Apparently you did not read (or did not like) my answer to Jason's interview question "Talk to me about the additional fiction that active management is a zero-sum game." In a nutshell, I said while it has to be zero sum over the entire stock market, it does not have to be the case within each segment of the market (more detail in my interview response).

The fact that the vast majority of active equity managers are skilled stock pickers means that if the fund avoids asset bloat, benchmark tracking, and overdiversification (what I call portfolio drag), there is an excellent chance the fund will outperform.

One of the interesting research results is that the top 20 or so relative weight holdings outperform while the lower relative weight holdings underperform. In one of the stranger twists, we find that a fund's high conviction stocks do well at the expense of the low conviction stocks in the same portfolio. This is a consequence of imposing drag on the portfolio after purchasing the manager's favorite stocks.

As I also noted, Wermers finds that the average stock held by a fund outperforms by about 130 bp , which is roughly equal to the average fees charged by funds. As a takeoff on Sharpe's comment you quoted, in the world of active equity mutual funds, the average stock held outperformance equals average fees, leading to near zero alpha across funds. That is, skill is widespread but its benefits are wiped out by portfolio drag and fees.