DETRENDED FLUCTUATION ANALYSIS OF THE US STOCK MARKET

Abstract

This paper extends the work in [Serletis & Shintani, 2003; Elder & Serletis, 2007; Koustas et al.; Hinich & Serletis, 2008] by re-examining the empirical evidence for random walk type behavior in the US stock market. In doing so, it uses daily data on the Dow Jones Industrial Average, over the period from January 3, 1928 to March 15, 2006, and a statistical-physical approach — "detrended fluctuations analysis" — providing a reliable framework for testing the information efficiency in financial markets as shown by Uritskaya [2005a, 2005b] and Uritskaya and Uritsky [2001]. The approach eliminates nonstationary market trends and focuses on the intrinsic correlation structure of stock market fluctuations at different time scales which is studied relative to random walks models. Our results indicate that the US stock market operates close to the state predicted by the efficient markets hypothesis. The observed transient deviations from this state are shown to have a statistical origin, consistent with a purely random geometric Brownian motion.