Information being used as part of fundamental analysis for investing
By Rick Baert
Location, location, location.
That’s not just a catchphrase in real estate, it’s data some active fundamental money managers are using to help determine whether to buy or sell securities.
And the information, called location-based data, is supplementing traditional sell-side analysis of securities — something not lost on money managers in the era of research cost concerns surrounding MiFID II, sources said.
“It’s a new wave, a new way of doing fundamental investing,” said Michael Recce, chief data scientist, Neuberger Berman Group LLC, New York. “This data provides a higher-resolution microscope to see things and make them more visible. There’s an advantage to seeing this information earlier and in more detail.”
Added Mark Ainsworth, head of data insights at Schroder Investment Management, London: “We make extensive use of location-based data in active fund management, both in equities and fixed income. Analysts and fund managers look at our data reports to assist them in management decisions. Our role is boiling down key points to make the information not too big or too complex.”
With location-based data, the anonymous tracking of smartphones based on global positioning satellites and cell towers, as well as information gathered from satellite photos, provides not only information on the number of people at retailers, but also how many are working in factories and at what times of day.
Data scientists then narrow the information to specific companies to determine information on things like same-store sales estimates in nearly real time — as opposed to at the end of a month or quarter — or whether manufacturers are increasing production before a business expansion is announced.
According to a report released April 27 by Opimas, a Boston-based capital markets consultant, the market for mobile device location data in the money management industry is expected to grow 40% annually to $250 million by 2020. That includes spending on data sources, predictive analytics, infrastructure and related data management, according to the report, “Generating Alpha with Mobile Device Geolo-cation Data.”
“In one hour, we have data on about 500 million phones worldwide,” said Greg Skibiski, founder and CEO of Thasos Group, New York. The firm takes the real-time location data, analyzes it and then sells that information to clients that include money managers. Mr. Skibiski said no personal information is gathered.
“Nobody wants raw data,” Mr. Skibiski said. “They want data that’s easily understood … We’re one piece of the puzzle that people use to confirm whether to invest or to understand the risk of an investment. If a business metric is turning down, managers and traders can take some action. Location data help the buy side fill in the missing pieces in their research. Location data is special because it’s a completely independent source of information that can supplement other data sets,” like credit-card transactions, he said.
Hedge funds and managers of real estate investment trusts have used Thasos data for short-selling and mall traffic measurement, respectively, for several years, Mr. Skibiski said.
But now fundamental managers and their trading desks are looking to use location-based data as a research tool in their long-only strategies as a partial supplement for sell-side analysis — particularly as managers monitor their research budgets after the Markets in Financial Instruments Directive II regulations went into effect in January, he said.
MiFID II “is a real boon for alternate analysis providers,” Mr. Skibiski said. “We never saw that coming.”
Schroder’s Mr. Ainsworth said his team, which buys cleaned up data and then analyzes it for application to the firm’s strategies, is in competition with sell-side analysts for the attention of Schroder investment professionals. But it doesn’t provide buy, sell or hold recommendations.
“There’s no hard numbers on substituting location data for sell-side analysis,” Mr. Ainsworth said. Parent Schroders PLC buys the data with hard dollars on behalf of the entire company, he said, adding: “We’re competing for the attention of investment professionals with the sell side. It doesn’t feel like it, because nothing we gave our portfolio managers came with a recommendation to buy or sell. It’s structurally different from the sell side, which does recommend whether sell, buy or hold. We don’t make those recommendations. The portfolio manager and analysts decide.”
Mr. Ainsworth, who has run Schroder’s data analytics unit since 2014, leads a team of 25 data scientists that works closely with portfolio managers and analysts for all of Schroder Investment Management’s fundamental active equity and fixed-income strategies as well as credit, multiasset, global macro and real estate strategies.
Schroder’s data insights unit has three main responsibilities, Mr. Ainsworth said: to provide insight to portfolio managers and analysts; to translate “big, messy information” into targeted and understandable information; and to work regularly and directly with the firm’s investment professionals.
“You need to know what actually matters to that industry, that company, that strategy,” Mr. Ainsworth said. “We depend on working in partnership with (portfolio managers) to give them what they need to make educated investment decisions.”
Added Neuberger Berman’s Mr. Recce: “The sell side is not equipped to think about these things in all these ways. Think of data as allowing you to reconstruct a dashboard of a firm to see how healthy it is. The sell side can charge $10,000 to talk with a company CEO and get his take on what’s going on with the company, but what would you rather do: Talk with the CEO or read his dashboard? That can provide more information on what a company is worth and you can still calibrate it with what the sell side says.”
Neuberger Berman investment professionals still work with sell-side analysts and meet with company CEOs, Mr. Recce added.
Mr. Recce, who joined Neuberger Berman in December, said location-based data lessens the possibility of “surprises” when company earnings are announced.
“If you follow a sector where there has been a large amount of surprises, you can look at data that overlaps other analytics to lessen the possibility of those surprises,” Mr. Recce said. “But we’re long-term holders. Here’s how we’d use the data: I tell people to think about this like you were buying a private company; think about other aspects of the business beyond the traditional.”
Location-based data aren’t perfect, Mr. Recce added. “It’s a little bit early days,” he said. “There are ways it can go wrong. In one case, there was data that showed peak visitors in a Home Depot in California … (came) at 8 a.m. The store opened at 10 a.m. But that store has a loading dock that abuts a freeway. The data was also including commuter traffic from passing cars. Even if you stand still with a cell phone, the signal can hop from tower to tower. That can change the information. It’s a lot harder to get this data right than some people think.”
Another issue that’s yet to be addressed is what the impact of the use of location-based data is on a portfolio’s returns. “That’s not something we have measured,” Schroder’s Mr. Ainsworth said. Parsing out data-based returns and those from fundamental active portfolio managers could “endanger the relationship between data scientists and portfolio managers, who want the independence to take risks. Taking that away from them (by detailing the benefits of data use) creates an adversarial relationship,” Mr. Ainsworth said.
Michael Falk, partner at money manager consultant Focus Consulting Group LLC, Long Grove, Ill., said there is “no documented data out there that shows the efficacy of alternative data sets.”
But that might not always be the case.
“What we don’t know is if that answer today will be the answer tomorrow,” Mr. Falk said. “Given the speed and extent of technological development today, to say we couldn’t do better in the long term is being shortsighted … There’s the potential that we can know in the future, but where we are today, we can’t say alternative data sets have generated returns.”
SIDEBAR: Location Data Provides New Insight Into Investments
Location-based data are being used by fundamental active money managers in a variety of ways to help decide whether to move on a stock or bond — often in ways that aren’t otherwise obvious, sources said.
“Basic data would be one guy at a McDonald’s,” said Michael Recce, chief data scientist at Neuberger Berman Group LLC, New York. “But with more data, you could find out that guy is at McDonald’s three times a day. That would show brand loyalty. If McDonald’s has 10 times growth in sales year over year, in financials, you don’t know if that sales growth is from new customers or loyalty. With location data, this provides one to two degrees more information, so you can determine how much of the business is from new customers and how much from existing customers.”
Greg Skibiski, founder and CEO of Thasos Group, a New York-based provider of analyzed real-time location data sold to money managers, gave three hypothetical examples of location-based data use by investment professionals:
- Monitoring employment at manufacturing plants to forecast whether the stock of the companies could go up if production increases or down if the number of workers declines;
- Gauging the number of people at mining facilities and quarries to determine if production is increasing, and if that is reflected in the current stock price; or
- Comparing the activity in retail stocks with the number of shoppers at retail outlets. If shares are down but store traffic is up, portfolio managers might use that as a reason to increase their holdings, hoping to cash in if the price goes up when same-store sales figures are released.
At Schroder Investment Management, Mark Ainsworth, London-based head of data insights, provided examples of how the firm’s portfolio managers acted on location-based data. In one, Schroder decided against investing in a retailer that had intended to do an initial public offering for its Brazilian operations after geospatial analysis of location data showed that many of the company’s stores were located in close proximity to its competitors. “We laid out precisely how many stores there were near the competitors’, and it showed the level of competition was much higher than thought, so the team avoided the IPO.”
In fixed income, Mr. Ainsworth said, the data were used to see if the firm would buy debt being issued by a retail pizza chain to pay for store expansion in the U.K. The data allowed Schroder to determine whether the targeted growth areas would put the chain in direct competition with other casual dining restaurants and if so, whether there was enough foot traffic in the area that it could attract new business. Schroder eventually purchased the debt.