Investigating Time Series Data for Self -Similarity Estimation

Abstract: An investigation has been prepared to explore concept of self-similarity and stationarity nature of time series data of a stock market. Two time series data namely the Bombay Stock Exchange (BSE) as well as, the Stock Exchange of Hong-Kong (SEHK), the average parameters have been observed from January 2007 to November 2017. Analysis of both parameters has been carried out, through statistical techniques, to establish the nature of scaling pattern and non-stationarity. The calculation of Hurst exponent done by WVA and VGA showed that the time series is anti- persistent. Augmented Dicky Fuller Test (ADF), Kwiatkowski–Phillips-Schimdt-Shin test (KPSS) and Continuous Wavelet Transform (CWT) have been used to test for non-stationarity.

Keywords: Stock Market; Hurst Exponent; Fractality; Stationarity; Continuous Wavelet Transform