This paper examines the combined role of momentum and term structure signals for the design of profitable trading strategies in commodity futures markets. With significant annualized alphas of 10.14% and 12.66%, respectively, the momentum and term structure strategies appear profitable when implemented individually. With an abnormal return of 21.02%, our double-sort strategy that exploits both momentum and term structure signals clearly outperforms the single-sort strategies. This double-sort strategy can additionally be utilized as a portfolio diversification tool. The abnormal performance of the combined portfolios cannot be explained by a lack of liquidity, data mining or transaction costs.
Commodity futures have become widespread investment vehicles among traditional and alternative asset managers. They are now commonly used for strategic and tactical asset allocations. The strategic appeal of commodity indices comes from their equity-like return, their inflation-hedging properties and their role for risk diversification (Bodie and Rosansky, 1980, Erb and Harvey, 2006, Gorton and Rouwenhorst, 2006, Chong and Miffre, 2010 and Baur and McDermott, 2010). Recent research has also established that commodity futures can be used to generate abnormal returns. For example, Erb and Harvey (2006) exploit the term structure signals of 12 commodities and implement a simple long-short strategy that buys the six most backwardated commodities and shorts the six most contangoed commodities. In a similar vein, Erb and Harvey, 2006 and Miffre and Rallis, 2007 follow momentum signals and tactically allocate wealth towards the best performing commodities and away from the worst performing ones. These simple active strategies have been shown to be capable of generating attractive returns.1
This paper digs deeper into the tactical opportunities of commodity futures by introducing an active double-sort strategy that combines momentum and term structure signals. This novel strategy aims at consistently buying the backwardated winners whose prices are expected to appreciate, and shorting the contangoed losers whose prices are expected to depreciate. While doing this, we expand on the term structure-only (hereafter, TS-only) strategy of Erb and Harvey (2006) by assessing the sensitivity of the TS profits to the roll-return definition, the frequency of rebalancing of the long-short portfolios and the date of portfolio formation. We also provide an in-depth analysis of the risk, performance and trading costs of the single-sort (momentum-only and TS-only) and double-sort portfolios.
Three contributions to the empirical literature on commodity futures markets are worth noting. First, we show that combining the momentum and term structure signals enhances the abnormal performance of either of the individual single-sort strategies. On a yearly basis, while the profitable momentum-only and TS-only strategies earn on average an abnormal return of 10.14% and 12.66%, respectively, the combined double-sort strategies, with an average annualized alpha of 21.02%, clearly provide the best signal on which to allocate wealth. A robustness analysis suggests that the superior profits of the double-sort strategies are not an artifact of lack of liquidity or data mining, and are robust to alternative specifications of the risk-return relationship. They are also robust to the high level of volatility experienced since January 2007. Second, the new commodity-based relative-strength portfolios emerge as excellent candidates for inclusion in well-diversified portfolios given the very low correlations between their returns and those of traditional asset classes. Hence, commodity futures may be tactically added to the asset mix of institutional investors not exclusively to earn abnormal returns but also to diversify the total risk of their global equity and/or fixed-income portfolios. Third, the proposed double-sort strategies are implemented on a small cross-section of contracts that are cheap to trade, liquid and easy to sell short. Net of reasonable transaction costs, they still generate a yearly net alpha of 20.41% on average.
The article proceeds as follows. Section 2 presents the dataset. Sections 3 and 4 analyze the profits of the individual momentum strategies and term structure strategies, while Section 5 studies the performance of strategies that jointly exploit momentum and term structure signals. Section 6 provides robustness checks and Section 7 concludes.
