JPMorgan is quietly building what it calls a 'next-generation prediction platform' and it's betting eventually the whole world will use it

JPMorgan's investment bank is working on a new platform that would use crowdsourcing to accumulate massive amounts of data in a bid to help its clients make better business decisions.
The platform, called Roar by JPMorgan, is also meant to allow clients to exchange and safely store sensitive data, such as hospital records or satellite imagery, that's not in the public domain.

JPMorgan's corporate and investment bank is best known for advising businesses on billion-dollar acquisitions, helping private unicorns tap into the public markets, and managing the cash of Fortune 500 companies.

But now it is quietly working on a new platform that would go far beyond anything the firm has previously done, using crowdsourcing to accumulate massive amounts of data intended to one day help its clients make complex decisions about how to run their businesses, according to people familiar with the project.

For JPMorgan's clients like asset-management firms and hedge funds, it could provide new data sets to help investors squeeze out more alpha from their models or better price assets. But JPMorgan is looking to go beyond the buy side to help its large corporate clients as well. The platform could, for example, help retailers figure out where to build their next store, inform manufacturers about how to revamp systems in their factories, and improve logistics management for delivery services companies, the people said.

The platform, called Roar by JPMorgan, would store sensitive private data, such as hospital records or satellite imagery, that's not in the public domain. Typically, this type of information is exchanged between firms on a bilateral arrangement so it is not improperly used. But Roar would allow clients to tap into this data, which they could then use in a secure fashion to make forecasts and gain business insights.

Roar is being spearheaded by Peter Cotton, an executive director at the firm, under Samik Chandarana, a 19-year veteran of JPMorgan who was appointed in 2017 to a newly created position to unleash machine learning and data solutions onto the corporate and investment bank.

In a sense, the platform bears some resemblance to IBM's Watson, which culls through loads of data to make recommendations for businesses. While Roar would aim to solve similar problems for JPMorgan's clients, it's better thought of as a marketplace for data, according to people familiar with the project.

It's also a different mission than that of the bank's JPMorgan Chase Institute, which bills itself as a global think tank that develops analysis and insights using its own proprietary data for economists and policymakers.

Roar is still in the early stages of development and is being operated essentially as a startup within JPMorgan. The team right now is focused on creating the right types of security safeguards for the platform and has recently been seeking to hire data scientists and cryptographers.

In a recent job ad for a Roar engineer, the bank says it's trying to build "a next-generation prediction platform used by the bank, the bank's clients, and eventually the entire world."

Right now, the platform is being tested internally with public data and JPMorgan is collaborating with academics to answer questions such as predicting traffic patterns or future air pollution.

"Every single one of the big banks is trying to use alternative data sources, whether it be external or internal," Erkin Adylov, the founder of the data provider Behavox, said in an interview. "It is a rat race and everyone is trying to take a different approach."

JPMorgan isn't the only financial-services firm looking into new data projects. Goldman Sachs has also been building out a team to sell its own alternative data, as reported by Risk.net.