A1256
Title: On the predictability of stock returns using predictive equity analytics with dynamic state space
Authors: Mark Zanecki - IHA Consultants (United States) [presenting]
Abstract: Predictive equity analytics with a dynamic state space framework is introduced to measure both the static and dynamic components of equity return processes. A state is associated with each close return followed by state compaction to enable matrix eigenvalue quantification. The framework allows for identifying and quantifying outlier processes separate from the core return process. Short-run predictability is demonstrated where the signal is sufficiently intense above the background, and long-run predictability uses maximal eigenvalue as well as the first occurrence of meaningful change in eigenvalue. Predictive equity analytics with dynamic state space amplifies spectral analysis by providing a dynamic Kalman-like filter as a first step.
