Abstract:
Active portfolio management techniques and tactical asset allocation in particular may be growing in importance for traditional assets, but relatively little research has been conducted on hedge funds. We seek to build on the work of. Amenc (2002) which tried to predict nine CSFB/Tremont hedge fund indices utilizing financial variables and lagged index returns. Instead of predicting individual indices, we use Principal Components Analysis on twenty-five indices of two providers for the 1996-2005 period and find three common factors explain 64% of the variation between index returns. The three factors relate to equity market performance, and the relationship between corporate credit spreads and equity market performance and expected volatility. Further departing from Amenc (2002), we lengthen the forecast window to three and six months because a well-known "return smoothing" effect impacts the less liquid hedge fund strategies in particular, but may not be exploitable due to liquidity constraints. We develop regression equations to predict the tree factors, which incorporate financial variables on the equity market, credit spreads, and expected volatility. Although the first factor is only marginally predictable using market valuation ratios, the second and three factors appear to be more predictable. An asset allocation optimization over seven years (1999-2005) incorporating the predicted equations show modest improvement of five to eight basis points a month in realized return over an allocation using only historical factor exposures and returns.
