Ride the trend: Is there spread momentum profit in the US commodity markets?

Some previous researchers have argued that trading strategies based on calendar spread time series momentum (STSM) can deliver significant returns (Szymanowska et al. 2014; Boons and Prado 2019), which, if true, is at odds with the efficient market hypothesis. These arguments however, do not exclude the unrealisable futures contract roll yield and are also affected by other empirical and statistical issues that may lead to misleading results. With more than 30 years of data, we investigate STSM in 22 US commodity futures markets. First, we assess whether past spread returns can predict future returns, a necessary condition for the existence of momentum. We find predictability to be very weak after correcting for the issues affecting prior research. Second, we implement STSM-based investment strategies. We compare STSM profits for individual markets and portfolios to profits generated by a simple long-only benchmark strategy that does not require any predictability. STSM does not generate returns statistically different from the benchmark trading strategy, with both strategies generating very low or negative returns. For the momentum to outperform the benchmark strategy, predictability should be three times larger than observed from real data, but would entail substantial downside risk. In sum, the empirical evidence indicates that returns from STSM-type strategies are illusive for the commodities and period studied. Our results strongly suggest that inclusion of unrealisable roll yield generates the illusion of profitable STSM trading strategies in previous research.