Efficient Volatility-Managed Portfolios

Abstract

We propose a theory of efficient volatility-managed portfolios (VMPs) that allocate the variance budget between intraday and overnight periods based on volatility forecasts. The efficient VMP achieves the lowest volatility-of-volatility among all VMPs targeting the same volatility level. The optimal portfolio weights depend jointly on the accuracy of the volatility forecasts and the noisiness of the volatility proxies that are used to measure the portfolio’s ex-post realized variance. Volatility forecast errors generate an efficiency loss even when they leave the optimal portfolio weights unchanged, while distortions in the optimal weights create an additional loss. We propose a simple Diebold-Mariano-type test to rank competing volatility forecasts by their economic performance in implementing the efficient VMP. Empirically, the efficient VMP not only reduces the volatility-of-volatility but also improves risk-adjusted returns relative to simple VMPs and the buy-and-hold portfolio. Joint work with Lara Schadwinkel.