A combination of shorter-term models and adaptive models, which can toggle to shorter timeframes, were helpful in the first, third and fourth quarters of 2020. In the first quarter, the models exited equities faster; in the third quarter they latched onto the dollar downtrend and reflationary commodities theme, and in the fourth quarter they picked up some perky agricultural trends in corn and wheat. The use of variable timeframes is different from a trend approach such as MACD (moving average convergence divergence): “It is very ambitious to try and slow down or speed up individual markets, and we have succeeded with some very robust and sophisticated techniques that are also transparent to implement. We are committed to a dynamic and pure trend approach, which captures trends and changes with markets as they move,” says Chief Research Strategist, Dr Kathryn Kaminski. 

Though time frames – and cross-sectional or asset class risk allocations – vary, AlphaSimplex nearly always aims to manage their portfolio-level risk. The strategy uses a strictly systematic process, although very occasionally events occur that have a sudden and wide-reaching impact on economic conditions (such as the 2011 tsunami in Japan) and that have the potential to disrupt markets before any quantitative model would be able to process the new information. In those cases, if the firm’s Chief Risk Officer and the Investment Committee as a whole believe the standard risk estimate does not fully or accurately capture the current risk of markets, the team may intervene to reduce overall strategy risk. However, no such interventions were made around the Covid crisis. It was the fastest crash and recovery seen in decades, which created potential for models both over-shooting and under-shooting volatility targets, but AlphaSimplex’s realised volatility remained very near the forecasted range: “The strategy stayed close to its volatility target throughout the year, thanks in part to model enhancements introduced in 2018 that incorporated faster and intraday data into volatility estimates. Some sorts of slower volatility estimation models would have resulted in an overshoot in volatility, and this is one factor explaining return dispersion between CTAs,” says Kaminski.

Objectives and crises versus corrections

In addition to recently outperforming other flat fee trend funds, AlphaSimplex has generated a higher Sharpe ratio than the SG Trend Index over the past decade, due to a mix of lower fees and systematic implementation, though AlphaSimplex has a relatively modest standalone Sharpe ratio target of around 0.5 through a full cycle. The aim is more about providing a diversification benefit that will sometimes, but not always, enhance a portfolio Sharpe. Indeed, apart from the usual dialing up and down of volatility targets in managed accounts, AlphaSimplex sees demand for customization tweaked towards a greater emphasis on tail risk protection, or “crisis alpha”. Kaminski coined this term and co-authored a book on it with Alex Greyserman (Trend Following with Managed Futures: The Search for Crisis Alpha). 

It is important to precisely define what is meant by “crisis alpha” and how a crisis should be distinguished from a correction. Going back to 1992, Kaminski defines a “correction” as an equity market drawdown of 15% or less over two months or less and a crisis as a larger or longer-lasting drawdown. AlphaSimplex research shows that the SG Trend Index has profited in seven of the ten worst quarters for the S&P 500 over the past ten years. On average, the SG Trend Index has done better during crises and the SG CTA Index, which includes managers with generally higher weightings in non-trend models, did better during corrections. Faster models have also done relatively well in a crisis situation but might still not capture a short sharp correction that is quickly reversed through “buying the dip”, which has itself become a meme. These episodes included 2018, which is often seen as a recent example of when CTAs failed to provide crisis or correction alpha but was also special in other ways: “The year saw two very short V-shaped recoveries in February and October, and it was also unusual in that larger markets performed much better for trend followers,” says Kaminski. In 2018, AlphaSimplex saw negative returns from all four major asset classes, which was also very atypical.

Flat fee trend followers, hedge fund replication and alternative risk premia

The CTA and related space is a diverse one where pure trend followers, predominant trend followers, hedge fund replication and alternative risk premia products can apply different strategies with different objectives. AlphaSimplex’s flat fee trend following strategy has been the firm’s flagship CTA strategy since the strategy’s inception in 2010. “In general, the firm’s ethos was partly about disintermediating hedge fund fees and providing best ideas to all clients with no unintended tracking error across the UCITS, US ‘40 Act, private fund and separately managed accounts,” says Kaminski. The investor base is 90% in North America and includes US pension funds. The strategy runs the equivalent of around $5.5 billion at moderate volatility targets, and estimates a strategy capacity of $10 billion. 

In contrast, some other managers’ flat fee trend followers were launched more recently than the AlphaSimplex strategy and apply somewhat different signals and models to a different investment universe. Like many flat fee trend followers, the AlphaSimplex strategy is pure trend, while other managers may be around 80% trend – and the two can perform differently through market regimes and cycles. AlphaSimplex’s research team has used statistical analysis to infer growing value and carry factor exposures in some CTAs. These and other non-trend exposures are sometimes branded as alternative risk premia strategies.

Flat fee trend-following strategies are sometimes a building block for multi-strategy hedge fund replication or alternative risk premia (ARP) offerings, both of which are broad churches spanning a variety of strategies and approaches. AlphaSimplex is also active in both areas, but they are distinctive strategies. “Though our trend strategy provides some insights and techniques that feed into its ARP, it is not the same as the trend component of our ARP offering. This is partly because our approach to ARP involves some non-linear exposures and has somewhat different performance and risk objectives and characteristics, which might sometimes underperform when trend does well,” says Kaminski.

Differentiators versus other trend followers 

AlphaSimplex’s Managed Futures strategy is a pure trend-following strategy, but it uses a mix of core, shorter term and adaptive models. “The philosophical focus is on more robust and simple measures that minimize over-parameterisation. The different groups of models are different ways to measure trends. For instance, shorter term models are more about overcoming noise in shorter term signals,” says Kaminski.

This mix of models is driven partly by the firm’s investment philosophy, the Adaptive Markets Hypothesis. Over time, the adaptive weight has grown: “The capital allocated to the adaptive models has grown over time as we have gained confidence out of sample. This is a key differentiator, as is robustifying allocations between the models. Machine learning and non-linear techniques have been used from the start and have steadily expanded to a 40% allocation embracing broader and more complex approaches and techniques. The non-linear techniques include kernel regressions, scenario-based analysis and random forests,” says Kaminski.

As well as toggling between time frames, AlphaSimplex is relatively dynamic in moving around asset classes according to the strength of signals: “In 2019 we had very little in commodities because there was too much noise in the signals. In the second half of 2020, we had a very low fixed income weighting. Trend opportunities are clumpy and episodic,” says Kaminski.

Trend signals can also be somewhat different between asset classes. They are derived mainly from markets’ own price histories, but include some cross-sectional signals, particularly in currencies, which trade against one another rather than in isolation. “Different dynamics are needed to understand cross-asset relationships in currencies. For other asset classes, models are mainly based on a market’s own price history but do incorporate some cross asset information on clustering, global trends and relationships between asset classes,” says Kaminski.

Though AlphaSimplex could often have higher or lower allocations to different asset classes, markets or time frames versus other CTAs, it is not deliberately trying to deviate from the broader trend following universe and will sometimes be heavily exposed to the most popular markets and timeframes: “Crowding is a controversial topic in trend following. It is a double-edged sword. Trends are by definition a crowd, but sometimes reversals in crowded trades will hurt. There are also different ways to measure crowdedness,” says Kaminski. AlphaSimplex research finds that the performance of crowded trends varies between and within asset classes and over time periods and is ultimately agnostic on whether the trends it trades are crowded.

Ongoing strategy evolution and research

The AlphaSimplex strategy’s investment universe has grown from around 60 markets to over 90 over the past ten years. The Chinese equity index is an example of a recent addition. Markets have occasionally been discontinued if they migrated onto another exchange or for structural reasons; LIBOR will be deleted soon. AlphaSimplex is open minded about new markets that meet the right criteria, including volume and liquidity. Bubbles such as the seventeenth century Dutch tulip mania could have been profitably traded by trend followers, Kaminski pointed out in her book. “Today, cryptocurrencies are a candidate but there are some issues. Volume is improving but is not yet large enough, and collateralization is not yet there in size. Additionally, some investors have ESG concerns about electricity generation using fossil fuels,” Kaminski points out. AlphaSimplex recently signed up to the UNPRI and is collaborating with corporate parent Natixis to explore how ESG might be applied to a CTA strategy. “These insights will take time to work out. ESG is more advanced for single name equities and companies. It is not as cut-and-dried for futures,” says Kaminski. AlphaSimplex is also working with Natixis to research Chinese commodities, where the issues include infrastructure and access. 

Diverse thinking informs the research agenda: “We have an interesting research culture that is very open and accepting. Being based in Boston and near Cambridge, Massachusetts helps us to attract great staff, recruiting from Harvard, MIT and other local universities. There is a strong focus on inclusivity and diversity at all levels, with one third of the team being female, which is rare for a CTA,” says Kaminski, who has been featured in The Hedge Fund Journal’s 50 Leading Women in Hedge Funds report and who recently won the Systematic Leader of the Year award from Women in Asset Management. 

The pace of innovation has been steady rather than accelerating. Internally, AlphaSimplex carries out “walk forward” and “walk backward” exercises comparing previous and current versions of the models. “These reveal that the latest models have produced some moderate improvements in terms of better performance, been more accurate at estimating volatility and robustified allocations and breadth of models. Equally they remain parsimonious models,” says Kaminski. The current research agenda is continuing to hone and refine areas that have been developing for many years: “Future research is emphasizing new techniques in machine learning and other adaptive models, shorter term factor models and volatility estimates. Different methods are applied to gauging shorter term trends in rapidly changing environments,” says Kaminski. 

Macroeconomic climate for trend following 

Macroeconomic trends could be on the verge of a sea change towards higher inflation and higher interest rates (already being seen in rate rises in some emerging markets), according to some economists. As a systematic manager, AlphaSimplex does not have a fundamental discretionary house view, but has researched how its strategies might have performed in different economic climates. This research predates the near 40-year bond bull market and low inflation regime that began in 1981 and considers how trend-following strategies might have fared not just in the 1960s and 1970s but also in previous centuries. 

Kaminski and Greyserman’s historical analysis of trend followers over the centuries suggests that they would have performed better under inflationary climates. A future episode of inflation, which might be associated with Modern Monetary Theory, could also be different from historical ones which generally saw interest rates rise as well. Kaminski has also researched how CTAs performed under rising interest rate climates. CTAs in general have already capitalized on the reflation trade in 2020 and 2021, while AlphaSimplex’s shorter term models have been nimble enough to cut and reverse from long to short bonds exposure in 2021. The firm’s systematic approach to trend following may allow it to weather potential macroeconomic changes and to provide crisis alpha and diversification to a traditional portfolio.