How Trend-Following and Mean Reversion Affect Long-Run Price Variance

Summary

This work studies how aggregate trend-following or mean-reverting trading behavior can alter asset-price dynamics. It models the direction of trading explicitly and links those styles to serial correlation: trend-following behavior can create positive correlation, while mean reversion can create negative correlation. The models are used to examine how these correlations affect long-run variance relative to a random-walk benchmark.

The theoretical analysis implies that trend-following raises long-run variance, whereas mean-reverting behavior can lower it. Applications to large U.S. stocks and high-frequency EUR/USD data show greater predictability than under a random-walk assumption. The supplied description does not give the model specifications, sample periods, or measures of predictive improvement, and it does not demonstrate that a profitable trading strategy follows from the predictability. The evidence is therefore best treated as a framework for understanding how trading styles may shape return dependence and risk.

Key ideas

• Trend-following and mean-reverting trading styles can induce positive and negative price correlations, respectively.
• The authors use probabilistic models that represent the direction of trading explicitly.
• Trend-following increases long-run variance relative to a random-walk model in the theoretical results.
• Mean-reverting conditions can reduce long-run variance relative to the random-walk case.
• Applications to large U.S. stocks and high-frequency EUR/USD data report increased predictability.

Tags

# Trading styles and long-run variance of asset prices


# Trading styles and long-run variance of asset prices









Trading styles can be classified into either trend-following or mean-reverting. If the net trading style is trend-following the traded asset is more likely to move in the same direction it moved previously (the opposite is true if the net style is mean-reverting). The result of this is to introduce positive (or negative) correlations into the time series. We here explore the effect of these correlations on the long-run variance of the series through probabilistic models designed to explicitly capture the direction of trading. Our theoretical insights suggests that relative to random walk models of asset prices the long-run variance is increased under trend-following strategies and can actually be reduced under mean-reversal conditions. We apply these models to some of the largest US stocks by market capitalisation as well as high-frequency EUR/USD data and show that in both these settings, the ability to predict the asset price is generally increased relative to a random walk.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.