Machine learning could be adapted by superannuation funds in Australia as the next phase of the evolution of selecting investments and portfolio construction.
Seth Weingram, Vice President of Acadian Asset Management, had been touring Australia speaking to superannuation funds about improving using machine learning.
Weingram said there were two different elements superannuation funds should look at when assessing the application of machine learning to the management of their funds.
One of them would be the extent super funds are investing in external managers and if they’re effective.
The second was if there was a subset of asset owners who are thinking about employing machine learning in various parts of their own process.
“How can they judge the reality of the situation from hype or from slick presentation? We talked about that explicitly in these conversations with superfunds,” Weingram said.
“What are the types of questions you should be asking? Or what are the things that you should be looking for? How should you be tuning your expectations in these types of conversations?”
Slightly different to artificial intelligence (AI), machine learning used algorithms to spot complex patterns in investment data, which helped funds be more efficiently managed.
“Our perspective is AI is the concept of developing machines that make decisions at least as well as human beings,” Weingram said.
“Machine learning is essentially means to that end, it’s using a data driven algorithmically orientated approach in order to train computers to make those decisions.”
For example, if trying to predict returns on an asset, there would be variables and attributes of data that would be used to make that prediction.
“You would define the general concept of a predictive algorithm and you would allow a computer to essentially train the algorithm in using data where you know what the answers are using historical data,” Weingram said.
“It’s where you know the values of what you're trying to predict, in order to come up with the most effective model for making those predictions.”
Acadian Asset Management started in Boston in 1987 specialising in systematic investment strategies, with now $100 billion funds under management, including just under $10 billion in Australia.