Trading signal, functional data analysis and time series momentum
Résumé
Prior empirical results show that the time series momentum portfolio outperformed the buy-and-hold benchmark well from 1985 to 2009, but this profitable pattern unexpectedly vanishes after 2009. In this paper, we reconstruct the time series momentum portfolio by applying new trading rules derived from the functional data analysis approaches. Using a dataset that contains 24 commodities from January 2010 to December 2018, our daily-based strategy documents an improvement in the Sharpe ratio of 0.75 compared to 0.07 in terms of the original time series momentum portfolio. This finding offers an alternative strategy for trend-following investors in the commodity futures market.