Why Wall Street Is Now Selling More Data, Less Analysis

The increasing importance of algorithmic and artificially intelligent decision-making systems for investors and managers means that users of information are less interested in what they consider opinion, and more in high quality data which they can apply themselves.

And for which they are willing to pay a premium. JL

Telis Demos reports in the Wall Street Journal:

Banks for years have crunched data on company earnings, price targets and other metrics for clients who might use the information to make investing decisions. Now they are pulling data from social-media sentiment, geospatial mapping and other unorthodox sources (and) making their data feeds available directly to clients. “People say data is the new oil, but there is a refiner needed, and we’re an integrated refiner of that data.”
Wall Street analysts are doing data differently.
Banks for years have crunched data on company earnings, price targets and other mundane metrics for clients who might use the information to make investing and trading decisions.

Now they are pulling data from social-media sentiment, geospatial mapping and other unorthodox sources. They are also increasingly making their data feeds available directly to clients, without the surrounding research notes that often go unread.
The changes are the banks’ latest strategy to try to juice up interest in—and revenue from—their giant research arms that are struggling to stay relevant. Banks have long provided research as part of a bundle of services to trading clients, but now many clients are either pushing to lower their trading bills or are more likely to base their decisions on quantitative algorithms than qualitative research. Only about 21% of research emails tracked by Street Contxt, a startup that distributes and tracks Wall Street content, were even opened during the second quarter.
UBS Group AG recently spun out the data team it built for its research business into a separate unit inside the bank. Research analysts will continue to use the service, but now clients can also tap into it directly.
“People say data is the new oil, but there is a refiner needed,” said Barry Hurewitz, UBS’s global head of the group, called Evidence Lab. “And we’re an integrated refiner of that data.”
The unit houses hundreds of data experts and specialists that consume thousands of raw data sources and shape them into usable information for investors. For example, the group analyzed job reviews on the website Glassdoor for several insurers, to judge which might lose underwriters. It also tracked Google searches for the Netflix Inc. show “Marvel’s Luke Cage,” and found less interest in the show when it made its second season debut.
Before UBS, Mr. Hurewitz worked at Morgan Stanley , where just over a decade ago he helped start a data unit now called AlphaWise.
AlphaWise, which today employs more than 100 data scientists and other engineers, is part of Morgan Stanley’s research unit. Research analysts tap it for its analytics services and nontraditional data to formulate views on the industries they cover.
In a recent note, Morgan Stanley’s real-estate investment analysts cited road-network maps to determine which malls had more or less spending power from nearby potential customers, based on how long it takes to drive to the mall rather than just a simple radius. The analysts hypothesized that people would drive to a better mall, but only if it was less than five minutes further away.
The group’s data resources are now increasingly shared across the firm, including with Morgan Stanley’s trading desk, which employs its own data experts that work with clients.
At HSBC Holdings PLC, research analysts recently used software to read the transcripts of 20,000 corporate earnings calls to spot trends.
For example, the software could pinpoint when executives talked about business difficulties, and distinguish that from talk of “technical difficulties” with the conference call dial-in. The bank is exploring how to keep running the software to generate ongoing alerts for clients.
But some trading desks don’t want to handle new kinds of data directly. One concern is verifying that the data don’t contain personal information. Banks don’t want to get pulled into the privacy controversies engulfing social media firms.
Research arms gained notoriety after some analysts leaked private information to clients in the 1990s, spurring a regulatory crackdown. Since then, the units have pulled other levers to try to gin up business. Analysts now often arrange meetings for clients with top executives at companies they cover. Some banks recently began collecting data on their own clients, like what research they click on and when. Banks say the information lets them better price their research products, but the tracking has rankled some clients who don’t want to be watched.
Jefferies Financial Group Inc. has made itself more of a conduit between clients and data. A team on its stock-trading desk advises clients on finding, evaluating and using new data sources, and the bank sponsors conferences connecting investors and data vendors. Jefferies’ parent group separately owns M Science, a major seller of alternative-data based research.
Other banks believe their edge is offering up their own data. Goldman Sachs Group Inc. has long collected data that informs its own business decisions—such as specialized measures of volatility—into a centralized repository. It would occasionally share some of that data with clients, typically in spreadsheets.
Now Goldman is providing more of that data to clients in a formal way, via real-time direct links through its Marquee web software. “Clients now want efficiency and access,” said Stacy Selig, the bank’s New York head of equity structuring.
Wall Street is also recruiting new kinds of people for these efforts. Barclays PLC over the summer hired digital media startup Buzzfeed’s principal data scientist, Adam Kelleher, to lead a new data group within the bank’s research unit.
It is a new step for both Barclays and Mr. Kelleher, who has a Ph.D. in physics. To become a research analyst, he had to take two exams with more than 300 questions.