a quantitative model for asset allocation to hedge strongfundsstrong.pdfVIP

TECHNICAL ARTICLE 5 A Quantitative Model for Asset Allocation to Hedge Funds Hari P. Krishnan and Norman E. Mains Morgan Stanley Co., Floor 37, 440 So. Lasalle St., Chicago IL 60605 Introduction historical data. For example, using CSFB/Tremont rather than HFRI indices as a proxy for hedge funds would result in a significantly differ- How much should an investor allocate to hedge funds? This is a question ent asset class mix. Using a CAPM approach to calculate a long term that has been debated for some time, but (to our knowledge) has not been expected return for hedge funds may also be misleading, since their adequately addressed in a systematic way. Numerous practitioners have volatility is probably understated by index data. emphasized the importance of asset allocation relative to manager selec- Our approach to asset allocation uses a technique from information tion, which makes this question an important one. Certainly, the appro- theory called the maximum entropy method as a supplement to mean priate allocation to hedge funds depends on a set of investor preferences variance optimization. Maximum entropy methods can be useful when- and assumptions about different asset classes. However, under a specific ever we are trying to extract information from noisy or missing data. set of assumptions, it is possible to develop an asset allocation framework They have been remarkably successful in image processing, as in (Gull for hedge funds. That is precisely what we do in this paper. We do not and Daniell 1978) and more recently, in a variety of financial contexts. claim that all