Non-efficiency in Automated Markets for the US and Mexico

Abstract

This work analyzes the Efficient Markets Hypothesis (EMH) introduced by Fama (1970). The notion has been a cornerstone in theory and practice in financial markets, given its implications for predicting future prices. The efficiency of markets has been widely explored in the literature. Still, recent technological advances have made it possible to trade faster than ever. Hence, the question arises again to examine the efficiency of high-frequency markets, which has not yet been investigated in depth. The present work explores market efficiency at various frequencies, from 1 second to 10 days, in the United States (US) and Mexico’s stock markets. The empirical distribution and correlation of the assets return series were modeled to try and gauge whether markets follow a random walk. Thus, one can conclude that information is instantly incorporated, and the market is efficient. Results show that the markets are not efficient for high frequencies, but above the 10-day threshold, these series follow a random walk with an asymptotically normal distribution. The conclusion is important for practitioners and academics since it suggests the possibility of forecasting prices in high frequencies using statistical tools.