Muang Thai Life of Thailand is moving towards a goal of integrating an internally developed artificial intelligence (AI) technology into its portfolio management, three years after having first developed it.
The life insurance company, which has Thb519.1 billion ($16.4 billion) in assets, has been conducting live investment portfolio tests to see how well the AI functions. It is designed to use quantitative algorithms and deep learning to identify risk signals, and the team can then switch assets between equities and cash depending on these indicators of how markets looks set to move.
It has focused so far on equities. Equity securities took up around 8.51% of Muang Thai Life's investment portfolio at the end of September.
“The way we have tested it over the past 10 years was that when we assumed that it's risk-off, we go to cash; so switching between cash and the equity market,” Sutee Mokkhavesa, senior executive vice president of risk, strategy and transformation at Muang Thai Life told AsianInvestor.
“What we are really interested in doing is quantitative investments, and really using deep learning to try and see if we can find alpha in the market,” he added.
According to Sutee, the conclusions so far had been encouraging. The investment team used the AI to calculate that there had been at least 350 basis points of potential unrealised alpha over the past 10 years in the Thai equity market, he noted.
Most recently, Muang Thai Life moved to set up a sister company called Aigen – short for artificial intelligence generation – in November to assist with the AI and broaden the scope for analysis. The sister company next aims to advance from analysing Thai equities to global equities, said Sutee.
“Once we show that after five years that it really does work, then we may implement it in our equity strategy,” he said.
In other words, the AI project needs two more years to prove its worth. Foreign equity securities took up around 3.55% of Muang Thai Life's Tb469.5 billion investment portfolio.
The Thai life insurer’s pursuit offers the latest example of Asian asset owners ramping up their efforts to use technology to enhance their investment efficiency.
For Thai insurers, AI investing solutions could help them cope with demographic challenges and market volatilities.
Sutee said the project aims to supplement the investment team’s decision-making. He said the project has already used techniques including machine learning to detect fragile spots in the market and has made use of a technique in network theory to find out how much one equity’s price movement can affect others.
“The system gives the investment risk team one more piece of information so that we can lower the risks, especially when the signal is able to predict big market downturn quite well,” he said. “So maybe we can hedge our equity portfolio or we can even lower the equity allocation.”
The AI project was developed by three Muang Thai Life executives, one of whom had an investment management background. The insurer is looking to add more talents to the team. Aigen is now in search of a director of software engineering and machine learning engineer, according to Jobsdb.
However, Sutee conceded that the biggest challenge facing their AI implementation plans now is data availability.
“In order to build an AI platform, you need lots of data, and especially … the right data,” he said. To solve this, the insurer has been working with partners, but Sutee did not elaborate who these partners are.
Other asset owners have taken a similar approach. Japan’s GPIF, for example, partnered with Sony CSL for its AI project. The project leader, Takao Tajiri at Sony CSL recently told AsianInvestor that the project has widened the prototypes dataset as it tried to quantify the extent to which fund performance is related to skills versus market movements.
For more information on how asset owners and fund managers in Asia Pacific are using technology for portfolio management, look out for our coming webinar ‘New technologies for a new investment era’.
The webinar will be held on Wednesday 26 February at 11.00am Hong Kong time. Click on this link for more information and to register to listen free of charge.