some do very well and some do very badly, but this is exactly what you would expect from a large number of market participants pursuing different rational strategiesThis suggests a possible scam (probably not an original suggestion). Set up some large-enough number, maybe 30 or 40, shell investment companies. Each one trades according to a very specific but randomly generated strategy, which it publicly announces in advance. (Some care would have to be taken in generating the strategies, but they should be such that they're expected to match the market, when taken as an ensemble.)
After a few years, some of these strategies will have turned out to be successful, others haven't. Quietly wind down the unsuccessful ones. The successful ones now look prescient: this firm announced in, say, 2008, that the market conditions were such that strategy X is clearly correct going forward. And history has now borne them out: here, five years later, their prediction was right and their portfolio is up 1000%. Of course, there is missing information, that this was just one of 40 shell companies, and overall returns across all companies did not beat the market. But now this successful company can sell investment services/consulting/whatever on the basis of its individual track record.
I suspect actual investment consultants are effectively doing a distributed, non-coordinated version of this, even if not intentionally.