A0327
Title: Warping time: Improving efficiency of tick-by-tick data in portfolio optimisation
Authors: Vitali Alexeev - University of Technology Sydney (Australia) [presenting]
Giovanni Urga - Cass Business School (UK)
Abstract: A unified framework is developed that allows analysis of unevenly spaced tick-by-tick data. Up to 90 percent of tick-by-tick data is lost during pre-processing of the data to fit existing models in finance applications. Compared to existing methods, the proposed approach avoids unnecessary loss of observations and allow for flexible time shifts by warping the time domain. The resulting framework is capable of direct analysis of tick-by-tick financial data while simultaneously addressing the main empirical issues identified in the contemporary high-frequency literature for such data (asynchronous trading and microstructure noise). Correlation structures estimated based on properly aligned time series at high frequencies allow for improved portfolio allocation strategies and decision-making process for investment professionals.
