This paper compares a number of different approaches for determining the Value at Risk (VaR) and Expected Shortfall (ES) of hedge fund investment strategies. We compute VaR and ES through completely modelfree methods, as well as through mean/variance and distribution model-based methods. Among the models considered certain specifications can technically address autocorrelation, asymmetry, fat tails, and time-varying variances which are typical characteristics of hedge fund returns. We find that conditional mean/variance models coupled with appropriate distributional assumptions improve our ability to predict VaR, 1% VaR in particular. We also find that the goodness of ES prediction models is primarily influenced by the distribution
model rather than the mean/variance specification.