The hard part is not regressing to the mean, and consistently beating the market, which sure, nobody can do. But if I learn today that a stock is going to rise tomorrow, and I buy as much as I can, the stock will rise a bit today. So the market has become a little bit more efficient.
That I happen to then base all my subsequent trades on overconfidence and tossing darts doesn't retroactively invalidate that trade. You can even make a market more efficient on losing trades- you short a stock, it goes down a bit, but you spend more maintaining the short than the stock goes down. Then, the stock crashes after your short expires. Oops! But your prediction was still partly right, and that information was integrated into the price a bit earlier than it would have been without you.
Finding someone successful at stock picking and asking him what he did to be successful is like finding that 1 out of 1024 person who flipped a coin heads 10 times in a row and asking him what makes him such a great coin flipper.