AI for investors: Potential applications

In a recent survey of its Asia Pacific clients, fund manager Northern Trust found that artificial intelligence is widely expected to become commonplace within institutional investors by 2020. About 60% of the fund’s Australian-based institutional investment clients believe meaningful adoption will occur over the next two years, while Singapore and Hong Kong-based clients anticipate broad adoption in two to five years.

AI implementation in the investment community is unavoidable, but what specific applications can we expect for institutional investors?

The magazine Institutional Investor tries to answer this question by observing the types of technology the largest US asset managers have been investing in. It notes that State Street Global Advisors has been building up its AI infrastructure by investing inmassively parallel processing, a technology through which multiple processors work on different parts of a program, helping it deal with large datasets to forecast company fundamentals and find anomalies in fundamental data.

Meanwhile, Blackrock has mainly focused on factor-based investment - building portfolios based onrisk factors rather than asset classes for more efficient returns - since its acquisition of Barclays Global Investors in 2009. This strategy requires it to analyse large sets of data using artificial intelligence. “Machine learning and artificial intelligence techniques allow us to comb through this often messy data to glean insights never before thought possible—like the speed of construction in China, foot traffic into major department stores and sentiment picked up from thousands of online employee reviews. When combined with millions of other data points, these factors help asset managers make smarter investment decisions that impact our clients’ financial well-being,” wrote Blackrock’s chief engineer Jody Kochansky in a March 2018 blog post.

The asset manager also launched a “Lab for Artificial Intelligence” in February, aiming to speed up efforts to bring the benefits of these technologies to its clients.

Another significant development is the growing use of robo-advisors, whereby hedge fund managers and software developers have teamed up to create trading robots that execute trades on their behalf. This has been made possible by improvements in machine learning technology, allowing the robots to continuously learn without requiring human input. According to Cerulli Associates, the digital investment advice market will grow to US$489.1bn in assets under management by 2020.

In a recent post for Enterprising Investor, Dan Philps, the head of Rothko Investment Strategies, argues that human portfolio managers should still be part of AI-powered investment workflows, if only for sense checking. He believes neural-symbolic reasoning - cognitive computational systems integrating machine learning and automated reasoning- are showing a lot of promise in terms of creating fundamental concepts that can be read and understood by humans. “Purely data-driven or black box decision making — and that includes factor investing — is not acceptable for the high magnitude decisions of long-term investing,” he says.

Meanwhile in the lending sector, machine learning is changing the way credit scoring is conducted, with the potential to benefit SMEs in particular. The eventual intersection of lending and investing AI applications has the potential to unlock a whole new layer of liquidity for trade finance.