Blog - Investing Notes

August 7, 2008 - Combining Momentum and Roll Return Signals for Commodity Futures

Does combining two commodity futures trading signals shown to be effective in prior research, momentum and roll return (term structure), improve on both? In the May 2008 version of their paper entitled "Tactical Allocation in Commodity Futures Markets: Combining Momentum and Term Structure Signals", Ana-Maria Fuertes, Joëlle Miffre and Georgios Rallis measure the combined value of momentum and roll return signals in the design of commodity futures trading strategies. They test combinations that iteratively buy backwardated (positive roll return) winners and short contangoed (negative roll return) losers. Using daily closing prices on the nearby, second nearby and distant contracts for 37 commodities as available over the period January 1979 through January 2007, they find that:

• The momentum (roll return) strategies that are significantly profitable earn an average risk-adjusted annual return of 10.14% (12.66%) before transaction costs, compared to just 2.48% for a passive long-only portfolio that equally weights all 37 commodities.
  • Trend following rules with formation periods of one, three and 12 months and a holding period of one month (Mom1-1, Mom3-1 and Mom12-1) are the best momentum strategies.
  • The best roll return strategy (TS1) buys the 20% of commodities with the most positive roll returns and shorts the 20% with the most negative roll returns each month.
• Double-sort strategies that exploit both momentum (Mom1-1, Mom3-1 or Mom12-1) and roll return (TS1) signals with monthly holding/rebalancing generate an average risk-adjusted annual return of 21.02% before transaction costs. The most profitable double-sort strategy is TS1 combined with Mom1-1 (see the chart below).
• These double-sort strategies involve a small subset of contracts that are liquid and cheap to trade, such that they offer an average risk-adjusted annual return of 20.21% after reasonable transaction costs.
• The abnormal returns of the double-sort strategies are not a result of data mining.
• Returns from the double-sort strategies have very low correlations with those of traditional asset classes and are therefore promising as portfolio diversifiers.

The following chart, taken from the paper, plots the value of a $1.00 initial investment in TS1 - Mom1-1 combined, Mom1-1 only, TS1 only and the passive long-only benchmark over the entire sample period. It illustrates both the strong performance and high risk of the active strategies. The outperformance of the combined TS1 - Mom1-1 strategy appears to be driven by momentum until 1998 and by roll return since.

In summary, commodity futures trading strategies that combine momentum and roll return may offer strong performance largely uncorrelated with those of stocks and bonds.

For related research, see Blog Synthesis: Investing/Trading in Commodities and Commodity Futures.