Machine Learning Decoded

Over the last two weeks, Bloomberg hosted its inaugural “Machine Learning Decoded” summits in Hong Kong, Tokyo, Singapore and Sydney to uncover the latest developments and applications of machine learning in finance, and how it is impacting investment strategies.

More than 500 participants from the financial community in these four cities, including the buy-side, hedge funds and quant funds, convened to hear from Bloomberg’s global experts and senior executives at leading asset management firms.

Gary Kazantsev, Bloomberg’s head of the Machine Learning Engineering team, and Gideon Mann, Bloomberg’s head of data science, discussed how machine learning and data science are transforming and impacting the financial sector. Gary looked back at how machine learning started in the 1950s with the birth of artificial intelligence, and the diverse set of machine learning applications available today, including sentiment analysis.

Gideon spoke about how a greater volume of data is inherently creating more problems as machine learning techniques improve, and how a consequence of machine learning is the ability to digest unstructured data. Looking ahead, he expects a greater application of machine learning on end-to-end strategy development, deep learning for time series analysis, and that we will get better and faster at looking at data across languages.

We also invited Professor Pascale Fung from Hong Kong University of Science and Technology to share her pioneering work and the latest on sentiment and emotion-aware natural language processing. Senior executives, including Virginie Maisonneuve, Chief Investment Officer of Eastspring Investments, Jinesh Patel, Partner at Dymon Asia Ventures and Chauwei Yak, Founding Partner at GAO Capital as well as Paddy McCrudden, Head of Data Science at Magellan Asset Management and Girish Nair, Global and Asia Pacific Quantitative Strategist at Bank of America Merrill Lynch, discussed how they are using AI, big data and machine learning for better alpha capture, and better understand fund performance and the changing investment environment.

Renowned quant finance researcher Bruno Dupire, Bloomberg’s head of Quantitative Research, closed with a compelling presentation on the latest applications of machine learning in finance, including time series prediction and sentiment based strategies.

At the events across Asia, we conducted a survey with participants and discovered the following insights:

• Machine learning (ML) adoption: The majority of participants in Hong Kong (39%), Tokyo (40%), Singapore (60%) and Sydney (55%) are at the exploration stage in applying machine learning in their processes and business operations. Only 4.6% in Hong Kong, 2.5% in Tokyo, 3.3% in Singapore and 4% in Sydney say they are using ML for end-to-end strategy development.
• Key drivers of ML: Nearly half of the participants in Singapore (48%) and Hong Kong (47%) and the majority in Sydney believe that the rise of quantitative investing is the key driver fueling investments in machine learning in the finance industry. In Tokyo, a third attributed ML investments to data interactions becoming more complicated.
• Key problems solved: The majority of participants in Hong Kong (62%), Singapore (50%) and Sydney (46%) say that ML helps deliver more sophisticated intelligence into trading, while most in Tokyo (33%) indicated that ML addresses the challenges of processing complex data better.
• Top ML applications: The most popular applications of ML in firms include portfolio selection and execution (26% in Tokyo, 31% in Singapore), forecasting returns (30% in Hong Kong), and strategy development (28% in Singapore, 23% in Tokyo, 27% in Sydney).
• ML spend: About a quarter in Hong Kong (27%) and Singapore (29%) indicated that they intend to increase corporate spending on ML by 10-30% over the next year.
• ML challenges: Key challenges faced when using ML include sourcing high quality data (40% in Hong Kong, 32% in Singapore) and a lack of expertise (31% in Hong Kong, 38% in Singapore, 33% in Sydney).

For several years now, Bloomberg has been implementing and investing in machine learning. We have more than 200 engineers working on data science problems and nearly 100 data science experts, and our resources dedicated to machine learning are growing. We are investing in natural language processing, which is machine learning methods applied to text, as well as information retrieval and search, and core machine learning (or deep learning).

Our efforts have focused on deriving intelligence and insight from data – so our clients can make smarter, more informed decisions about their business and financial strategies.

Read the special Machine Learning edition of Bloomberg Briefs here, featuring Q&As with some of our top experts.