Heck mm not only do I have the knowledge in statistics but I made a living with a Global company, leading a team forecasting inventory on 'service parts' using both historical models and causal models that we developed. So please make no assumptions about what I know or do not know! Some people have an education from a fancy school, some people have an eduction from a good school with years of practical experience.

I get what you are saying but their is another statistical factor that talks about using the size of a particular data set to make what is normally defined as an outlier as a recurring event or within a normal frequency. This also leads to a discussion of what it the statistical smallest or largest data set that should be used for a statistically valid model. Again this comment was in relation to your discussion on the 3%....you data set is to damn big!