Regression to the Mean

Extreme results are part skill and part luck, and on the next draw the luck component resets, so outliers drift back toward average. Investors apply it by distrusting extrapolation: the fund on a hot streak, the company at record margins and the star analyst are each likelier to get worse than better. Where luck demonstrably drives the extreme, trimming after abnormal goodness is the model executed in one move.

5 books · 3 people behind it · 4 resources