Well, the lack of detail in that paper is by design -- it's a self-described intro, focusing on examples and results... not much more than an executive summary, you could say. I don't agree on the over-fitting part at all. Look for the other papers by Kat and Palaro at the same SSRN website for a lot more detail. I can easily see certain institutional investors buying into this idea of synthetic HF returns replication, with enthusiasm. Obviously, Goldman must see large potential bottom-line value there, to go ahead with the launch of that ART fund.
I think you miss-judge the number of investors who invest in FOFs just to diversify risk in their portfolio's. Not all investors are looking to maximize returns. Many are just trying to reduce the unit of risk per unit of return. A stellar performing fund might have great returns, but they might also have a lot of volatility in those returns. If you are an adherent to generally accepted portfolio management theory, then you will try to find ways to diversify you risk away. What many investors are looking for is an investment that has a low correlation to their current holdings... -Neo
I am sure you are right. Pesonally, I would not invest in something just for diversification if it did not provide the returns as well. Historical correlation is often a very poor indicator of future correlation so adding something just for diversification can be folly. Return/risk is important but the returns distribution is also very important IMO.