Decomposing Momentum Spread

42 Pages Posted: 14 Jun 2019 Last revised: 16 Sep 2019

See all articles by James Tengyu Guo

James Tengyu Guo

London School of Economics & Political Science, Department of Finance

Date Written: July 26, 2019

Abstract

Since momentum arbitrage activity, buying winners and selling losers, effectively enlarges the return spread between these two groups, I find that the momentum spread (the difference of the formation-period recent 6-month returns between winners and losers) negatively predicts future momentum profit in the long-term, but not in the following month. I further decompose the momentum spread into the spreads of old or young momentum stocks based on whether a stock has been identified as a momentum stock for more than three months. I show that the negative predictability is mainly driven by the old momentum spread. For the top 20% of the sample period associated with the highest values of old momentum spread, the momentum reversals happen sooner (only six months after formation) and stronger (more than 120 basis points per month from month 7 to month 24 after formation), relative to negligible momentum reversals observed following the bottom 20% period with low old momentum spread. As these old momentum stocks are more likely to be exploited by arbitrageurs, these findings suggest that momentum is amplified by arbitrage activity and excessive arbitrage destabilizes asset prices and generates strong reversals.

Keywords: Momentum, Return Spread, Underreaction and Overreaction, Destabilizing Mechanism of Excessive Arbitrage

JEL Classification: G12, G14

Suggested Citation

Guo, James Tengyu, Decomposing Momentum Spread (July 26, 2019). Available at SSRN: https://ssrn.com/abstract=3386828 or http://dx.doi.org/10.2139/ssrn.3386828

James Tengyu Guo (Contact Author)

London School of Economics & Political Science, Department of Finance ( email )

London
Great Britain

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
180
Abstract Views
1,042
rank
189,331
PlumX Metrics