The argument stipulated by those who believe in efficient markets goes something like this: If 50 million people flip a coin twenty times, and get paid $1000 for every "head", there could be around 50 people that will have rolled "heads" all twenty times. Are such people particularly skilled? Should we copy their methods? Upon first glance, probably not.
But what if such coin flippers lived in the same neighbourhood? Or what if they all learnt how to flip from the same person? Wouldn't it be worthwhile to determine whether this group has some knowledge or skill which is allowing them to profit?
Warren Buffett makes such an argument in his article The Superinvestors of Graham and Doddsville. He shows us the investment returns of the people who studied with him under Ben Graham, demonstrating that the returns derived from this method of investing are no fluke. Below are the returns Buffett references from the investors he has personally worked with (despite the fact that there is little overlap between the stocks owned by these investors). I encourage you to read the full article here.