One size does not fit all – Adapting to meet investment goals

Guillaume Arnaud, global head of quantitative investment strategies (QIS), and Sandrine Ungari, head of cross-asset quantitative research at Societe Generale, explore the benefits of QIS for investors, why flexibility is crucial for investors to meet their specific objectives, and the biggest opportunities for growth in QIS 

 

What are the main benefits of quantitative investment strategies (QIS)?

Guillaume Arnaud, Societe Generale

Guillaume Arnaud, Societe Generale: QIS can be a valuable tool within the context of an existing portfolio to enhance performance or mitigate drawdowns. Depending on the tools implemented, QIS can help isolate specific exposures and, in turn, improve diversification. Their daily liquidity and cost-efficiency is an attractive argument compared to the alternatives. One size does not fit all, so flexibility is another important feature, should investors need to customise their strategy to meet their specific investment objectives. 

Sandrine Ungari, Societe Generale: QIS provide direct access to a wide range of risk premia strategies with attractive risk/return profiles that were previously only available to specialist traders or hedge fund investors. In addition to providing access to a wide investment universe, the strategies offer greater transparency than traditional alternative investments since they are rules-based and usually supported by strong academic and quantitative research. 

 

Which criteria should investors prioritise when selecting a strategy?

Guillaume Arnaud: Prior to conducting due diligence on a specific strategy, investors must have clear investment objectives. Is the goal to exploit a short-term market dislocation, to hedge a portfolio or to harvest a strategic premium?

Then, when digging into the mechanism of a strategy, a thorough understanding of the fundamentals and the rationale is key – the performance of some strategies tends to vary depending on the macro context or market cycles. For instance, QIS offer a wide range of strategies based on harvesting risk premia, which, by definition, involves getting paid a premium for taking a known risk. Identifying the risk and the scenario in which it would materialise is paramount.

Investors also need to take into consideration the range of implementation choices and their potential impact in terms of performance or alignment with their investment objective. Price observability of the instruments traded by a strategy should be part of that assessment. 

 

To what extent has poor performance in 2018 changed investors’ perceptions of QIS? What lessons can be learned?

Sandrine Ungari, Societe Generale

Sandrine Ungari: After a solid performance in 2017, 2018 was a challenging year for risk premia strategies. Idiosyncratic risks and market context put downward pressure on risk premia performance. It is safe to say equity factors have fallen out of fashion as portfolio construction – rather than factor choice – has dominated returns. However, even though the skilfully selected combination of strategies tends to build a strong risk/return profile over the long term, diversification had only a limited impact last year. Monetary policies clearly still matter, and market microstructure is becoming more and more important. Within this demanding environment, carry was a silver lining. Carry strategies have been consistent performers over the past few years. For instance, the fixed income universe may offer some interesting possibilities depending on the risk profile of the investor. 

 

How should success be measured?

Guillaume Arnaud: Benchmarking performance is one of the most widely discussed issues related to risk premia investment. A few solutions have been developed in the past two years to help investors navigate the range of investable strategies, such as non-investable indexes measuring the largest risk premia funds and benchmarks based on bank strategies. Some of the later ones use quantitative methods to remove the noise coming from differences in implementation. 

Sandrine Ungari: Strategies also need to be benchmarked against an investor’s individual objectives. When entering a new risk premia strategy, investors typically use the backtested performance, the live performance and its fundamental rationale to derive a set of expectations. Once invested, investors need to compare those expectations with reality: Does the strategy perform in-line with backtested and past live returns? Has the strategy behaved as anticipated given the macro and market environment? Has the strategy delivered according to the investment objective?

For instance, when adopting a defensive strategy, the investor expects to benefit in a risk-off environment. Some strategies, such as investing in good-quality stocks, have kept this promise in the recent past. Others – such as trend-following strategies – have struggled in the recent bear market, raising questions around their true defensiveness across all types of crises.

Another example is diversification. Risk premia strategies are often used as a diversifier or a decorrelated alpha overlay. But some strategies, such as volatility selling on equity, become very directional when uncertainty rises. Choosing the right strategy to fit an objective process is key for a successful investment in risk premia.

 

Which macro factors are likely to have the greatest influence on performance over the next 12 months?

Sandrine Ungari: The macro environment of early 2019 was largely influenced by the reversal in central bank policy, the resulting collapse in interest rate expectations and the ongoing tensions between China and the US. 

Going forward, the dynamic of the trade negotiations between China and the US should continue to shape market sentiment. The direction of US and European Union monetary policies will also play a significant role, not only influencing investors’ growth and inflation expectations, but also balancing supply and demand for some investment products. Those two factors could ultimately impact the performance of risk premia strategies. 

Guillaume Arnaud: Extremely polarised valuations within equity markets and the abnormally high levels of corporate leverage in the US are also important risk factors. This leaves markets at risk from both significant upside and downside risk, from either a deterioration in credit conditions or significant rotation if some of the macro issues currently vexing investors fade or indeed are resolved. 

 

How are risk management practices evolving in this area?

Sandrine Ungari: As the risk premia concept matures, so does the associated risk management. At the strategy level, there is now a better understanding of the potential implementation biases. As small companies largely underperformed large companies last year, the size bias became evident in some of the implementations of the equity risk premia. As the US Federal Reserve increased rates by over 2%, a hidden long-duration bias also hurt strategies that were not truly duration-neutral. Low volatility in equity or some fixed income carry strategies are a case in point. While it is difficult to completely suppress all implementation biases, better strategy construction can help reduce them. Societe Generale Quant Research has spent a significant amount of time over the past two years highlighting the main biases stemming from classical implementation and proposing innovative frameworks to neutralise them. 

Guillaume Arnaud: At portfolio level there is now a greater awareness of potential hidden risk – such as tail and duration risk. To mitigate these, a better allocation framework and a well-designed hedging overlay are required. In light of recent history, it has become clear not all downside risks are the same. Last year saw two very different bear markets – one led by a crash in volatility and another by a general equity meltdown. Each bear market calls for a different hedge and, again, allocation across various solutions is key. Beyond hedging equity risks, solutions have started to emerge to hedge other asset classes, such as rates or credit.

 

What influence are machine learning and new technologies having?

Sandrine Ungari: The wider adoption of new technologies and machine learning in investment strategies is essentially creating a new way of doing things by radically changing how we analyse markets and data. Machine learning is not new, neither is having a lot of data to analyse. But, as these tools become more broadly available, they are providing quants with new insight into their data and the performance of their strategies.

For instance, backtesting typically forms the backbone of most QIS strategies, but it is fraught with biases and hindsight bias. Machine learning provides some extremely powerful and perhaps more robust data-mining tools, which – if used correctly – can help us not only improve on what we have, but also offer new insight into how markets behave. 

Guillaume Arnaud: One of the applications of machine learning that we see – based on Societe Generale Quant Research work – is, for example, to apply those sophisticated algorithms to the dataset we use to design our factor indexes. The aim is to overcome some of the biases prone to classic multi-factor exposure and open new perspectives for factor rotation models. 

 

What role will big/alternative data play in finding new sources of returns? 

Sandrine Ungari: Forecasting prices is notoriously difficult and many alternative data sources have stumbled with their offers of alpha. Alpha generated by alternative data sources is most likely to be short-lived, with very shallow market depth. But the industry is maturing, with alternative data now incorporated into existing investment process, such as forecast earnings, identifying environmental, social and governance issues and other fundamental processes. Web-scrapped data, satellite data or consumer-driven data can also help provide a more accurate picture of the macro environment, with the emergence of concepts such as ‘quantamental’ and ‘nowcasting’. Flow-driven data, such as intraday price and volumes or fund flows, is also becoming more commoditised today and could be classified as alternative data to the classical end-of-day observation of prices. This data, coupled with the use of machine learning techniques, could potentially lead to a better understanding of flows in the market. Alternative data does not change the investment process, but makes it more robust, accurate and reactive, thus potentially generating better returns.

 

What is the biggest growth opportunity for QIS in the future?

Guillaume Arnaud: Societe Generale is dedicated to continuous development of its QIS platform features to make it more flexible, transparent and cost-efficient for the investors. The platform is progressively evolving in two main directions – the enhancement of bespoke execution capabilities on Societe Generale’s platform and the expansion of Societe Generale’s solutions into broader investment universes and geographies. 

On the execution side, Societe Generale is constantly developing smarter systematic trading algorithms to allow for cost-efficiency and overall improvement of the execution capabilities. The idea is to offer clients challenged by management fee compression an ‘à la carte’ outsourcing opportunity to invest directly using our execution platform without necessarily investing in an index, thereby outsourcing their administrative and operation functions to Societe Generale. 

We also constantly innovate by implementing new strategies that give investors access to specific dislocation in the market, or by completing our risk premia range, backed by the research of the Societe Generale Quant Research team. Our analysis shows that large asset management and pension fund portfolios have significant biases towards equity, creating the space for diversification in the fixed income and commodities space. Bringing diversification benefits, robust performance and risk neutrality is key. Societe Generale differentiates itself by exploring new opportunities for the next generation of strategies, such as intraday market microstructure and machine learning. The goal here is to reap the benefits of the research while prioritising the capacity considerations and investability of strategies. For all our strategies, we have a methodical process to test and determine the risk premium resilient to the risk of crowding – the decrease in the ‘reward’ for taking the risk as the risk premia becomes more popular.

Opinions expressed are current as of the date appearing in this publication and neither Societe Generale nor its subsidiaries or affiliates accept any responsibility for liability arising from the use of all or any part of this document.

 

Quantitative Investment: Uncovered – Special report 2019
Read more 

Sponsored content