But that assumes that one could predict in advance what those highest performing funds would be. Of course the highest performing XXX will be above average. The point is that no one can consistently predict which ones will be the highest performing XXX. Worse still, regression to the mean indicates that if, in 2021 you invest in the highest performing XXX of the preceding X years, you're more likely to get returns below average.

And let's suppose that there really is a fund staffed by a group of super-geniuses who can reliably pick the best performing stocks. There are two possibilities: They take on more and more investment capital until their returns end up getting closer to the mean or they don't accept new investments and the hypothetical new investor is left out in the cold.