Australia's State Super has hired Neuberger Berman LLC for an equity mandate and to help the fund accelerate development of data science and machine learning tools that can complement its more traditional investment capabilities.
The move reflects continued concerns that conventional approaches to managing the Sydney-based fund's A$44 billion ($30.2 billion) portfolio may not meet the moment in unconventional times.
"My biggest concern is what happens if the market is behaving in an abnormal way," outside of the industry's knowledge base and modeling conventions, said Charles Wu, State Super's deputy chief investment officer and general manager, defined contribution investments, in an interview.
In that regard, Mr. Wu said the emergence of negative sovereign bond yields three years ago was a warning bell.
In the current environment, "a different way of thinking is required" and machine learning — with its potential to come to a problem without prejudices or preconceptions — can help State Super's team navigate a world growing evermore different from the one everyone has been trained to think about, he said.
On May 21, citing data science capabilities as a "key differentiator," State Super awarded Neuberger Berman a mandate for a concentrated global equities portfolio of 35 stocks.
State Super selected the New York-based firm from more than 100 managers responding to its request for proposals at least partly on the strength of what Andrew Huang, the fund's senior investment manager of equities, termed a "unique and compelling" approach to having data scientists and fundamental analysts work hand in hand to add alpha.
Mr. Huang declined to reveal the size of the mandate. Lucas Rooney, Neuberger's Sydney-based managing director and head of institutional business, Australia, would only say it was "large."
The mandate is the first for Neuberger's global equity data-science integrated strategy, the culmination of three years of effort to complement fundamental analysis with capabilities to mine data for better insights into company-specific fundamentals, Mr. Rooney said.
As part of that integration, "we see data scientists go to company research meetings" and traditional analysts learning to write code and understand data sets, Mr. Rooney said.
Neuberger's data scientists, working in tandem with fundamental analysts, comb through credit card data, job postings and social media to better understand the "long-term intrinsic value" of the companies that are part of the strategy, said Hari Ramanan, New York-based managing director of Neuberger's research-centric core and thematic funds, in an interview.
Data scientists can effectively bring "power tools" to the table, capable of looking beyond key metrics such as same-store sales to, for example, a more useful analysis of those sales on a store-by-store basis, said Michael Recce, the veteran of Singapore sovereign wealth fund GIC and Point72 Asset Management who joined Neuberger in 2017 as managing director and chief data scientist.
Mr. Ramanan said his team's analysis of credit card data — including distinguishing brick and mortar vs. online transactions and using statistical inference for demographics and customer cohorts — "increased our confidence" that Nike Inc., one of the strategy's top holdings, could maintain midteen margin levels even as consensus estimates pointed to a decline to low-teen margins. Meanwhile, the Neuberger team's ability to track sales on China's TMall.com online platform allowed it to confirm that Nike's strong momentum in the country, which accounts for 20% of the company's total sales, wasn't being derailed by the COVID-19 crisis.
State Super's Mr. Wu said Neuberger's approach of using data science to empower rather than replace the firm's fundamental analysts meshes well with the superannuation plan's own approach in recent years.
In popular culture terms, it's more "Ironman," a melding of man and machine, than "Terminator," he said. The nature of the decision-making process, aimed at generating unconventional insights, "is the same," even if Neuberger does it at the company level and "we try to do it at the asset class level," he said.
Mr. Wu cited the recruitment of Mr. Huang two and a half years ago as part of a push to take State Super's machine learning efforts "to a different level," with subsequent efforts to build up the governance framework, the appointment of an independent consultant and moves to set up advisory committees. The appointment of Neuberger, "was one of the key pieces of the puzzle ... how to gel the things together," he said.
Beyond the mandate, Neuberger's data scientists will work with State Super "as we share some of our data science/machine learning experience toward improving their and our dynamic asset allocation capabilities, in partnership," said Alexander Samuelson, Neuberger's New York-based spokesman.
Computers can still enhance average human capabilities but when to apply machine learning insights, "how to make sense of it and when you should turn it on, that's what we're focused on," Mr. Huang said.
In that regard, the partnership with Neuberger will "help us generate ideas more quickly (and) implement some of these concepts without having to relearn the lessons that they've had to learn" rolling out their capabilities, he said.
Mr. Wu said the evolution of his team's build up of data science and machine learning capabilities since negative sovereign yields emerged came against the backdrop of governance changes at the fund five years ago delegating more decision-making responsibilities from an investment committee to a quartet of executives — State Super's CEO, chief investment officer, chief operating officer and himself.
As a result of those governance changes, decisions are being made "a lot closer to the ground," giving State Super a "hybrid structure" from a capital management perspective, somewhere between a traditional asset owner and a money manager, he said.
That delegation of decision-making responsibilities had considerable ripple effects, prompting Mr. Wu's team to construct its own quantitative models and market dashboards.
The deputy CIO said State Super doesn't have the resources to compete in the realm of money management with the likes of external managers such as BlackRock Inc. But if those "big league" managers are tasked with making good decisions every day, State Super, investing patient capital with a long-term, seven-year outlook, can still act to take advantage of market dislocations that offer payoffs over a one-to-three year period, Mr. Wu said.
"machine" - Google News
July 13, 2020 at 11:00AM
https://ift.tt/2WdYbmX
State Super moves to add machine learning tools - Pensions & Investments
"machine" - Google News
https://ift.tt/2VUJ7uS
https://ift.tt/2SvsFPt
Bagikan Berita Ini
0 Response to "State Super moves to add machine learning tools - Pensions & Investments"
Post a Comment