Parametric adopting AI, machine learning for its back office

Parametric Portfolio Associates LLC is planning its first use of artificial intelligence and is looking at a machine learning application to help the $240 billion Seattle-based firm better manage its complex back- and middle-office functions, said Brian Langstraat, CEO.

"We have our first (AI) project being scoped right now," Mr. Langstraat said in a telephone interview. "One of the things that's unique about Parametric is we manage somewhere between 35,000 and 40,000 individually managed customized separate accounts. The care and feeding of those accounts and identifying those that need attention and arriving at the (solutions) is in effect a factory-like, industrial process. We think there's some applications for machine learning and artificial intelligence in helping our teams identify which portfolios need attention when."

Mr. Langstraat said the AI project is "early in the first inning … It hasn't really launched yet; the project is being spec'd by one of our longtime engineers here and we have to approve it."

The money manager also is looking at the potential use of robotic process automation, Mr. Langstraat said. "In a high-volume back office where you have a lot of workflow processes connecting reconciliation systems with custodians and trade settlements, there's inevitably relatively routine and repetitive data-processing functions that are performed by human beings," he said. "The robots are not physical robots, but they're software programs that can help monitor, observe and take over some of the most repetitive parts of that."

Mr. Langstraat gave an example of a possible robotic process automation application: "Every day your job is to log on to a system, download a file, manipulate that file for exceptions, reformat it and upload it into another system, which probably happens here and in a lot of other places thousands and thousands and thousands of times. The (RPA) systems can help do that and free up the human resources to do something much more valuable. This is completely different (from AI), there's nothing artificial intelligence about it. The people in the business call these 'swivel-chair functions,' as in you're at one terminal looking at one (piece of) data and then swivel over and type it into another terminal. The robotic process replaces that function. Maybe that can be reduced to a series of rules."

Parametric is going to test RPA functions "over the next six months," Mr. Langstraat said.

Of Parametric's $240 billion in assets under management, $120 billion is for institutional investors and the rest is managed for private and high-net-worth clients.

Mr. Langstraat said AI and RPA are applicable to Parametric's investments — overlay, liquid alternatives, factor-based and systematic alpha strategies. "I think they become more attractive to a manager who has a high-volume processing environment, a complex, diverse client base, custodial connections," he said. "The (return on investment) is less clear and present if you're managing a couple pools of long-only equities because your operations and administration burden is lower on your (profit and loss)."

Technological applications serve multiple purposes, Mr. Langstraat said, not just in efficiencies and cost savings in an increasingly fee-constrained business environment, but also in providing opportunities for back-office employees to advance at the company.

"We want to operate with as few errors, protecting our financial resources," he said. "We want to offer our employees work that they find varied and meaningful, with opportunities to increase their skill sets and retain them as employees as they move up the experience and value chain. The drive to become more efficient is enhanced or amplified in a declining fee environment, but it's always there. Any for-profit business is attracted to a proposition of increasing its efficiency."

Mr. Langstraat described Parametric "as much an implementation engineering firm and a technology firm as we are a money manager."