A0379
Title: Extracting efficient prices using intrinsic time information
Authors: Roxana Halbleib - University of Freiburg (Germany) [presenting]
Lukas Schmidt-Engelbertz - ETH Zurich (Switzerland)
Abstract: A comprehensive framework is presented designed to extract efficient price processes on financial markets, acknowledging and mitigating the impact of market microstructure noise inherent in high-frequency financial data. The framework comprises three pivotal stages: sampling, filtering, and optimizing. In the sampling stage, various frequencies and schemes are explored, both calendar and intrinsic time-based, to capture noisy ultra-high frequency prices effectively. Following the sampling stage, the methodology employs a Kalman filter in the filtering stage. This filter is utilized across multiple window sizes, aiming to extract the fundamental price process while minimizing the impact of market microstructure noise. The optimization stage uses a non-linear optimization approach to minimize the autocorrelation left, thus aiming to obtain a price process that exhibits martingale properties. The results demonstrate the success of this methodology in removing significant autocorrelation present in observed high-frequency financial prices. Furthermore, the approach offers adaptability and versatility, permitting methodological substitutions at each stage. This flexibility empowers researchers to tailor the framework to better suit the specific nature of their data or underlying models.
