The purpose of this course is to provide an up-to-date treatment of the core topics in the modelling of high-frequency data. Advances in computing and data technology make it possible to observe markets at very fine intervals of time. Using high-frequency data permits the calculation of realized measures which are superior to volatility measures generated from GARCH and stochastic volatility models. However, the processing and financial modeling of high-frequency data remains a challenge to both researchers and practitioners. This course aims to provide guidance on the techniques involved in processing, filtering and modeling such data. Using data from TAQ and TICK- DATA databases, the attendees will have an intensive introduction to both the theoretical and empirical aspects of high-frequency data.
The object of the 2-day course is to demonstrate the empirical techniques and methods employed to analyze high-frequency data with special emphasis on the calculation of realized measures, forecasting and Monte Carlo methods and design.
- Familiarisation with Matlab syntax, functions and writing new functions
- Computation of realised measures of volatility
- Introductions to theoretical foundations and mathematical models of continuous/discontinuous time modeling
- Forecasting techniques
- Monte Carlo Simulations: Design and implementation.
- Fundamentals of programming in Matlab
- Importing and exporting data
- Descriptive statistics and Density/log-density estimation
- Inter and intra-daily plots
- Time stamp, frequency conversion and data aggregation
- Data bases comparison Tick vs TAQ
- Data Types (Equity, Forex and Indices)
- Estimation of Quadratic Variation and its Components
- Stylized facts (Normality, persistence and noise)
- Intra-day periodicity
- Leverage eﬀect
- Jump estimation and identiﬁcation
- Forecasting using short and long memory speciﬁcations
- Monte Carlo Simulations