Portfolio optimization using forward-looking information

In this paper we develop the first estimator of the covariance matrix that relies solely on forward-looking information. This estimator only uses price information from a cross-section of plain-vanilla options. In an out-of-sample study for US blue-chip stocks we show that a minimum-variance strategy based on this fully implied estimator consistently outperforms a wide range of benchmark strategies, including strategies based on historical estimates, index investing, and investing according to the 1/N rule. The outperformance is strong in periods of high information asymmetry, whereas in quiet periods all strategies lead to similar results. The outperformance can only be reached using a fully implied approach; partially implied approaches that combine implied moments with historical ones might even perform worse than purely historical approaches.