In the scientific enterprise, outliers are often discarded as they skew the statistics. It’s understandable why this is done and the reasons make sense to me.

However, outliers are often what matter the most. Stock market crashes, bubbles, earthquakes, disasters of all sorts — your outliers are the most relevant data. Discarding outliers in many areas means that you now know much less than you did before. “It’s just an outlier” means you just threw away where something is going to break in the future.