We introduce a simple and intuitive approach of modeling and forecasting correlations for use in portfolio optimization. The model is composite in nature and consists of elements based on a bivariate realized volatility model. Importantly, our framework allows for volatility spill-overs between assets which provide an edge compared to competing models when forming portfolios. We apply the model to high-frequency data for commodity markets and demonstrate significant economic gains for an investor basing portfolio decisions on our modeling framework. This gain is significant in economic terms, even after imposing realistic constraints on short selling and portfolio turnover.

A press conference (pc) organized by the Federal Open Market Committee (fomc) followed half of the scheduled announcements from 2011 to 2018. We document that excess stock returns are strongly and positively related to their betas on announcement days with a pc. In addition, the cross-sectional dispersion in betas declines substantially on pc days when measured using both daily and intraday return data. These effects are absent on announcement days without a pc. Last, we find that stock-bond correlations are positive (negative) on pc (all other) days and that their variations are related to uncertainty and yield curve information. We discuss implications and possible explanations for our findings.

We present a new approach to selecting actively managed mutual funds that uses both portfolio holdings and fund return information to eliminate funds with predicted inferior performance through a sequence of pairwise fund comparisons. Our methodology determines both the number of skilled funds and their identities, and locates funds with substantially higher risk-adjusted returns than those identified by conventional alpha-ranking methods. We find strong evidence of time-series variation in both the number of funds identified as superior using our approach, as well as in their performance across different economic states.