New approaches to the HHT analysis: An application to the stock market

Although financial data analysis has become one of the most critical parts of financial economics, the limitations and assumptions of existing statistical financial analysis methods are incapable of analyzing stock market price series due to nonstationary and nonlinear characteristics of financial price series. Hence, previous literature on financial data anlaysis has mainly focused on financial return series instead of prices.This study examines the dynamics of the S&P 500 and ISE 100 stock market price indices for the period 1991 - 2011 using the new frequency - time methods of Hilbert - Huang Transformation (HHT). The HHT method is specialy developed for analysing nonstationary and nonlinear data. The method first decomposes complicated data into a Intrinsic Mode Functions (IMF) based on local characteristic time scale of data. This process is adaptive and applicable to nonlinear and nonstationary data. Then Hilbert transformation of IMF gives instantaneous frequency and energy patterns of data. Instantaneous frequency and energy of stock market prices give velocity and strength of price changes. Final frequency - energy representation of data reveals the timeline of events resulting the significant energy or frequency change. Hence, detecting the events triggering the stock market movements enlights the motivations behind stock market dynamics.This study also modifies the HHT method using the main properties of price indices, which are exponential growth, occasional price jumps , and having periods with low price changes. In the light of these properties, three models,Keywords: Hilbert - Huang Transformation, financial time series, spectral analysis, stock market price analysis