Technological innovation grows exponentially, which means that every year when the QuantMinds community gathers, there are more breakthroughs to share. QuantMinds International, however, is not just about the innovations that the community has been working on, but it’s also about the industry itself. What are the key trends in quant finance? And how will these trends impact the future of quant finance? That’s what we’ll explore below.
Causality: A game-changer in machine learning and artificial intelligence
“We're in a time when mathematical models can improve your work incrementally”, Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis, UCL, told us in her keynote speech.
Indeed, models have enhanced many things in quant finance, but being a regulated sector, broad adoption is still far away. The concept of explainable artificial intelligence (XAI) seems to be the answer – if it was possible to understand how the machine came to its conclusions, it would solve the issues of transparency that is associated with machine learning and AI. However, Fabien Choujaa, Global Head of Algorithmic Trading Model Risk Management, Morgan Stanley, noted that we don’t currently have a definition of explainability. Commonly, we use synonyms to imply explainability, for example transparency or causality – the latter of which is increasingly important for quants.
“Data can only tell us what happens. Not why”, Ioana Boier, Head of Quantitative Portfolio Solutions, Alphadyne Asset Management, explained.
If an AI can successfully sort and categorise photos of wolves and dogs, does that mean the AI knows what wolves and dogs are? Not quite – it turns out the AI noticed that the images of wolves have snowy backgrounds, hence snowy background equals wolves. Correlations are a key staple of financial modelling, but correlation is not causation.
Data challenges and an ESG-led future
There may be misconceptions around explainable AI and lower performance, however, the model will be as good as the data representation. And even though we talk about big data and leveraging it to our advantage (making data complexity a key focus area for quants), the data that’s available is not enough.
This year, when Covid-19 first hit, the question whether we can predict black swan events came up in conversations again, but this time with a twist: can financial models predict black swan events? In theory, yes, Marcos Lopez de Prado, Global Head – Quantitative Research and Development, Abu Dhabi Investment Authority, said in his keynote, but only if we had access to the right data.
This access is already key for quants to prepare for the future, for example when integrating ESG into the day-to-day. At the moment, ESG data is limited, therefore quants’ ability to make informed decisions on this is also limited. However, it is generally agreed that ESG is an area that needs improvement, not only because investor and client pressures are mounting (with regulatory pressure expected to follow), but because ESG is possibly the biggest long-term trend (and risk) that will impact all sectors. With that in mind, will we see the rise of climate quants, much like we saw the rise of risk quants over a decade ago? Will new frameworks and tools be needed? And will quants be buying or building the data and the tools?
Quants in the bigger picture
As innovations in financial modelling entered and algorithms became more sophisticated, the quant role evolved accordingly. The traditional role of quants was to develop models, but now, quants play an important role in risk and decision-making, and in managing and correcting inefficient processes.
“The environment changed”, Nadhem Meziou, Head of Fixed Income Quantitative Research, Natixis, noted.
Regulatory pressure impacts the whole organisation, but regulatory pressure on model risk management changes the role of quants. In the future, quants will have to go beyond answering regulatory requests, ensure the right and responsible use of machine learning, and furthermore justify the use AI & ML methods to the regulators.
Understanding machine learning is therefore a key skillset in a quant. However, Jeanine Kwong, Global Head of Investment Risk Oversight, Manulife, reminds us that when hiring, ML skills are not all you want. You’re still looking for a person to fit into and complement your existing team.