AnotherAspect: What was trendy in 2019?

Last year was an overwhelmingly positive year for traditional asset classes. Alternative strategies such as Managed Futures also generated strong returns. Whether we look at the SG Trend, SG CTA, BTop50 or CISDM CTA indices, 2019 was the best year for the industry since the commodity bear market in 2014. Yet it was not all plain sailing. Last year presented numerous challenging macro-economic conditions as well as a good amount of geo-political uncertainty. Only looking at headline indices, one can miss important details. 2019 was one of those years where the detail matters.

Figure 1 shows the dispersion of CTA manager returns through time. The selection of managers here includes the most significant 15-20 names that over the last 20 years have been Aspect Diversified’s peers. At any point in time these managers would have been members of either the SG CTA or the BTop 50 indices but we excluded those that were either explicitly Global Macro or Short-Term trading in nature. It is also worth mentioning that we include managers that are no longer around but were significant players in the space in the past and would have formed part of the available universe of CTAs for large allocators.

Really, 2019 was a good year ‘on average’ for CTAs, at a time when many commentators are talking about the commoditisation of trend following, it’s noteworthy that the inter-quartile range of return dispersion was the widest since 2008 and second widest in our 21 year history.

What was it about last year that resulted in such variable performance across the CTA industry?

As ever, there is no one answer: differences in market allocations, trend-following speeds, risk management and portfolio construction choices can play a big role. We cannot account for all these effects but we want to highlight some interesting observations by looking at the trend-following opportunity set by speed and asset class. Let’s first remind ourselves how the major asset classes behaved during 2019. We proxy these asset classes with broad based indices such as the MSCI All Country World Index as a proxy for Developed and Emerging equity markets, the Barclays Global Aggregate Bond Index as a proxy for global bonds, The Dollar Index as a proxy for FX and the Bloomberg Commodity Index as a proxy for commodities such as energies, metals and agriculturals.

Looking at Figure 2, one might easily infer that this was a very favourable environment for trend followers. One asset class was up ~25% on the year, two were up between 5-10% and, at worst one was basically flat. No wonder the CTA industry had one of its best years in a while… But, didn’t we say earlier that last year had quite a few challenging periods and led to wide dispersion in the performance of many CTAs? What’s going on? Let’s look at the simulated attribution from trend following these asset classes across a wide range of time frames. The results below show the application of some generic trend following models across approximately 200 markets spanning the four major asset classes: Bonds, Commodities, FX and Stock Indices. The time frames are looking for effects between 6-9 months (slower), 2-3 months (medium) and 1-2 weeks (faster). All the individual timeframes are simulated independently and are set to target 15% volatility.

Figures 2 and 3 seem counter-intuitive: regardless of speed, the only profitable opportunities appear to have been in bond markets.

Let’s try to make sense of this by looking at the same asset class data but visualise them a bit differently. By including extra price history we remind ourselves that trend-following strategies are highly path-dependent and a trend is only a trend if it’s been in place for some time – therefore looking at data only for the period we are analysing misses out vital information that is needed to understand strategy positioning.

Looking at Figure 4 we see that the so-called ‘risk assets’ such as equities and commodities reversed sharply at the beginning of 2019 after declining significantly in Q4 2018. The commodity rally faded at the end of April and spent the rest of the year predominantly trading within a 5% range. Equity markets interrupted their recovery with some sharp sell-offs in May and August and only persistently trended higher during Q4. Currencies were indeed mostly sideways and the only persistent multi-month trend was in bonds – but even that asset class became range-bound towards the end of 2019.

Rather than looking at summary statistics as in Figure 3, the dynamics of the various trend-following speed simulations can be very informative:

Currencies were generally rangebound but certainly seemed a lot choppier the faster they were traded.

Trading commodities quickly seems to have been the better choice during 2019 but even then, there were no profitable opportunities over the period. By trading quickly at the beginning of the year, strategies would have captured the early rally, while slower time frames would have been the wrong side of the trade for several months. By the end of April we note a near 10% potential differential between trend following commodities quickly and the other two speeds. After the reversals in May, commodity markets kept being affected by the ongoing trade war rhetoric resulting in losses across all time frames.

The stock indices conundrum: so significant were the market declines in Q4 2018, and so sharp the rebound in equities during Q1 2019, that only by trading quickly would a trend-following strategy have been able to participate in the equity market recovery in the early part of last year. However, the choppy nature of equity markets from May to September ensured that the early strong gains were ‘given back’ for the remainder of the year. At the other end of the spectrum, trading slowly meant that the early rebound was missed and it took the slow trend model until September to begin to recoup losses and end up flat on the year. Path-dependency is clearly evident in this example and looking at the medium time frame reinforces the point. Not only was it too slow to capture the rally at the beginning of the year but it was also quick enough to get caught out by the sharp reversals in May and early August.

Finally, we see that trend models on bonds were far more successful during 2019 but the slower and medium time frames outperformed the faster one. Quicker time frames had more opportunities to get whipsawed during the fixed income-to-stocks rotation during September and October.

Conclusion

Without needing to know the finer details between different managers’ attributes across the Managed Futures space we have identified a number of distinctly challenging conditions across the asset classes traded by CTAs. Differences in allocations to asset classes and to trend following speeds alone can account for large parts of the dispersion we observed.

Aspect Diversified’s strong performance during 2019 can be attributed in part to being able to successfully capture fixed income trends but importantly also to being able to navigate the stop-start nature of trends in the other asset classes through its dynamic risk management framework. There were many sharp reversals throughout the year, particularly in January when equities bounced back, in May and August when commodities and equities fell sharply and during September and October when there was a factor rotation from growth to value within equities and from risk-off to risk-on globally. In these scenarios, portfolio construction and risk management are very important differentiators.

Finally, most CTAs, even the trend-heavy ones, generally include non-trend models in their strategy mix and Aspect Diversified’s non-trend models also made significant gains particularly in challenging asset classes like equities and FX.