Risk Managed Time Series Momentum

Sammanfattning: This paper aims to investigate the crashes of time series momentum and to explore a systematic approach that mitigates the crashes of this strategy. Similar to cross-sectional momentum, time series momentum is also prone to severe drawdowns subsequent of a market decline when market volatility is high, contemporaneous with market reversals. However, such crash risk and option-like behaviour appear to be statistically predictable. Based on the insight on momentum crashes, we construct a risk-managed time series momentum strategy (RTSMOM) through a dynamic loading on the basic time series momentum (BTSMOM) strategy using in-sample predictions of the strategy's return and volatility. Our findings demonstrate that RTSMOM has a lower crash risk as negative volatility, maximum drawdown, VaR and expected tail loss decrease. Furthermore, RTSMOM has a higher average return and a substantial increased Sortino ratio. These findings are robust in subsample, back-testing, and cross asset analysis.