What it took to build ALTD, the Bloomberg Terminal’s new Alternative Data function

When members of Bloomberg’s Alternative Data team recently logged into a Zoom call to demonstrate the Bloomberg Terminal’s new ALTD <GO> function to a client, they were expecting to hear some positive feedback. What they weren’t expecting was to find the client firm’s entire U.S. Research team – 50 analysts – on the call, eager to see how the new tool worked.

Richard Lai, Head of Alternative Data in Bloomberg’s Office of the CTO, has discovered that a response like this is not uncommon.

“Another client has multiple investment analysts, but only one of them has a Terminal subscription,” Lai shares. “When they saw what we are doing with ALTD <GO>, they noted that this is the kind of value-add that could benefit all of their analysts.”

Earlier this summer, Bloomberg began deploying an early version of ALTD <GO>, an alternative data tool integrated into the Terminal that provides an unprecedented look into the operating performance of companies on a mere seven-day lag. The new feature, which was developed in less than a year, soft launched in late July. It allows all Bloomberg Terminal clients to see alternative data analytics alongside traditional market data, broker research, estimates and news for the first time, thereby democratizing historically inaccessible insights.

“Alternative data is all about bringing in richer data-driven insights that help analysts and portfolio managers at our clients answer questions about companies and about what’s going on in the world with more speed and depth than ever before,” says Lai. “Our strategy is about getting the highest-value alternative datasets and bringing them directly to the Terminal alongside existing research workflows. Acquiring Bloomberg Second Measure in late 2020 was the beginning of this strategy – that acquisition gave us access to consumer transaction data, which is the most in-demand type of alternative data among our clients.”

According to a report from Grand View Research, hedge funds spent billions of dollars in 2022 on alternative datasets, non-traditional collections of information that can include consumer transaction data, geolocation data, app usage, and web traffic, among other types of content. Typically used only by the most sophisticated quantitative investors and larger hedge funds to generate alpha, the alternative data market has historically commanded steep prices: Lai says the average fund spends as much as $1 million on this type of information a year, due to the rarified nature of such information and the resources needed to unlock its value.

As Arvind Seth, Bloomberg’s Head of Alternative Data Engineering & Data Science, recalls, in the early days of the iPhone, Apple wouldn’t release sales data, so analysts would purchase third-party data about phone activations and use it to predict the company’s performance.

With this new Terminal functionality, however, information about recent sales trends can be displayed alongside more traditional market insights and news items as an indicator of a company’s performance. Alternative data can help paint a higher resolution picture of the impact major events may have on a company’s sales, while tracking trends in customer engagement with a brand in near-real-time. Already, the team says, major customer firms are updating their research reports based on the information available through ALTD <GO>, which includes both consumer transaction data analytics and foot-traffic data at launch.

Building a “brain trust” through a combination of teams

Jacob Boerema, General Manager of Alternative Data at Bloomberg, likes to explain alternative data like this: If you asked him how many times he went to McDonald’s and what he ordered, he’d probably tell you he went once a week and ordered a salad. But if you could look at his credit card payments data and connect it to geolocation information, you’d find he went three times a week. It’s this gap between intermittent reporting and real-time data, he says, that ALTD <GO> bridges.

Insights like these have real consequences amid rapidly shifting markets and pivots in a company’s business strategy. Boerema and Lai use Netflix’s change to its password-sharing policy earlier this year as an example of the near-real-time insights that can be provided to Bloomberg Terminal subscribers through ALTD <GO>.

When this change was announced, the market was primarily concerned that Netflix would lose a lot of customers. Bloomberg Second Measure data showed that the opposite was true – Netflix was actually gaining new subscribers, and the company ended up reporting a strong beat on subscriber numbers in its next earnings report.

Boerema joined Bloomberg in 2020, when alternative data company Second Measure was acquired. At the time just a web application, the company deployed analytics based on billions of credit card and debit card transactions to help subscribers answer questions about consumer trends in near-real-time. By merging Second Measure’s team of data scientists with Bloomberg veterans, Second Measure’s startup mentality found extensive institutional support, Boerema says. The team became something of a startup-within-a business, moving quickly to bring ALTD <GO> to life.

The collaborative pairing of long-time Bloomberg engineers and Second Measure employees was a fruitful one. “They came together as a single brain trust,” says Lai.

“On the Second Measure side, we’re incredibly strong on data science,” explains Catharina Kronlein, Product Manager for ALTD <GO> in the CTO Office and another addition to the team following the Bloomberg Second Measure acquisition. “We have a lot of opinions about how we should calculate things and the best way we should represent data. But we’ve really benefited from Bloomberg in terms of putting all of that to paper and wrapping it up.”

To bolster the new division and help integrate the Bloomberg Second Measure team’s expertise within the Terminal, Bloomberg brought in engineers such as Peter Litskevitch, a 19-year veteran of the company, as well as Seth and his two decades of leadership at Bloomberg, to oversee the new Bloomberg Second Measure engineering team.

“Throughout my career at Bloomberg, I’ve gravitated towards startup-type projects,” says Litskevitch. “And I really loved the energy and excitement around this product.”

To kick off the process of building what would become ALTD <GO>, Bloomberg held a two-day summit in September 2022. Seth describes it as “drinking from a firehose” as new team members took a “crash course” in alternative data, aligned their product strategy, and brainstormed ideas.

Building a nimble, client-first product

Building a brand-new major function on the Terminal was no small feat. However, by December 2022, just three months after the summit, the team had a working prototype for ALTD <GO> that relied on a new and exclusive transaction dataset that Bloomberg Second Measure had launched earlier in the year, which derives aggregated analytics and insights from billions of purchases from a subset of a U.S. consumer panel that includes 20+ million members.

The rapid pace at which ALTD <GO> was built, engineers say, was largely thanks to the team’s flexibility. The Alternative Data team worked in small units of two or three people, allowing them to make decisions quickly and autonomously.

“We were able to move quickly because we made sure to have veterans who were familiar with the Bloomberg tech stack, while also using the technology Bloomberg Second Measure had already built for ingestion, processing, cleanup, and normalization of the panel data,” says Seth.

Utilizing Bloomberg Engineering’s home-grown software development workflow system with automated testing frameworks increased the speed of the application development process. This enabled the team to follow a merge-and-move-on workflow, something Litskevitch refers to as the “holy grail” of the development experience.

Fahd Arshad, the User Experience (UX) Manager for Bloomberg’s CTO Office, has been with the company for 16 years, and led the team responsible for designing the user interface (UI) for ALTD <GO>. The UX design process started early, he says, with a group of researchers who mocked up a series of wireframes based on dummy data, which they then took to clients for feedback. After a few rounds of testing, they hit on a multi-part Terminal screen that resonated with the investors most likely to use the function. But, he says, it was the close collaboration between designers, researchers, and front-end engineers that allowed for an unprecedented level of experimentation and a final product that equity analysts would find incredibly easy-to-navigate and understand.

“We were able to build a best-in-class user interface for ALTD that matches the strength of the underlying data in a way that I just haven’t seen before, and I’ve been here for a long time, working on many different things,” he says. “And that was a pretty awesome journey for us.”

So far, the response to all of this careful iteration has been “phenomenal” says Lai. Thousands of clients signed up to participate in the function’s beta test.

“In my career, I’ve scheduled a lot of client feedback sessions,” says Kronlein. “I don’t think it was ever easier than with ALTD. Our clients are very eager to partner with us to help make this better, because they’re immediately seeing its value.”

Democratizing access to new data

ALTD <GO> currently provides consumer transaction data analytics on more than 300 public tickers, primarily those with meaningful U.S. direct-to-consumer sales, and the function is already seeing real-world results. Recently, a number of sell-side analysts have published reports adjusting their price targets ahead of quarterly earnings reporting by some of the companies they follow, specifically citing trends in Bloomberg Second Measure credit and debit card data analytics that they accessed via ALTD <GO>.

However, the number of companies tracked in ALTD <GO> will grow rapidly in the coming months, and the team has already built infrastructure to significantly expand the tool’s functionality — a crucial part of the initial design process, given the number and types of different alternative data sources Bloomberg plans to make available on the Terminal.

“The product is great right now,” notes Kronlein, “but it’s very much in the beginning stages.” The team is currently working on an update to the Bloomberg Desktop API that will allow analysts to export data from ALTD <GO> directly into Excel to better integrate alternative data into their existing workflows, as well as adding additional datasets and building quarter-to-quarter trackers. And, in collaboration with Bloomberg’s Terminal Alerts & Insights team, Bloomberg’s CTO Office plans to provide more robust distillations of the kinds of information that alternative data can help surface for investors alongside company financials in other Terminal functions, like MODL <GO>, and through news articles.

For Litskevitch, part of the satisfaction is bringing a traditionally expensive and inaccessible type of information to Bloomberg’s clients. “How often do you get the opportunity to democratize a part of the financial market?” he asks. “And that’s the Bloomberg mission: bringing data and more transparency to financial markets, while democratizing them in the process. It’s part of what makes me so proud to work here.”