Time-series momentum in nearly 100 years of stock returns

Abstract

We document strong time-series momentum effects in individual stocks in the US markets from 1927 to 2017. Time-series momentum is not specific to sub-periods, firm sizes, formation- and holding-period lengths, or geographic markets. The effects persist after controlling for standard risk factors. Time-series momentum effects are conditional on the market state, the information discreteness of the constituent stocks and investor sentiment. We propose two alternative implementations, revised time-series momentum and dual momentum, which generate even higher profits than standard time-series momentum.