A novel integrated approach for analyzing the financial time series and its application on the stock price analysis

Abstract

With the advent of the big data era, the ever-increasing accumulated financial data plays a significant role in people's daily life. The ensemble methods are the most efficient and accurate methods for analyzing the essential phenomenon of the financial data. In this paper, a novel approach, which incorporates the complementary ensemble empirical mode decomposition (CEEMD) and semi-parametric linear non-Gaussian analysis (Spline- LCA), i.e., CEEMD-Spline-LCA, is proposed to analyze the nonlinear and non-stationary financial time series. The proposed CEEMD-Spline-LCA method consists of three steps: In the first place, the CEEMD is applied to obtain Intrinsic Mode Functions (IMFs) of the analyzed data. Next, according to the contribution coefficients between the IMFs and the financial time series, IMFs are reorganized to get a new collection of the IMFs (NIMFs) for the subsequent explanation of influence factors of financial time series. Furthermore, Spline-LCA is utilized to separate the NIMFs into independent components (ICs), reflecting the different inner driving factors. Concentrating the established model on the stock price (the Dow Jones Industrial Average, the oldest stock index generally used), by comparing with real economic indicators, we find that the obtained ICs are close approximation of the exchange rate (U.S. Dollar Index), interest rate (Fed funds rate), GDP growth rate, CPI and major events, respectively.