Econometric Methods for High-Frequency Financial Data Analysis

Project Start:01/2018
Status:Ongoing
Researchers:Aleksey Kolokolov, Davide Pirino, Roberto Renò
Category: Financial Markets, Systemic Risk Lab
Funded by:LOEWE

The project is dedicated to developing econometric methods for analysis of high-frequency financial data with the main focus on methods of model specification. Under no-arbitrage condition prices of financial assets ought to be semimartingales. Semimartingales form a rich class of stochastic processes including, in particular, the two classes of widely used models: Brownian semimartingales plus jumps (commonly adopted in financial modelling) and pure-jump models (extensively used for, e.g., derivative pricing and volatility modelling). The aim of the project is twofold. First, the project aims to develop data driven methods for distinguishing pure-jump and Brownian semimartingale models. Second, the project aims to study deviations from the semimartingale hypothesis observed in real stock prices (e.g., the presence of zero high-frequency returns), their economic determinants and their impact on the econometric methods. The two objectives are highly interrelated. On the one hand, the detection of deviations of the semimartingale hypothesis and market anomalies depends on the particular type of a semimartingale considered. On the other hand, as shown by the previous research, the presence of market anomalies, in particular flat trading and zero high-frequency returns, significantly bound (or even prevent) using the existing econometric methods. Simultaneous investigation of violations of the semimartingale hypothesis at micro level and model specification not only allows to provide more reliable econometric models but provides a bridge between the existing market microstructure theories and econometrics of discreetly observed continuous-time processes. The contribution of the project is expected to be important for virtually all fields of financial research and practice utilising continuous-time asset price models. In particular, I intend to study the possibility of occurrence of “flash crashes” in stock prices, which ought to be incorporated into classical continuous-time framework in different ways depending on the nature of the price process.

 

Related Published Papers

Author/sTitleYearProgram AreaKeywords
Aleksey Kolokolov, Giulia Livieri, Davide PirinoStatistical Inferences for Price Staleness
Journal of Econometrics
2020 Financial Markets, Systemic Risk Lab staleness, idle time, liquidity, zero returns, stable convergence

Related Working Papers

No.Author/sTitleYearProgram AreaKeywords
236Aleksey Kolokolov, Giulia Livieri, Davide PirinoStatistical Inferences for Price Staleness2018 Financial Markets, Systemic Risk Lab staleness, idle time, liquidity, zero returns, stable convergence
Back