Application of the Wavelet based Multi-Fractal for Outlier Detection in Financial High-Frequency Time Series Data

Financial market experiences a high degree of fluctuation that may be related to economic events. In this paper, we employed the multi-fractal formalism based on WTMM (wavelet transfer modulus maxima) to test the existence and the location of outlier in high frequency time series. On the foundation of empirical analysis, we drew the conclusion that it is reasonable to incorporate this wavelet arithmetic to analyze the properties of intra-day data which show different distributional characteristics from common low frequency data