AsianInvesterAsianInvester

What AI funds need to do in order to flourish

Artificial intelligence could revolutionise actively managed funds, but it cannot expand until regulators accept them and they build multi-year track records, say advocates.
What AI funds need to do in order to flourish

Financial regulators need to better embrace artificial intelligence (AI) technology and AI funds build demonstrable performance over several years if cheap, automated investment funds are to become widespread among investors, argue financial technology (fintech) advocates.

Paul Smith, the Hong Kong-based former global head of the CFA Institute and now a strategic adviser to portfolio software provider SuisseTechPartners, believes it could take 2025 until for active AI portfolio services are available unless local financial regulators become more comfortable with the technology and give the green light to fintech-driven business models.

He is frustrated by this, arguing the funds have a lot of potential.

“Today, robo advice is very much based around ETF platforms, but I don’t think it stops there. In due course, smaller investors will have active portfolios that are managed partly algorithmically. I’m pretty sure passive is not the endgame,” he said.   

“Financial technology ought to enable you, for $1,000, to run a single stock portfolio. There’s no reason why that can’t be done now. Trading costs have come down; the technology exists,” added Smith, who has been a long-time advocate of lowering the cost of investing.

A 2018 report by Opimas predicted that by 2025 the global asset management industry will have shed 90,000 of its 500,000+ workforce, to be replaced by machines.

Several asset owners are looking seriously at the potential of AI. Brisbane-based QIC is using AI to collect and analyse data to optimise its portfolios, though head of funds management David Asplin told AsianInvestor he couldn’t envisage AI machines evolving sufficiently to take over from humans completely.

Japan’s Government Pension Investment Fund (GPIF) is studying AI and has partnered with Japanese technology company Sony and consultancy Accenture to understand how AI funds work and how to use them.

But some industry participants say there are legitimate questions over the current sophistication of fully automated AI fund management technology.

Weaknesses in the algorithms supporting the funds can be found wanting during spikes in capital market volatility and it can be difficult for humans still to identify the reasoning and logic behind them, according to Singapore-based data provider Eurekahedge. These have been particularly evident in recent days.

Performance numbers for the last two weeks are not yet available, but based on previous periods of market volatility, the funds have yet to fully prove themselves.

TRACK RECORD NEEDED

The incompatibility of various computing protocols is also halting progress, as it has with blockchain technology.

“Investors want to know can the AI fund do well over five years,” said Musheer Ahmed, one of the founders of the Fintech Association of Hong Kong who now runs Indian start-up Finstep Asia. He noted that the products are being stymied by this lack of pedigree.

Musheer Ahmed, Finstep Asia

“How will it react over the full economic cycle and how will it behave in a bear market, such as we are entering now. Indeed, now would be a good time to assess how AI funds behave, in a completely new phase of the market.”

There are signs that AI funds have potential. Eurekahedge has an index that tracks hedge fund strategies that use machine learning in their trading processes. According to the index, these funds have consistently outperformed long/short equity funds and CTA/managed futures funds over the long term (see below).

Chart: Relative performance of AI-driven hedge funds (in green)

Source: Eurekahedge

Additionally, AI portfolios should be far cheaper than equivalent hedge funds. ChinaAMC launched an AI-managed multi-factor A-share strategy in 2018 which uses machine-learning and automation to pick and trade stocks. The strategy charges fees comparable to passive products in China – typically between 40 and 75 basis points. It has returned 17.26% from its launch in January 2018 to the end of February 2020, compared to a -5% return for the CSI300 index.

RISING R&D

While AI investment funds have yet to gain mainstream credibility, investment into technology-driven roles will increase as part of fund manager budgets. US fund giant BlackRock estimates that at least 40% of its current staff are involved on the technology side of the business and fund distribution platforms like its Aladdin expect to carve out a big market share from the future AI-driven market.

But Ahmed believes it won’t be the traditional fund houses that succeed with AI funds – “They will not be as quick and agile as the tech-based providers.” Instead he pointed to the likes of e-commerce companies Alibaba and Tencent.

The Chinese companies possess strong data and analytics about their customer preferences and also possess investment subsidiaries that offer them the knowhow of traditional asset managers.

“I wouldn’t be surprised to see the next big wealth manager being one of the big four Chinese technology firms – Alibaba, Tencent, Baidu or Xiaomi,” said Ahmed.

For more information on the growth of technology in investing, click this link to listen to a new webinar, "New Technologies for an Investment Era', conducted on February 26 in association with Refinitiv.

¬ Haymarket Media Limited. All rights reserved.