The trading of commodities futures dates to at least the 1850s, and commodity trading advisers represent the second-biggest hedge fund category. Yet meaningful performance measurement of CTAs and adequate benchmarks remain elusive.
CTAs are well known for their reluctance to disclose trades in detail — leading to the "black box" designation — so it is difficult to create benchmarks for them that are based on the underlying assets. As a result, investors have tended to benchmark CTA strategies against their peers. This is a relatively blunt instrument because peer benchmarking does not reveal or mimic the underlying drivers of CTA returns.
A new study, "Trends' Signal Strength and the Performance of CTAs," published in the CFA Institute Financial Analysts Journal, describes a potential solution to this issue. The study proposes a model that better replicates the performance of CTAs than a peer-group benchmark and enables investors to better select skilled managers.
Toward a better CTA benchmark
The model, which the authors Gert Elaut and Peter Erdos call "adaptive time series momentum," teases out the underlying assets that trend-following managers use. The authors mined a large amount of short-term and longer-term signals from a long series (1990 to 2015) of futures data. These signals were then compared with monthly returns from CTA funds in the BarclayHedge database.
The resulting performance measurement model is innovative, the authors argue, because drilling down to the asset level allows us to measure the strength of momentum signals from the prices of assets.
In short, the authors create a performance measurement model that reflects the ways trend-following managers actually invest. That is, they tend to load up on developing trends and underweight trends as they start to fade. The model consequently mimics CTA performance, allowing identification of skilled CTA managers among their peers.
How it's done
As is well-known, a momentum signal in a CTA strategy typically occurs when the price of an asset moves above or below a moving average.
However, by analyzing a large number of signals over a large number of time horizons, it should be possible to assess the strength of the signal rather than simply the direction of the signal, the authors argue. Capturing the strength of the signal is key to the ATSMOM strategy, which, the authors believe, is superior to the simpler time series momentum strategy. Adaptive time series momentum, which relies solely on binary long and short signals, traditionally has been the focus of attempts to measure and mimic performance in the CTA industry.
The ATSMOM model looks across numerous asset classes, including commodities, equities, fixed income and foreign exchange. Critically, by focusing on the strength of signals, it allocates disproportionately to futures contracts that display clear trends.
Revealing the contents of the black box
The model uncovered evidence about the CTA industry that could be revelatory for investors.
The study's focus on the underlying assets used by CTAs sheds valuable light on managers' allocation decisions. For instance, smaller CTAs allocate evenly across asset classes, but larger funds overweight the more liquid futures markets, such as fixed income.
Indeed, the research found that CTAs tend to be exposed to fixed income most frequently, with exposure to fixed income being significant for 70% of the funds. Commodities, being significant in 64% of cases, made up the second most important exposure.
The authors further discovered that fund size was negatively correlated with superior performance, whereas the age of a fund is positively correlated.
Fund style also appears to be a factor. For example, funds that engage in pure trend-following approaches or higher equity momentum tend to generate superior performance.
Meanwhile, the negligible impact of fund flows on this extra-benchmark performance suggests capacity constraints are not a big issue for CTA funds and their investors.
Funds that charge higher management and performance fees might be expected to achieve higher returns. However, the study debunks this, at least for CTA strategies.
Most importantly, perhaps, the authors appear to have created a model that can improve benchmarking in the CTA industry.
The study found that the adaptive time series momentum model is better at explaining CTA performance than other well-known models and benchmarks because it allocates more to higher-performing assets and limits drawdowns by more smoothly allocating less to lower-performing assets.
The speed factor, which was identified in previous research, is the excess return found in funds that buy longer-horizon (slower to react) momentum CTA strategies and selling shorter-horizon ones. The speed factor has been the driving force behind a great many CTA strategies. If this speed factor is combined with the ATSMOM strategy, the combination can be seen to be responsible for 40% of the variation in individual CTA returns, the study showed.
This, however, leaves 60% of their performance still unexplained. There are a large number of funds that display considerable alpha relative to the model. In fact, funds with positive alphas generated mean alphas of 4.8% a year. These positive alphas, the authors surmise, can reasonably be attributed to individual manager skill.
What do we conclude? The study reveals many characteristics and return drivers of CTAs that have been obscure. It shows that a proposed ATSMOM model, when combined with the speed factor, is better at explaining CTA returns than existing benchmarks. This key finding should help investors to better understand CTA strategies and improve CTA manager selection.
Heidi Raubenheimer is managing editor at the CFA Institute Financial Analysts Journal, Charlottesville, Va. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines, but is not a product of P&I's editorial team.