Dynamic Hedge Fund Portfolio Construction: Exponentially Weighted Returns Approach

This article proposes an exponentially weighted returns approach for constructing portfolios of hedge funds. This approach, which gives more weight to more recent observations, allows for volatility and higher moment dynamics in portfolio construction. Using monthly hedge fund index returns from the Hedge Fund Research Database for January 1990 to December 2020, we find that the proposed approach significantly improves portfolio performance in terms of enhanced returns, reduced risk, and improved risk-adjusted returns compared to the benchmark equally weighted approach. However, the exponentially weighted method generally requires more frequent portfolio rebalancing to capture the return distribution dynamics, and the turnover varies across different portfolio optimization models. Therefore, investors should select an appropriate optimization strategy when implementing this approach for hedge fund portfolio construction. Moreover, we show that the results are robust to the choice of decay factor for the exponential weighting, target returns, and estimation window size, and across low-volatility, high-volatility, and most recent evaluation periods.