The authors have two main goals. The first is to analyze whether allocating to commodity trading advisers (CTAs) results in better risk-adjusted returns for an overall portfolio. The second is to analyze whether CTAs generate improved risk-adjusted returns on a standalone basis.

Allocating to managed futures funds benefits an overall portfolio, according to the authors. Allocating to CTAs improves diversification benefits because of low or even, at times, negative correlations both among CTAs and between CTAs and other asset classes. The authors observe a median pairwise correlation coefficient of 0.08 and negative correlations among 36.6% of CTAs. Meanwhile, they obtain correlations between their CTA universe and equities, commodities, and bonds of 0.03, 0.07, and 0.08, respectively. In addition, CTAs perform well during adverse market environments and thus provide a tail risk hedge for an overall portfolio. Importantly, the authors show that the optimal weight of CTAs within an optimal portfolio is strictly positive.

In contrast to the favorable findings with respect to their first goal, the authors find that in aggregate, CTAs do not generate superior risk-adjusted returns on a standalone basis. In addition, the authors find no evidence of positive skewness, contrary to received wisdom. In fact, the authors observe return distributions that are leptokurtic (more concentrated about the mean) for the majority of CTAs. They construct a trend-following multi-factor model to serve as a CTA benchmark in order to identify the return sources of CTAs and test for alpha. The authors find that most CTAs show statistically insignificant alphas when regressed against this model, but they find strong performance dispersion, with top-quartile CTAs generating statistically significant outperformance.

These findings are relevant for investors considering allocating to CTAs and suggest that practitioners should focus on manager selection and/or lower-cost alternatives to CTAs.

The authors account for survivorship bias and address backfill bias. Backfill bias occurs when funds newly entering a database are allowed to include returns generated prior to the start of reporting to the database. This approach could result in an upward bias: Funds with poor performance prior to reporting to the database have no incentive to reveal those returns, whereas those that have performed well do. To avoid this bias, the authors exclude data from before the fund starts reporting to the database.

Following data cleaning, the authors are left with monthly performance data from 1971 to 2016 that includes 1,558 CTAs, with an average track record of 95 months. A broad spectrum of performance measures, such as the Sharpe and Calmar ratios, is used to study the dataset’s statistical characteristics.

Using standard portfolio optimization techniques, the authors find that adding managed futures to a hypothetical portfolio of equities, bonds, and commodities fixed in a ratio of 55/35/10 can raise annualized return to as high as 11% (versus 8.76% without managed futures) and can reduce annualized volatility to as low as 8.66% (versus 9.63%). The authors allow for a maximum allocation of 20% to managed futures in their overall portfolio.

A benchmark is created to examine whether CTAs provide alpha. The benchmark is constructed using a simplified trend-following strategy based on time-series momentum. More sophisticated features are gradually added to test whether they improve the benchmark’s explanatory power.

At a time when many investors believe that the long-term return outlook for equities, bonds, and commodities is poor, interest in alternative investments — an asset class that includes CTAs — is high.

The main usefulness of the authors’ work is the evidence showing that adding CTAs to portfolios can increase return and reduce risk and that CTAs in aggregate do not provide risk-adjusted outperformance on a standalone basis.

Research already exists on the diversification benefits and risk–return profile of CTAs, but most of it has been conducted on data samples that do not include the years following the 2008 global financial crisis. The inclusion of the crisis period in the author’s dataset makes this research particularly noteworthy given the perceived benefit that CTAs offer during times of market stress.

In addition, the authors conduct their research using a dataset from three sources, which improves on previous CTA research. This dataset is not limited to a single data vendor and thus does not suffer from lack of cross-data validation.

One possible limitation of the authors’ work is that hedge funds denominated in currencies other than US dollars are excluded from the analysis. Future research that examines CTAs denominated in other currencies could serve to make the conclusions more robust.