Doing the maths: Why quant funds could be attractive to burned crypto investors

As digital assets suffer massive losses in both value and confidence, quant funds have emerged as a potential alternative for tech-savvy investors. Dalvinder Kular reports.

Cryptocurrency and other digital assets including non-fungible tokens (NFTs) have suffered hugely over the past year through what has been described as a ‘crypto winter’.

This largely stems from the spectacular collapse of FTX, the world’s second largest global cryptocurrency exchange, which faced bankruptcy after the crypto equivalent of a bank run in late 2022. Its potential buyer Binance – which now finds itself at the centre of numerous regulatory scandals across the world – walked away from a deal, leaving millions in investor funds in limbo.

Sam Bankman-Fried, the chief exec of FTX, is facing two separate trials on 13 fraud and conspiracy charges. He is currently awaiting trial in his parents’ home on a $250 million bond, the largest such bond ever set in an American criminal proceeding.

The episode triggered a huge sell-off in the market, with some cryptocurrencies losing as much as a quarter of their value. Bitcoin – a barometer of trust in crypto – fell to its lowest level in two years from a high of $69,000 in November 2021.

Though Bitcoin has rallied in the months since FTX’s collapse, the fallout from FTX’s collapse and other crypto-related scandals have led to a high level of distrust in the digital asset market and many investors are looking for ways to invest their money that involve less risk.

One such option is quantitative funds.

Quant funds – an introduction

Quantitative funds, also known as “quants”, are investment funds which rely on mathematical models, algorithms and AI to make their trading decisions. While these funds are not necessarily better than those led by humans decision making, they are efficient and can analyse large amounts of data at a speed which humans cannot. Additionally, they require less overheads and can charge lower fees than traditional funds and, importantly, seek to remove human emotion and subjectivity to create purely data-driven investment decisions.

The approach appears to be borne out by the numbers. Quant funds seem to be outperforming other asset classes, with Bloomberg predicting that AQR Capital Management’s Absolute Return Strategy will have its best year ever after the fund surged 40 per cent in November 2022. Man Group’s $11.6 billion AHL Alpha gained 10.7 per cent in November while Aspect Capital’s Diversified fund jumped 37.9 per cent over the same time frame.

As the technology improves and becomes more accessible, an increasing number of quant funds have incorporated artificial intelligence (AI) and machine learning (ML) in their investment strategies, blending the latest technology with traditional money management.

Experimenting with hyperparameters

One company operating in this space is Axyon AI, a tech firm based in Modena, Italy, which helps asset managers apply structured machine learning approaches to their investment funds. Its chief executive and founder Daniele Grassi tells FStech that there are many funds which have used AI and ML in some form for many years with varying degrees of success for tasks such as data analytics and data processing.

“My perception is that in the industry even more mainstream funds that use traditional approaches, statistical approaches or traditional factor-based investment approaches have started to rely on ML processes,” Grassi says.

“ML for traditional managers can be a useful tool to augment their processes. It looks at different elements in different ways. However, it cannot evaluate some elements that humans can. At the same time, it does evaluate elements that humans cannot evaluate,” he adds.

He explains that AI and ML allow fund managers to experiment with hyperparameters, which in turn allows them to see the data in a way they could never have before imagined.

Haris Chalvatzis, head of quant research at German financial consultancy 2iQ Research suggests that AI and ML “don’t make much sense” in long term investments and are best used in funds which are trading daily.

“You have lots of data and you need to make decisions fast. You need to have some tools in your toolkit that will allow you to take them faster. But it is an auxiliary method rather than the core component,” he says.

Chalvatzis warns that attributing the success of a fund’s outperformance to one particular method is difficult, but the results of using AI and ML can be impressive.

“When you have a fund that is say eight per cent up year-to-date, we don't have any visibility to what strategies are pushing outperformance – we know they are using AI and ML, but we don' t know at which stage and how much. I would argue as a rule of thumb that about 25-30 per cent of returns are probably coming from AI and ML,” Chalvatzis says.

Avoiding Bias

Laurent Laloux, chief product officer at asset management firm Capital Fund Management (CFM) highlights an industry-wide shift he has seen in the industry over his 30 years career. Back then, he says, the models were simpler due to the technology available at the time. Now CFM has made use of the newer technologies available.

“We try to make sure that we understand how the model works and what it entails. It's not a black box for us. We need to understand what is happening,” Laloux explains.

He warns that AI and ML need to be back-tested to show that they are learning and not memorising to avoid bias, a common refrain levied at supposedly ‘impartial’ AI and ML tech in every area ranging from finance to law enforcement.

“What is interesting with ML is the capacity to have a much more powerful way of sifting through data. It is dangerous and interesting at the same time. You can do a lot more, but you need to be extremely rigorous about your methodology.

“However, on the other side, it's much more industrial in the way you can browse and see some data and find a new relationship in the data that is not immediately obvious to a human being.”

Vincent Berard, head of product strategy at BNP Paribas Asset Management, estimates that around €40 million of the assets that the Paris-based bank manages use some form of quantitative strategy as part of its investment process. However, just one quant fund at the bank relies on AI for most of its strategy.

“It's something that we are not pushing too aggressively and promoting too aggressively to clients for the simple reason that it's a little bit more ‘black box’ than what we usually do,” he explains.

“This technology is still relatively fresh. We used some basic models that our lab was using for other purposes in the bank to apply to this investment objective. We really want to take the time to see how it works out.”

He adds that technology can be relatively simple and not all quant funds require hugely complicated models to provide good returns.

AI and ML are not being used just to manage funds. The technology is being integrated into areas adjacent to asset management. BNP Paribas, for example, is using bots to read emails and draft replies for clients leading to increased sales.

Berard recalls a conversation with a client who complimented the bank’s sales staff without knowing of this practice: “I had a discussion with a client who said: ‘your sales guy is amazing! I receive your quotes much faster than everybody else’s – how do you do that?’”

No definition of normal

While a layman might typically view the role of an asset manager as gazing into a crystal ball and predicting the future for their funds, the technologies they are using for quant funds typically tend to look towards the past.

Quant funds often rely on historical events and market data to predict how securities will behave in the future. Curiously however, events such as the war in Ukraine and the pandemic fail to show up in historical data.

Axyon AI’s Grassi says that there are still opportunities to be found for quant funds, even when the magnitude of events like the pandemic, natural disasters and wars are difficult to measure.

“There are never normal years; there is no definition of normal. The past couple of years have been particularly volatile with strong macro-led movements,” he says.

“These events are actually opportunities as quant funds are naturally cooler at handling these events,” he explains. “It can obviously play against them in terms of being able to react to an environment which changes abruptly. But in the long run, especially when everything is macro driven, the strategies of quant funds will pay off and that’s what we saw in 2022.”

Chalvatzis explains that it is simply not a case of plugging historical data into an ML system and hoping it produces results. Rather, the ML system must be trained so that it learns from the patterns.

“How you train the models is critical and it’s more of an art rather than science. It's probably one of the main causes of failure. How you train the model, how you deliver robust training, how you make sure that the data is sufficient and representative of what you're trying to achieve? There are a lot of decisions that go into the model training,” he concludes.

Powering results at any cost

An issue increasingly being highlighted by critics is the catastrophic environmental impact of digital transformation of the financial sector and quant funds are no different. Laloux said that the practice consumes a lot of CPU power and the amount of carbon produced conflicts with the need to care about climate change.

Data from research firm MSCI shows that the more assets under management a fund has, the larger the carbon footprint it has. Funds can produce as much as 200 tonnes of C02 per million US dollars. Research found that funds managed in the US produced more carbon than those based in Europe and the asset class had little effect on the amount of carbon produced.

Beyond just the environmental concerns, Laloux adds that AI and ML technology should not be blindly applied to all investment vehicles and that it should be used by investors with an understanding of how it works.

“It is not just simply pressing a button and looking for the result. As with any complex technology you really need to understand what it does, what it is good at and what are the pitfalls of such powerful tools and it's not a perfect tool,” Laloux explains.

While quant funds need technology and electricity, it should not be forgotten that they also run on people power. Quant funds have traditionally been powered by quant analysts. Usually holding PhDs in maths, physics, or analytics. These quant analysts look at the data and concoct the underlying algorithms that power a fund’s strategy.

Asides from the environmental concerns, AI and ML are increasingly being used by technology firms to do the job of humans and many are questioning whether there is still a need for funds to employ quant analysts.

Grassi argues that quant analysts are even more important than ever: “They can look at the data and process it. They are like translators, capable of translating the raw data that the ML can understand and the system can leverage. There is a space between raw data and what machine learning can really exploit to extract value.”

He adds that quant analysts are experts in handling data and can use their judgement to relate this data to the market dynamics – seemingly contradicting one of the core ideals of quant funds to remove the human element of decision making.

Like many other industries, fund management has seen a shortage of skilled workers. Laloux says that there has been a struggle to find data scientists and engineers for a protracted period.

Tech layoffs

The most significant story in big tech in 2023 has undoubtedly been the brutal level of cost-cutting that firms like Google, Meta, Twitter and Microsoft have carried out through layoffs. According to Crunchbase , which has created a rolling Tech Layoffs Tracker to illustrate the severity of the sector's downsizing, more than 158,000 workers from US-based tech companies (or tech companies with a large US workforce) have been laid off in 2023 so far.

Additionally, there is the ever-looming threat of AI programmes like ChatGPT. While these have yet to fully replace workers, they have automated repetitive tasks freeing up some of the workforce.

CFM’s Laloux however thinks the tech layoffs could be of benefit quant funds as highly skilled, technically proficient workers seek employment.

He explains: “The big tech companies are now laying off people, and we’re definitely seeing a lot more people coming towards quant because they feel that it's an interesting space where they can deploy their talent and have an interesting job - not that different from what they were doing in big tech. But I don't see yet a massive shift in the space. It's more a gradual trend that is still materialising in the market.”

The executive adds that there is demand for people developing ML to predict internet habits or predict what people want to buy and that these skills can be transferred into predicting how the financial markets will behave.

“For the quant researchers, we typically ask people for PhDs, with postdoc experience. We're still hiring a lot from academia. That's our core expertise. We have a strong partnership with academia. It’s the gap between technology and quant research which people from big tech can fill – the data science space. We like people with interesting backgrounds and a lot of experience in dealing with massive data sets and cloud technology to leverage that,” Laloux says.

Berard also notes that people who used to work for technology companies are finding their way into finance and banking, but he isn’t convinced that this is a trend that is here to stay.

“Since November, AI has grown in popularity and we also see a lot of startup companies hiring. For some companies it can be expensive to hire in the tech field. But I expect a lot of smaller startups will build on all the capital flows being invested in a lot of these new companies. But I'm not sure if these people are fundamentally passionate about technology that they will not go back into pure tech or simply will cycle into smaller companies,” he says.

Writing the rules

Laloux expects quant funds to continue performing well but warns of the potential risks that come with any form of investment. “The volatility inflation, a lot of unknown in the market is potentially good for arbitrage. But there's always the risk that you're on the wrong side of the coin.”

He adds that the use of AI and ML will continue to grow as the technology develops: “As we get access to much more complex neural networks and things with different capacities, that definitely brings potential ideas and capabilities in terms of dealing with the different data sets.”

Berard meanwhile explains that asset managers will always be needed to constantly monitor the models that they use, regardless of whether they use technology like AI and ML.

“There are a number of things that we are not comfortable doing because they are moving so fast that we can't really set strict rules and follow them systematically. We tend to take a lot of time to ensure if we use these technologies, we can monitor them and see if they are stable before we launch a financial product attached to it,” he says, adding that smaller firms have more discretion in how they use AI and ML and may be faster in responding to changes in the market.

“We must write the rules so that someone can almost replicate our strategy if they have the ability to access the data,” suggests Berard. “But it is a challenge when the technology is moving so fast – how do we put that in writing and how do we ensure that it is still going to run a few months down the line?”

It remains to be seen whether investors who built up significant digital asset portfolios will move into quant funds, but Berard concludes that there is more money flowing into quant funds than other strategies. He predicts that this will probably continue in the short term.

While the long-term viability of NFTs, cryptocurrencies and other digital assets remains unclear, the demonstrable method behind quant funds – while often requiring a high level of technical knowledge to fully comprehend – shows that they are a more stable avenue for retail investors and pose less of a risk in 2023 and beyond.

    Share Story:

Recent Stories

Safeguarding economies: DNFBPs' role in AML and CTF compliance explained
Join FStech editor Jonathan Easton, NICE Actimize's Adam McLaughlin and Graham Mackenzie of the Law Society of Scotland as they look at the role Designated Non-Financial Businesses and Professions (DNFBPs) play in the financial sector, and the challenges they face in complying with anti-money laundering and counter-terrorist financing regulations.

Ransomware and beyond: Enhancing cyber threat awareness in the financial sector
Join FStech editor Jonathan Easton and Proofpoint cybersecurity strategist Matt Cooke as they discuss the findings of the State of the Phish 2023 report, diving into key topics such as awareness of cyber threats, the sophisticated techniques being used by criminals to target the financial sector, and how financial institutions can take a proactive approach to educating both their employees and their customers.

Click here to read the 2023 State of the Phish report from Proofpoint.

Cracking down on fraud
In this webinar a panel of expert speakers explored the ways in which high-volume PSPs and FinTechs are preventing fraud while providing a seamless customer experience.

Future of Planning, Budgeting, Forecasting, and Reporting
Sage Intacct is excited to present FSN The Modern Finance Forum’s “Future of Planning, Budgeting, Forecasting, and Reporting Global Survey 2022” results. With participation from 450 companies around the globe, the survey results highlight how organisations are developing their core financial processes by 2030.