Title: High-frequency volatility: Online and stream estimation perspective
Authors: Vladimir Holy - University of Economics, Prague (Czech Republic) [presenting]
Abstract: An important aspect of financial high-frequency data analysis is volatility estimation. Volatility of the price process is typically measured by the quadratic variation (volatility realized over a given time period), integrated variance (quadratic variation of the continuous part of the process) and spot volatility (derivative of integrated variance). These quantities are typically estimated by non-parametric methods under rather general assumptions about the price process and in the presence of the market microstructure noise. A specific process can also be considered and parametric methods utilized. From the computational point of view, it is natural to consider high-frequency data to be a data stream. The question is which volatility estimators can be computed by online and stream algorithms.