Stationarity and Self-similarity Determination of Time Series Data Using Hurst Exponent and R/S Ration Analysis (original) (raw)

Advances in Intelligent Systems and Computing, 2021

Abstract

Time series data is highly varying in nature. Determining the quality of predictability of the data is necessary to describe it. Self-similarity and stationarity are the key tools to determine the property. In this paper, visual and quantitative results to measure predictability of time series data are shown by rescaled ratio (R/S) analysis and Hurst exponent. We use several transformations and scaling to avoid the noise and vastness of stock data. Case-based studies are done on various kinds of stocks from Bombay Stock Exchange to establish the necessity of the R/S ratio with Hurst exponent. From the results of this study, an inference has drawn about the nature of stocks. The predictability is quantified depending on the value of Hurst exponent and Hurst co-efficient. Another factor named roughness factor is included for analyzing the result of the R/S ratio.

Debayan Ganguly hasn't uploaded this paper.

Let Debayan know you want this paper to be uploaded.

Ask for this paper to be uploaded.