Trend, Mean-Reversion or Random Walk? A Statistical Analysis of Price Behavior in Major Markets

We examine the price behavior of 56 major markets over the last 16 years applying a set of univariate and multivariate robust statistical tests across different time frequencies. Our results can be considered as an augmented true out-of-sample test of all previous research testing for time-series independence. We find no statistically significant evidence that price movements in calendar time scale consistently deviate from randomness. There are only limited departures that are split between trend and mean-reversion depending on the time frame, prices (spot or futures) and assets studied, but even those cases have to be interpreted with caution since we tested a large number of assets increasing the probability of Type I error. Our univariate results are also confirmed by the multivariate analysis that examines the joint distribution of assets using a pooled panel regression. Finally, comparing our results to previous research we find evidence that price dynamics change over time highlighting the difficulty of identifying price patterns ex-ante