Multiscale Volatility Analysis for High-Frequency Financial Data

University Of Connecticut, Storrs CT

Investigators

Abstract

ABSTRACT Proposal Id: DMS-0504323 Institution: : University of Connecticut Principal Investigator: Wang, Yazhen Title: Multiscale Volatility Analysis for High-Frequency Financial Data Volatilities of asset returns are pivotal for many issues in financial economics. They are the key ingredients for the pricing of financial instruments, portfolio allocation, managerial decision-making, and financial risk management. Existing financial modeling mostly employs parametric models like GARCH and stochastic models. These parametric models are used for modeling volatility dynamics of low-frequency interdaily data and generally fail to describe the volatility patterns in high-frequency intradaily data, partly because high-frequency data have complex structures such as jumps and market microstructure noise contamination. The investigator initiates an innovative research on modeling and analyzing of high-frequency data. By taking advantage of the fact that jumps, volatility and microstructure noise tend to behave differently at different scales, he uses wavelets' multiscale structure as a platform to realize the fact and perform efficient volatility analysis for high-frequency return data. The proposed nonparametric multiscale methodology is very flexible and applies to high-frequency data with (a) microstructure noise, (b) jumps, and (c) leverage effect. It achieves the following goals: (1) utilize all available data efficiently, (2) test and detect jumps, (3) construct noise resistant realized volatility for estimating integrated volatility, and (4) separate the jump variation from the total quadratic variation. Financial data are collected at increasingly shorter time horizons. Over past decade there has been a radical improvement in the availability of high-frequency financial data (which is referred to as intradaily data, while low-frequency data mean financial data at daily or longer time horizons). Nowadays high-frequency financial data are available for financial instruments on markets of all locations and at scales like minute and even individual bids to buy and sell. One example of high-frequency data is the currency transaction data that are collected at every minute. High-frequency finance is an emerging new and challenge research field. The proposed research invents new statistical techniques to analyze high-frequency financial data and accurately model and forecast financial markets. It advances the understanding of high-frequency volatility and provides better volatility evaluation for practitioners in financial industry, where mistakes and wrong judgments can lead to catastrophic consequence for the industry and the rest of society.