37 Pages Posted: 30 Jan 2009 Last revised: 3 Feb 2009
Date Written: October 30, 2006
Hedge Funds are private, lightly regulated pooled investments which have the aim of obtaining absolute returns. The name hedge fund (HF from here on) is misleading since the word hedge implies defensive management or insurance against bad times. We can say that there is a common misperception through which one assumes HFs have a market neutral position.
This comes from the fact that the original idea of market neutral Hedge Funds from Alfred Winslow Jones (aim at zero exposure to specific equity market factors) has now developed in a larger and impressively growing investment sector. Freedom in the managing process and lack of regulation are probably the main reasons for the heterogeneity and variety of HFs. Another peculiarity of Hedge Funds is that the Fund Manager usually puts his own money at risk participating to the fund with a substantial part of his liquidity; this behaviour is called eating one's own cookies. The private nature and the lack of regulation takes us to the main limit in academic works on HFs: the limited information often plagued with some biases.
In this paper a broad and well known set of indexes, the CSFB Tremont Hedge Fund is taken in exam. I chose this one because of its long history and from an academic point of view it seems to be less biased compared to other non-investable indices. This Index is sold outside the USA and has an investable version with a participation of $1,000, clearly with a target also on retail investors.
The first step is the description of the main univariate statistics of HFs, from which the drawbacks and advantages encountered in HF investing are highlighted and in particular: non normality (e.g. the so-called negative skewed trading strategies) and autocorrelation (liquidity risk and data smoothing). In order to get rid of autocorrelation a filtration is performed. I apply a four moment CAPM to get the statistical properties of HFs over the simple mean variance framework in order to understand where their abnormal return comes from and to give a better idea about how much their value is when added to a general equity portfolio.
The second part is going to provide a brief State of the Art in the Hedge Fund Evaluation Research. I apply the Style Analysis to replicate the CSFB Tremont Hedge Fund Index by setting up reasonable common location factors and putting a gearing factor by variance targeting. Before doing the linear replica I check for multicollinearity and statistical significance in the location factors adopting the approximated t-statistic proposed in Lobosco and DiBartolomeo (1997). The model is as general as possible intended to avoid non-linearity (as in Jaganathan et al 1994). The location factors are a wide representation of the alternative risk factors hidden in the CSFB Tremont.
In the end of this first replication approach I take the first 6 years of returns and perform 12 months out of sample linear replica. Statistic of the linear replica are given and compared to Index. To justify this passive buy and hold strategy I give CUSUM Test results in order to check if there is stability in the weights of the model. Finally the out of sample statistics for each strategy are given and compared with the CSFB/Tremont Indexes.
Keywords: Hedge Fund, Hedge Fund Indexes, Style Analysis, Omega Ratio
JEL Classification: G10, G20, G32
Suggested Citation: Suggested Citation
By Bing Liang