Some of Canada’s big banks are turning to Big Data to try to rejuvenate their capital markets research departments, using tools most commonly associated with quantitative hedge funds.
Advanced analytics, such as satellite-image intelligence, consumer trends gleaned from web scraping and investor sentiment indicators pulled from social media, have popped up in sell-side analyst reports in recent months.
This is a big leap for an industry that needs to evolve beyond backward-looking analysis and postearnings commentary, said Carl Kirst, director of U.S. equity research at BMO Nesbitt Burns.
“We’re trying to figure out how not to keep doing the same thing Wall Street and Bay Street have been doing for the last 20 years,” Mr. Kirst said.
As quantitative funds have long understood, the reams of data being generated by the digital economy can yield tremendous insight into conditions affecting specific industries, and how individual companies are performing.
As a complement to fundamental equity research, the hope is that alternative data can boost the relevance of an industry in decline.
The demand for brokerage research has generally been eroding since the financial crisis in 2008, as reputational challenges, regulatory restrictions, heavy competition, declining profitability and the rise of passive investing strategies have led to a culling of the ranks of stock pickers around the world.
At the same time, alternative-data strategies have been spreading beyond the hedge-fund industry and into mainstream investing.
A recent study by Greenwich Associates found that half of investment managers surveyed already use alternative data, with another quarter planning to do so over the next year.
Alternative data is anything beyond traditional financial data, such as what is found in quarterly financial statements, but that still can say something about a market or a company’s performance.
Satellite images of oil-storage tanks, for example, can suggest whether crude inventories are rising or falling. About two years ago, RBC Capital Markets partnered with Orbital Insight, a leader in that kind of geolocation data.
Energy analysis incorporating satellite imagery is now among RBC’s most-read research, said Fardeen Khan, RBC Capital Markets’ head of strategic initiatives. An RBC note from May, for example, pointed to global oil inventories “piling up,” which might be considered a bearish sign for the price of crude.
RBC has since enhanced its alternative-data capabilities to include natural language processing and web scraping.
The bank, for example, developed a sentiment and emotion tracker for Freedom Mobile, the regional carrier owned by Shaw Communications Inc.
An algorithm sifts through posts by Freedom’s customers on social-media channels such as Twitter, Reddit and Instagram, looking for indicators of positive or negative sentiment.
In July, an RBC analyst noted a distinct increase in negative posts about Freedom over the prior three months. “The negativity appears to reflect an uptick with various issues Freedom Mobile has been managing, such as poor reception, customer service and technical issues,” the analyst wrote. (Shaw said earlier this year it continues to invest in improvements in Freedom’s quality and customer experience.)
In many cases, shifts in sentiment prove to be leading indicators of changes in quarterly earnings, Mr. Khan said.
Web-scraping tools, on the other hand, which extract data from websites, were used to help RBC analysts get a handle on Amazon.com Inc.’s strategy in fighting the “delivery wars.”
Using data from Amazon’s own website, the team scanned more than 1,600 Canadian postal codes to discover the company had increased its same-day and one-day delivery in Canada by fourfold over the past four to five years.
BMO, meanwhile, said its analysts were working on 48 web-scraping projects, at last count, with another 115 in the pipeline.
There are legal considerations, however, in harvesting this kind of data. Some sites explicitly prohibit web scraping in the legal terms and conditions posted on the site.
The legalities of some kinds of alternative data are still up in the air. Who owns what data and who has the right to sell it is not always clear. Additionally, some kinds of proprietary or internal data could, in theory, constitute material non-public information.
Mr. Khan said RBC spent a year preparing to launch its alternative-data team, establishing a governance framework and getting the bank’s legal and compliance team on board.
“We're comfortable at this point that much of this stuff is going to create value for clients, and that we're not doing anything that will get us into regulatory hot water,” he said.
BMO’s Mr. Kirst said that its data-science efforts are not meant to replace fundamental research, but that this kind of quantitative analysis is quickly becoming “table stakes.”
“All of the things we have been doing, we still need to do – our interactions with management teams, our ability to project their financials. This is just an extra tool. It’s just a very, very important tool.”
Tim Shufelt