Some of functionality may not work while you disabled JavaScript. Enable JavaScript for better User Exprience.
Access here alternative investment news about The Rise Of Machine Learning: Why Institutional Investors Should Be Wary
LPNEWS

The Rise Of Machine Learning: Why Institutional Investors Should Be Wary

by trusted insight posted 1year ago 1235 views

Machine learning and artificial intelligence (A.I.) may seem like just buzzwords, but an arms race is quietly underway to dominate institutional investment, an industry controlling more than $40 trillion, according to Trusted Insight estimates.

Machine learning -- where computers, typically software, learn and execute complex tasks -- is an evolving technology that aims to replicate how human brains learn. Over time, this technology will become better, faster and cheaper than human labor.

While financial service providers and asset managers dived head first into these new technologies, adoption of machine learning by asset allocators is still early, at best, according to interviews with 14 senior investment professionals at foundations, family offices, corporations, health care systems and public pension funds.

The main reason is quite straightforward: machine learning is the latest step toward automating human processes; the easiest tasks to automate are those that are quantitative, and institutional investing -- particularly in the alternatives space -- is very much a qualitative, relationship-driven business.

So far, the most viable products impacting the lives of asset allocators focus on back office tasks and data aggregation and analysis.

The back office of Aflac’s investment office, managing more than $100 billion, is completely automated, which “allows us to keep our staff very efficient,” said Eric Kirsch, chief investment officer.

“Through technology, we also have access to rich data,” he said. “My investment team can slice and dice our investment portfolio using many different metrics.”

Ambitious machine learning aims to not only replace human jobs, but eventually make robots irreplaceable by humans, particularly in areas where data suffice and humans simply can’t compete with computers.

In hedge funds, for example, algorithm-driven trading, or quantitative funds, have gained popularity for its lower costs and real-time reactions to market information. Assets under management for quant hedge funds rose to more than $500 billion, or 17 percent of hedge fund market share, in 2017, according to data by Barclays Plc and reported by Bloomberg in June.

Nonetheless, LPs remain skeptical. Clark Cheng, CIO of Merrimac, said, “[algorithm-driven investment] products are designed to give an average hedge fund return, but not the top-quartile or decile, which is what most allocators are trying to achieve.”
 

"If this business was simple, and you could systematize it into some kind of algorithm, I don’t think any of us would be in business." 

 

“The more predictive quantitative models get, the more difficult it is to produce excess returns,” Robert Soros, George Soros’ son, told Bloomberg in June upon his departure from Soros Fund Management to start a new fund targeting private market investments.

Where private markets and manager selection are concerned, LPs describe their job as “a people business.” The current consensus is that humans will remain the brain of the industry long before robots take over.

One reason might be a lack of demand in that space. Computers are capable of processing high volumes of information at a high speed, which makes sense for investors that need help expediting due diligence in manager selection, but institutional investment offices increasingly don’t have this problem.

“I think one of the trends -- and it's a very clear trend I identified from talking to peers -- is this whole focus on reducing the number of manager relationships that an institutional investor has in their portfolios,” said Al Kim, director of investments at The Helmsley Charitable Trust.

The 12-person investment team at Helmsley manages $5.5 billion concentrated in fewer than 50 asset managers. The benefit of a small manager pool, Kim said, is it enables his team members to do thorough due diligence and “make a compelling case” when presenting to their investment committee, the ultimate decision maker in committing capital.

Investors also argue that technology is far behind in informing complex decisions, particularly in private markets. As mentioned previously, machine learning improves on the basis of data, but in private markets, there is simply not enough structured data.

“If this business was simple, and you could systematize it into some kind of algorithm, I don’t think any of us would be in business,” Cheng said. “Reliable, robust data is the biggest problem in the industry. Selection bias, survivorship bias and voluntary reporting of performance to public databases create flawed research results.”

Another family office investor, Biren Bhandari, director at CM Capital Advisors, mirrored Cheng’s point: “ On the private side, it would probably take a little longer because the data's not as structured.”

“Perhaps asset allocation can be quantified into an algorithm, but selecting managers in the alternative business is more difficult,” Cheng added.

Although the technology is far from perfect, service providers are optimistic about the prospect and the potential demand for machine learning by asset allocators, as market climates shift and the technology is better understood.  

 

"This is definitely an area of evolution. We continue to see it getting better, and we'll continue to explore how it's integrated into the process." 


 

Data aggregation and analysis tools will be increasingly valuable to LPs as they increase active asset management in their portfolios in the post-2008 market, particularly those with large asset pools.

“Before the Financial Crisis, most large institutional investors were relatively passive in their approach to the markets. Their investment analysis was very qualitative and much less quantitative than a typical asset manager would be today. Only in the last 10 years have we seen a paradigm shift in institutional investing toward active quantitative analysis,” said Josh Smith, CEO of cloud-based portfolio management platform Solovis.

“Institutional investors are aware of A.I. in the same way they are aware of blockchain and cryptocurrency, which is from the perspective that it is a new field, but they are not exactly certain how to implement it, at least from the perspective of their own portfolios,” he added.

While the swiftness with which these new tools and technologies will change the lives of asset allocators remains remains murky, the signs of change are on the horizon.

“This is definitely an area of evolution,” Kirsch, Aflac’s CIO, said. “We continue to see it getting better, and we'll continue to explore how it's integrated into the process. I think if you ask me a year from now, I wouldn't be shocked if there's two or three new things that we’re doing that we’re not doing today.”
 


For comments, tips and story ideas, please contact us at editorial@thetrustedinsight.com