Hilbert-Huang Transform Based Jump And Volatility Analysis On High-frequency Data

For adapting nonlinear and non-stationary time series data, empirical mode decomposition (EMD)and Hilbert-Huang transform(HHT) provided a effective and adaptable method. HHT not only provided a more precise definition method of time-frequency than other normal method, but also provided more physically meaningful interpretation about underlying dynamic process.EMD method also acted filter for extracting different scales variations in nonlinear and non-stationary data. HHT had be applied to many field, including Finance, biology, meteorology, etc.This paper based on research about Hilbert-Huang transform method, first reviewed the traditional spectrum analysis method, and discussed the basic idea of EMD and Hilbert spectrum and the corresponding algorithm, further researched several issues and basic solutions about Hilbert-Huang transform in practical applications need to be addressed. Objectives and innovation of this paper is to estimate the jump point and the volatility by EMD method and Hilbert-Huang Transform on high frequency data, particularly financial high frequency data. To achieve this goal, the paper discussed the correlation method based on traditional methods (especially the wavelet transform) and Hilbert-Huang transform method were applied to estimate the jump point, de-noising and volatility estimation. Finally, the above method was applied to the price of a stock5seconds interval frequency data, and the experimental results were analyzed and compared each step. In the case study section, respectively, of the data preprocessing, the jump point estimates, data adjustments (remove the mutation point), de-noising and volatility estimated five steps, the final number of points calculated jump within the stock price and the corresponding time of day point, the number and volatility of volatility under different ranges.