take a random sample on the sample mean. If the distribution comes back normal, bingo you have the correct mean.
2. You could crunch numbers and attempt to fit a distribution and then use the statistic to calculate a sample size.
3. You could analysis price relative to your entry point for your trading method and then fit a distribution. From that distribution you can find the probability that price will be above or below your entry and optimize stops and profit targets that way. (That is what I do-it seems to work pretty well too!)
Of course the drawback to this is that your system would have to be 100% mechanical to effectively analyze (If you consider that a draw back).
Can you give an explanation of your spreadsheet? Also if there was some comment in the workbook or code to see how you are calculating these numbers it would be helpful too.