AnotherAspect: What Was Trendy in 2020?

Examining the liquid alternatives landscape in a turbulent year

In this latest instalment of "What Was Trendy?", we ask: which models were most in vogue in 2020 and what was it about their style that made them turn heads all the way from London to New York to Tokyo?

Over the years, research has led us to create hundreds of models which try to harness globally investable effects across multiple asset classes, sectors and timeframes. They are 100% computer-driven and use evolving technology such as machine learning and forward-looking alternative data. In this piece we review how those different quant investment styles performed over the last 12 months through Aspect’s lens.

As we look ahead to 2021, there is widespread belief that expected returns on traditional investments are likely to be lower for longer, particularly in the case of ultra-low bond yields. This makes the case for liquid alternatives more compelling and that is exactly the area in which our models live. Each model seeks to capture effects found in liquid assets and we group them into themes based on common premises (see table below). These themes also help to explain the extent of trendiness in 2020 alongside other groupings such as speed and sector focus.

Around 150 multi-asset derivatives models (almost all of which trade real client money as part of Aspect’s programmes) were scaled to target 1% annualised volatility and their 2020 returns (gross of transaction costs and fees) were analysed. Fast, medium, and slow speeds denote holding periods of <1 month, 1-6 months and >6 months, respectively. Where possible, we delineate sectors to the lowest level of granularity found within the model; for example, the commodities sector models may trade any combination of agriculturals, energies or metals whilst the metals sector models would strictly only trade metals.

In 2020, average model profitability within and across all sectors was positive, except credit. The following heatmap illustrates sector (model count in parentheses) profitability by speed (green is positive, white is flat, red is negative and grey means not applicable):

Overwhelmingly, the order of profitability by speed was fast, medium, and then slow in all sectors except:

• Energies, where slow carry models did well to short heavily contangoed markets
• Fixed Income, where slower-moving relative plays on short-term interest rates did well
• Volatility, where medium-term harvesting of volatility futures risk premia outperformed quicker options-based models due in part to inclusion of agile conditioning variables

2020 was a year where markets broke records by moving so quickly from risk-on to risk-off and back again, and this meant that more nimble trading strategies were often required to keep pace.

All model types had a positive 2020 on average. Relative (cross-sectional) strategies generally had a better 2020 than directional because Aspect’s brand of relative strategies trade faster. Blend models are not explicitly directional or relative, and they fared similarly to directional strategies. The weakest area was medium-term directional, where the effects of rapidly gyrating markets were not hedged out. One of the strongest areas was in faster directional strategies, which were agile enough to avoid being excessively whipsawed. The recurrent theme of speed may be reflected in the information ratios of industry benchmark indices: On top was SG Short Term with 0.82, it tracks managers of all types with holding periods of less than 2 weeks. Second came SG Trend: 0.76, which is mostly directional trend-following managers with fast and medium-term outlooks. Finally, SG CTA: 0.45, which has an assortment of directional and relative model types and operates over similar speeds to SG Trend as it contains all its managers and also contains some SG Short Term names.

Let us highlight some notable outcomes from 2020 by focusing on average model profitability by timeframe, sector and finally investment theme. Please note that for timeseries charts, the mean monthly return by grouping is shown and monthly totals are not expected to be the same across groupings.

It is not surprising that slow models struggled in the first quarter, given that March saw the sharpest crashes and subsequent rallies in living history. As the severity of the pandemic dawned on the globe, the S&P 500 broke records, declining over 30% in little over a month before gaining most of it back before the end of the second quarter.

Fast models navigated each quarter well, including the second quarter which also saw decent gains from medium and slow models. Perhaps risk-on positions which caught the slow models off guard in Q1 were somewhat vindicated by a risk-on Q2, supported by historic central bank liquidity provision. Steady risk appetite throughout Q2 helped medium-term models prevail in May before rotations occurred in June as virus cases resurged.

Q3 witnessed large sudden moves in commodities against established patterns, which proved tricky for slow models. Positive vaccine developments and US election results preceded cheerfulness in markets in Q4. This led to a significant factor rotation in November, at which point fast models again came out on top. December’s geopolitics removed some market uncertainty and all speeds, sectors and themes were positive.

So, trade swiftly forever more, I hear you say. Not so fast! Just take a look at the same set of models in 2019:

It is also important to remember that transaction costs and capacity become more of an issue as trading speeds increase, albeit our “fast” models sit in a speed range where such considerations do not invalidate the takeaways from this analysis. Overall, we still remain diversified across speeds.

2. Sector Focus

The chart below shows that there were nicely diversified sources of returns from sectors. The bulk of the noticeable credit sector losses came in the last week of February, when corporate borrowing almost came to a standstill amid piercing upsurges in credit spreads. The volatility sector was caught on the wrong side of historic spikes in risk aversion but recouped those as the year went on, amid elevated volatility gauges in riskier assets as many markets melted up. Interestingly, it was the same medium-term energy models which foundered in August that came back strongly in December. The big weather-driven moves in natural gas had something to do with that!

3. Theme Focus

The chart below illustrates how much monthly variation there was amongst investment themes. Few investment themes achieved gains as consistently as ‘Flows’ and it earnt the top spot in 2020. The expansion of central bank balance sheets by trillions of dollars cascaded through the financial system. This generated large differentials in fund flows over a short period of time, to the benefit of our flow-based models. Central bank largesse was also the primary driver for interest rate movements across tenors and these dynamics provided lots of opportunities for valuable flattener and steepener trades in global fixed income markets. These models came from the ‘Slope/Curve’ theme which had the second highest average profitability. The ‘Value’ theme finished third and with it came the dichotomy between relative against directional and fast against slow. Quicker relative value models which calculated mispricing using exogenous datasets outperformed in almost all sectors, particularly in March. As the year went on, ample liquidity and hints of economic recovery appeared to boost previously unloved market segments in the form of factor rotations.

’Economic Fundamentals’ was the only theme to finish negative on average. There was an apparent disconnect between risk asset strength and economic woes. Unconventional market moves reflected exceptional policy support and a reduction in the risk that COVID-19 would cause worldwide collapse. Whilst forward-looking alternative data models managed to predict inflationary pressures, stock indices defied deep recessions and job losses. Major indices have become less reflective of the real economy. For instance, hard-hit hospitality and leisure industries employ around a tenth of the US workforce but only have a 2% weight in the S&P 500.

4. Conclusion

What are some things worth remembering about liquid alternatives based on 2020?

• The order of trendiness by speed appeared to be fast, medium, then slow in 2020. However, just like the fashion industry, previous years were markedly different and every year is unpredictable. Spreading risk across time horizons is our preferred approach as it can expand the opportunity set.
• Most themes had a positive year. In particular, models based on shorter-term market flows and yield curve plays, and it was good to see our old friend "value" rather revitalised – at least in our formulation. Eventful years tend to bring to light a whole host of factors that drive markets. Risk-adjusted returns can be maximised by combining as many of these factors as possible alongside thoughtful portfolio construction and risk management.
• There were profitable models in most sectors. Many asset classes were ripe for picking, as outsized moves popped up far and wide. This reinforces how useful it is to disperse risk across sectors in addition to diversification across speeds and themes.

Looking forward to this year, questions are focused on whether the global economy can withstand various pandemic stresses with or without stimulus-laden intervention, whilst negotiating the spectre of inflation. Can markets be navigated successfully by combining our strong, diverse set of predictive models into a performant, dynamic and risk-managed ensemble? Now that would make for a trendy 2021!