Funds of Funds – Questions

I’m curious about funds of funds – that is, hedge funds whose primary investment is other hedge funds. (Full disclosure – yes, this is for a current project, and yes, I have data, but its proprietary and I’m not allowed to share, though perhaps some of the results may be shareable at some later point.)

I understand that most hedge funds don’t actually beat the market, but a few of them do manage to do so systematically. Presumably, some hedge fund managers really do have some skills and/or knowledge that is “above and beyond.”

Presumably, we would see something similar among funds of funds, after a while. Sure, there is the added problem of the dual layer of fees (there are fees for the hedge funds, and then fees for the fund of funds), but perhaps that can be gotten around. (A fund of funds might, due to its assets and connections, be able to negotiate fees for the hedge funds in which it invests down. It might also be able to keep its own fees down, in part by hiring outsiders – say, for example, a number cruncher – to act as consultants on a specific task instead of retaining that expertise in-house in a place like Westchester or NYC or Bermuda.)

So here are some questions…

1. Do we see successful fund of funds? Is there any specific behavior that works well?
2. What are the things a fund of fund should look for among the funds in which it might potentially choose to invest. Obviously, it would want funds that perform well over time (how long?), have relatively low volatility, and whose performance (and style and investment strategy) are uncorrelated with each other, so that if the doo-doo hits the proverbial fan with one or two of the underlying investments, it won’t hit it for all of them.
3. Can you determine if a given style or approach is going to stop working? For example, a fund that invested in real estate would probably have done very well from about 1998 until recently. I believe many funds (I assume the small ones?) make money mostly by writing options which rarely get cashed.
4. How far back should one look? Is data from 1975 really relevant? What about 1989?
5. What am I not asking?

Thanks.