Recognition Model Study Of The Singular Points Of Financial Time Series Based On Wavelet Transform

The financial data are usually highly noisy and contain outliers.And efficient markets hypothesis demonstrates that the market prices fully reflect all available information. Furthermore, previous studies suggest that public information arrivals could lead to volatility of stock prices. Therefore, the formation of financial time series are frequently affected by the outliers which are caused by some unpredictable or uncontrollable information, such as policy changes, large international or company-specific events and natural disasters.On the other hand, the existence of outliers may affect the trend of the financial time series, reflect some important character and bring much significant investment information. such as the abnormal market effects and financial volatility. The abnormal fluctuation of financial market and market effect frequently reflect the correspondence between the financial market and domestic or international significant economy events. Therefore, how to efficiently locate and estimate outlier has become a significant research topic.Wavelet transform, a theory developed from Fourier theory, is a math tool of time series analysis, image processing, signal denoising, etc. It has some unique characteristics in detecting singularity of time series and signal. Wavelet analysis is a time-frequency local analysis method which can adapt the requirements of time-frequency signal analysis automatically and focus on any signal details.In this paper, the author presented a simple method using Wavelet Transform Modulus Maxima Method to detect outliers and locate them with corresponding information and obtained some conclusions:1) The author presented a identify model of outliers in securities time series using Wavelet Transform Modulus Maxima Method, and testified the accurate and effective in detecting and locating outliers in time series through this model.2) Through detected the location of outliers of time series, we find out some information sources may affect these outliers. Then, the correspondence between the macroeconomic environment and the outliers of time series is able to be observed.