Title: The informational content of implied correlation
Authors: Nikolaos Voukelatos - University of Kent (United Kingdom) [presenting]
Abstract: Option-implied correlation has been shown to be an efficient predictor of market returns. We examine the source of this informational content. We document that the predictive ability of implied correlation stems from the interplay between its high-frequency and low-frequency components. The high-frequency component captures short-term trends, and it is found to be a robust predictor of market returns at short horizons, outperforming the original series of implied correlations. The low-frequency component reflects longer-term trends and optimally predicts market returns at longer horizons. We also provide evidence that decomposing implied correlation substantially improves the out-of-sample predictability of market returns at horizons of up to one year.