Multi Fractal Detrended Cross Correlation Analysis based on Indicator in Financial Time Series: Case Study of Forex Market

Subject Areas : Financial engineering

Zohreh  Alamatian ¹ (Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran)
Majid  Vafaei Jahan ² (Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran)
Reza  Sheibani ³ (Department of Computer Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran)

Received: 2021-11-30 Accepted : 2022-01-23 Published : 2022-09-23

Keywords: Correlation analysis, Time series, Indicators, Multifractal Detrended Cross Correlation Analysis,

Abstract :

Modeling synchronous time series in financial systems is very complex. In order to analyze such series, we require procedures that can determine long-term relations with high accuracy. Multifractal detrended cross correlation analysis (MFDCCA) is a technique to analyze long-term relations through detrending the time series. In this work we propose a novel technique for a more accurate detrending of a financial time series, called indicator-based multifractal detrended cross-correlation analysis (IMFDCCA).We aim at using financial market technical analysis indicators to better determine correlations between financial time series.We investigated our method on currency pairs EUR/USD and USD/JPY and their long-term and short-term relations of these series were determined as multifractal.In order to evaluate the effectiveness of IMFDCCA, we used R.S and GHE techniques for the Hurst exponent estimation. The evaluation results on a collection of 8 years data (2011-2019) show that the proposed method compared to the baseline (MFDCCA) reduces the RMSE by 30% and 26% using R.S and GHE respectively.