A New Climate for Systematic Investing?

Stockholm (HedgeNordic) – No one can really know what will be the long-term consequences of the recent market and economic turmoil. Aspect Capital holds the view that the next decade could be much tougher than the last one for investors who have recently enjoyed historically extraordinary returns from the very simple strategy of buying and holding equities and bonds on a long only basis.

The firm has authored an insight paper entitled “The Return of Stagflation: Post-Pandemic Implications for Asset Owners”. One potential game changer is that broken supply chains may reverse the megatrend of peak globalization. A return to 1970s style stagflation could result in negative returns for both equities and bonds, and create a more promising climate for various systematic strategies.

Yet Aspect is also open minded about the possibility that some systematic strategies might also diverge from recent history. “We are rethinking whether changes in human behaviour will mean that the heuristics of the past 30 years, such as “buying the dip, will still be valid”, says Razvan Remsing, Director of Investment Solutions, (pictured) who joined Aspect nearly 10years ago and has participated in HedgeNordic’s roundtable on CTAs several times.

“The Coronavirus crisis is not a financial crisis per se. It is a wholesale economic shutdown, and is a unique unprecedented crisis as measured by the speed of the crash, and the extent of the liquidation, by both systematic and discretionary managers, which has dwarfed anything seen in 2008.”

Indeed, it is possible that financial market regimes have radically changed. “The Coronavirus crisis is not a financial crisis per se. It is a wholesale economic shutdown, and is a unique unprecedented crisis as measured by the speed of the crash, and the extent of the liquidation, by both systematic and discretionary managers, which has dwarfed anything seen in 2008”, he continues.

Liquidity

The liquidation panic was manageable from a trading perspective: “liquidity did get stretched, but we were able to trade everything. It cost more in March, even in Treasury markets, where bid/offer spreads blew out to 8 times the normal level”, says Remsing. The worst liquidity was seen in emerging market interest rate swaps, but here Aspect benefitted from having multiple trading counterparties. The model is to ask each counterparty for a two-way quote, which results in an aggregated bid/offer spread narrower than the individual quotes. “We typically get responses from 6-8 counterparties, but sometimes only got two or three back in March”, he recalls. Overall, despite trading three times as much as an average month, the bleed rate per day of trading was only twice as high. As of May, “spreads have now narrowed, but the order book is not quite back to where it was”, says Remsing.

A High Speed Crash

The extra trading costs did not prevent many CTAs from generating positive returns, but the crisis has thrown into sharp relief the differences between how systematic strategies performed. Given the speed of the crash, it is unsurprising that short term traders have somewhat outperformed traditional trend following CTAs: Societe Generale’s SG Short-Term Traders Index advanced 4.3% in 2020 to April while the SG Trend Index was up 2.47%.

Similarly, at Aspect, their shorter-term trading strategy, the Aspect Tactical Opportunities Program was up 11.86% as at end of April while the medium-term flagship Aspect Diversified (which is around 80% trend) was flat. Aspect Core Diversified HV was still nimble enough to have rotated from a risk on to a risk off stance over the first quarter. “We started 2020 as risk on as you can get, long of equities and, commodities, with very little in normal defensive fixed income plays. As US equities made an all time high on February 21, our first moves away from risk came in other asset classes: oil had already entered a bear market in January, and industrial metals had already turned. By the end of January, we had also started rebuilding longs in bonds. Between February 21 and March 10, we were just cutting risk, losing on equities but making up for it on other asset classes”.

“Between February 21 and March 10, we were just cutting risk, losing on equities but making up for it on other asset classes.”

The mechanics are worth investigating here: “the trend programs generate signals purely within asset classes. The other asset classes did not generate cross-market signals to reduce actual equity exposure, but having shorts in energy and metals, and longs in bonds and USD, did reduce the overall portfolio’s equity beta”, explains Remsing.

Though the trend program repositioned its exposures towards a risk-off stance, it did not produce the highest returns, partly due to risk management. “The spike in volatility also meant that Aspect was reducing position sizes across all markets in order to maintain its steady volatility target for the trend program. By late February, March and April the program was running at below average risk”, points out Remsing.

Relative Value Challenges

Relative value approaches faced headwinds. Some CTAs are in effect multi-strategy quantitative funds, running trend strategies alongside others including relative value, which may help to explain why the SG CTA Index was slightly negative at -0.3% for 2020 to April. “Some relative value spread strategies were decimated in March, particularly where slower moving or static models were upset by the new market environment”, says Remsing. Aspect’s relative value program, Aspect Systematic Global Macro (ASGM), managed to stay in positive territory not least because its models are somewhat faster than those of other relative value traders. Dynamic risk filters shifted exposures, using data not based on fundamental data.

“Some relative value spread strategies were decimated in March, particularly where slower moving or static models were upset by the new market environment.”

Old or New Data and Techniques?

Alternative data was also helpful. “ASGM has also benefitted from “nowcasting” techniques using real time data and sentiment to pick up the lurch into recession that could not have been forecast using backward-looking data. Nowcasting and natural language processing based models showed the economic slowdown before the PMI (Purchasing Managers Index) or unemployment data even came in. This meant the strategy was to some degree able to navigate the regime shift”, he points out.

By way of contrast, some other new techniques – machine learning models – used in some of Aspect’s programs did not always adapt well to the market climate in March: they lost money in fixed income, and made only small amounts in stock indices and energies amid record intraday volatility in Treasuries. “Perhaps counter-intuitively, machine learning works better when things are quite stable and does not necessarily cope as well with jump-risk or previously unseen things. One hypothesis is that large and rapid unwinds of basis trades disrupted patterns in bond markets”, says Remsing.

“Perhaps counter-intuitively, machine learning works better when things are quite stable and does not necessarily cope as well with jump-risk or previously unseen things.”

Another possibly surprising situation occurred in the single stock equity trading that resides in Aspect’s alternative risk premia strategy, the Aspect Absolute Return program (ARP), where 2,000 equities are traded: backward-looking fundamental data signals worked better than forward-looking analyst forecasts. This is possibly because the sell side analysts did not update their models fast enough or because the near-term uncertainty was so high that backward-looking financial statement data provided a more reliable picture of a company’s resilience

Of course, systematic managers regularly reshuffle and refine models. The crisis could lead to more extensive revision – and even revamping – of both old and new models.

 

This article featured in HedgeNordic’s report Systematic Strategies: When Numbers are the Key!

Share.

Leave A Reply

Time limit is exhausted. Please reload CAPTCHA.