Emerging AI adoption drives change in asset management in Asia

For all the excitement about artificial intelligence in asset management, AI adoption is relatively more nascent in Asia, according to panelists at Bloomberg’s Machine Learning Decoded 2017 conference which took place in Singapore, Hong Kong and Sydney.

One reason for this is quantitative strategies – the main focus for artificial intelligence in asset management – being far less prevalent in the region, where there is more of a focus on fundamental investing. Currently, only about one third of Asian institutions include quantitative investment strategies in their portfolios, according to a recent study by Greenwich Associates.

Two developments are taking place that will help push the adoption of AI past this nascent stage in Asia: AI will be applied to fundamental investing in a broader capacity as applications become more commonplace and affordable; and quant trading strategies will gain in popularity in the region. Investors, seeking to diversify into new sources of alpha with low correlation to fundamental managers, are keeping an eye on quant strategies as a solution.

“At this point, someone somewhere has automated every part of the investment process from alpha generating strategies to risk analysis. It behooves us to be aware of this dynamic and try to keep up with it,” Gary Kazantsev, Head of Machine Learning Engineering at Bloomberg, told attendees.

Augmenting core strengths

For now, asset management firms in Asia are slowly rolling out AI in niche applications that tend to fall into three categories: helping quant traders set up proprietary algorithms; helping fundamental managers incorporate more data and non-financial data into their analysis; and lastly helping IT and administrative staff spend less time on repetitive “janitorial” tasks.

One quantitative firm, for instance, is using AI to improve data capture and to help its fund managers extract more alpha from data. On the fundamental side, the firm is aggregating large data sets through a cognitive AI engine to deliver actionable investment insights to fund managers in its equities business. The firm is also using artificial intelligence to reduce various types of risk – reputational or cyber – and keep tabs on what traders are doing.

Firms are also using machine learning, a subset of artificial intelligence that focuses on giving computers the ability to learn, to come up with new investment ideas. One startup, for instance, created a tool that can digest data and act as an additional “vote” in investment decisions. On the administrative side, one firm is using machine learning to track funds and read through reports to identify red flags.

Bloomberg is introducing AI to make data more accessible; for instance, working on making unstructured data, such as data tables, digestible to machines and integrating natural language interfaces into search functions.

Leaping into data acquisition

In a rush to be ahead of the curve, many investment firms in the region are buying more data to use in AI or big data applications.

“Some people are buying a lot of data – not just quant funds but fundamental funds that have never touched quantitative strategies before – they are spending money and moving into big data,” Chauwei Yak, founding partner at GAO Capital said at the Machine Learning Decoded conference in Singapore.

Despite the flurry of activity, Yak said, there’s still much development to be done and firms are encouraged to examine how buying data can translate into higher earnings or a more accurate estimate of share prices.

Machine learning, in particular, sometimes leads to further investment insights, said Khoi Le Binh, Head of Global Quantitative Strategy Asia Pacific at Deutsche Bank, but “it’s no magic wand. Just because you have data or machine learning doesn’t mean you suddenly have a great model,” he reminded conference participants.

One consideration for firms that are looking to integrate big data and quant strategies is transparency. One of the difficulties of using artificial intelligence to find correlations or investment recommendations is a lack of transparency – it’s not always clear how or why a recommendation is made and correlations can be spurious. It’s going to be particularly important for managers to explain how artificial intelligence is being used in investing models.

Transparency and interpretability are particularly significant issues in Asia where investors can be a bit wary of “black box” strategies. A recent study by Greenwich Associates found that 45% of Asian institutional investors surveyed found quant strategies to be “too opaque” and would like managers to better explain their strategies. Bloomberg’s Event-Driven Feeds help build transparency into strategies by delivering real-time, machine-readable data for black box applications.

Interpretability, particularly as it applies to finance, is a key issue with machine learning, Bloomberg’s Kazantsev said, because some investors will want an understandable thesis, depending on the investor’s risk tolerance profile.

Artificial intelligence will have a widespread impact on the asset management industry, just as it is effecting change in other sectors, but firms in Asia are just starting to dip a toe into the possibilities. To fully take advantage of AI development, asset managers in Asia will need to assess the value of acquiring data and building transparency into their evolving strategies.

Request a demo.