How fund houses are looking to build artificial CIOs

Move over robo funds; the next generation of artificial intelligence-based funds is being rolled out. But while they could spell big changes for asset managers, they also have drawbacks.
How fund houses are looking to build artificial CIOs

Increasingly, the world’s fund industry is grappling with the impact of two letters on its business: AI. 

Artificial intelligence looks likely to disrupt global portfolio management in the years to come. Robo advisers, which automatically put customer assets into exchange-traded funds based on risk tolerance, are already proliferating. The next wave of AI funds is set to be disruptive in more radical ways.

Many fund houses already use technology to analyse data, churn out ideas or refine their trading signals, but a human fund manager still controls how much exposure to take and how to execute the trade. Virginie Maisonneuve, Singapore-based chief investment officer (CIO) at Eastspring Investments, has worked with local universities to study the role of AI in asset management. She said AI’s ability to calculate and quantify masses of data could better analyse corporate transparency, treatment of employees and environmental impact. 

“Strategies integrating ESG principles and/or thematics will form crucial parts of product offerings in the future,” she predicted.

But some fund houses are taking AI portfolio management to the next stage: fully robotically controlled funds that automatically conduct analysis, security selection and trading. 

Global fund giants such as BlackRock and hedge fund specialists Man Group and Two Sigma are leading the field. Florian Spiegl, co-chair of the artificial intelligence committee of the FinTech Association of Hong Kong, said such companies have been exploring how to integrate machine learning into investment strategy for some time.

“Some Asian regional firms are increasingly active in developing machine learning solutions or forming research collaborations,” Spiegl said. China Asset Management is one of a handful of regional firms to have launched AI-managed investments. 

Investors and fund houses are taking the potential seriously. Japan’s Government Pension Investment Fund (GPIF) is studying AI and its likely impact on their use of third party funds. 

GPIF’s CIO Hiromichi Mizuno is a firm believer in the transformative nature of AI, and the pension fund has partnered with Japanese technology company Sony and consultancy Accenture to understand how to understand how AI funds work. 

“I started thinking how in five years’ time our team can do due diligence on fund managers,” he said. “If one manager comes to us … and says ‘we have 200 analysts and 50 very experienced portfolio fund managers’, and then other people come in jeans who just graduated from Stamford and have just created an AI manager but their performance using the simulation is 10% over the last 10 years, how can we judge each side? That’s the question I try to address and try to make our team prepare for.”


AI has myriad potential uses in fund management. It can be applied into data aggregation and interpretation, correlation analysis and predictive modelling, all of which can directly feed into the investment decision-making. 

To date there are relatively few AI-supported or fully managed funds, but that looks set to change. A report by Opimas in March 2017 predicted that by 2025 the global asset management industry will shrink by 90,000 people (from today’s total industry figure of 520,000), with most the jobs being assumed by machines.

Renaissance Technologies is one of the most progressive firms. The famously reclusive hedge fund’s Medallion fund has been supported by AI tools for quantitative number-crunching; it has returned more than 35% annualised over a 20-year span. Advances within the field of machine learning means AI tools will be increasingly used to support mainstream investment management over the next few years. To facilitate this fund giants are hiring a particular type of professional—the data scientist—to develop and optimise algorithms.

“Technology spending by financial firms has risen significantly and this trend can be observed across all the larger players, including UBS, Goldman Sachs, BlackRock, Morgan Stanley, Deutsche Bank and JP Morgan,” he said.

That shift looks set to accelerate and transform the fund industry, as cost pressures continue to mount. 

Gaurav Chakravorty, CIO of Qplum, a US-based asset management firm that offers AI-based trading strategies, said the asset management industry takes in about $1 trillion in revenue every year, more than half of which is invested into human capital. He believes fund houses will replace more humans with computers to cut this expense.  

“$100 billion of annual investment will go into technology-driven roles in the next two-to-three years,” he predicted. “A typical asset management firm will soon have more machine learning engineers and data scientists than people with experience in financial markets.”

Much of the investment will utilise AI in a supportive capacity. Brisbane-based Queensland Investment Corporation (QIC), for example, is using AI investment to collect and analyse data. 

“We are looking to use technology to gather and store as much as information as possible, in order for us to have an information advantage in terms of making investment decisions,” David Asplin, managing director of global business development at Brisbane-based QIC said.


But having AI data analysis and quantitative functions is not the same as using algorithms to make active investment decisions. 

So far a few hedge fund operators have created what they claim to be fully AI managed funds. Singapore-based hedge fund data provider Eurekahedge has built an index to track AI hedge fund strategies that use machine learning in investing and trading processes. It has 28 constituents, of which only 16 are still trading. The index includes the others to account for survivorship bias and offer a true picture of the sector’s performance. All-told, the functional firms represent about $1.8 billion in AUM.

The funds that still exist have offered some encouraging results to date (see chart on AI Hedge Fund Index versus quantitative and traditional hedge funds). But the limitations of the technology were highlighted in the first quarter of this year, when volatility in the world’s capital markets spiked. 

The volatility spike was a rude shock for the AI funds, said Mohammad Hassan, head of hedge fund research at Eurekahedge. “All of last year the VIX index (which measures market volatility) was down below 10 and suddenly this year it shot up to 50. Whatever was happening in the AI portfolios was suddenly upset by this volatility swing.”

That wasn’t the only factor. Returns from individual AI funds varied depending on market exposure, said Hassan. While he declined to disclose the identity of specific funds, he noted that one fund that took a big hit uses AI to invest in over 500 stocks in the US. 

“It is churning through tonnes of data and based on that is rotating its portfolio very often. If you’re holding a lot of small and mid-cap stocks, the losses will be huge.” Another fund investing in Topix futures incurred heavy losses when the Japanese market experienced
a correction. 

In essence, the key failure of the AI funds was an inability to adapt to rapidly shifting market conditions. They lacked the smarts to change as conditions did. 

Look out for the next part of this feature from our April/May magazine edition, in which we consider the limitations and possibilities of AI funds. 

¬ Haymarket Media Limited. All rights reserved.