Startup digs into public records with GPU-driven machine learning to offer alternative financial data services

When Rachel Carpenter and Joseph French founded Intrinio ten years ago, the fintech revolution was just beginning. But they saw an opportunity to apply machine learning to large amounts of financial documents to create an alternative data provider among the giants.

The St. Petersburg, Florida-based startup provides financial data to hedge funds, proprietary trading stores, retail brokers, fintech developers and others. Intrinio runs machine learning on AWS instances of NVIDIA GPUs to analyze mountains of publicly available financial data.

Carpenter and French realized early on that this data was being sold at a premium, and that machine learning offered a way to sort through free financial documents to come up with new products.

The company offers information on stocks, options, estimates and ETFs, as well as environmental, social and governance data. Its most popular product is data on equity fundamentals.

Intrinio took an unbundling approach to traditional product offerings, creating a la carte data services now used in some 450 fintech applications.

“GPUs have helped us unlock otherwise expensive, manually obtained data,” said company CEO Carpenter. “We built a lot of technology with the idea that we wanted to unlock data for financial services innovators.”

Intrinio is a member of Creation of NVIDIAa free global program designed to support cutting-edge startups.

Partnership with Fintechs

With reduced overhead using GPU-driven machine learning to deliver financial data, Intrinio has been able to offer products at lower prices that appeal to startups.

“We have a much smaller, more agile team because a small team — in conjunction with NVIDIA GPUs, TensorFlow, PyTorch, and everything we use — makes our work much more automated,” she said.

Its clients include fintech players like Robinhood, FTX, Domain Money, MarketBeat and Alpaca. Another, Aiera, transcribes live earnings calls with its own automated speech recognition models driven by NVIDIA GPUs and relies on Intrinio for financial data.

“Our use of GPUs has made our data packages affordable and easy to use for Aiera, so the company integrates Intrinio’s financial data into its platform,” Carpenter said.

Aiera needed financial data cleansing services to get consistent information on company earnings and more. By leveraging Intrinio’s application programming interface, Aiera can access standardized corporate financial data in a fraction of a second.

“GPUs are an essential component of Intrinio’s underlying technology. Without them, we could not have applied machine learning techniques to cleaning and normalizing fundamental data and financial statements,” said Carpenter.

Equity, Options, ESG Management

For stock pricing, Intrinio’s machine learning technology can resolve price discrepancies in milliseconds. This translates to significantly higher data quality and reliability for users, according to Carpenter. With Equity Fundamentals, Intrinio automates several key processes, such as entity recognition. Intrinio uses machine learning to identify company names or other key information from unstructured text to ensure the correct categorization of data.

In other cases, Intrinio applies machine learning to reconcile financial statement line items into standardized buckets so that, for example, you can cleanly compare revenue between companies.

Using GPUs and machine learning in both of these cases results in higher quality data than a manually oriented approach. Using Intrinio showed an 88% decrease in the number of errors requiring correction compared to manual sorting, according to the company.

For options, Intrinio uses the raw feed from the Options Price Reporting Authority (OPRA) and applies state-of-the-art filtering, algorithms and server architecture to deliver its options API.

ESG data is also an area of ​​interest for investors at the moment. As retail investors begin to be more environmentally conscious and institutions feel the pressure to invest responsibly, they want to see how companies fare with this information.

As regulations around ESG disclosures solidify, Intrinio says it will be able to use its automated XBRL standardization technology to unlock these datasets for their users. XBRL is a standardized digital information exchange format for businesses.

“On the retail side, app developers need to show this information to their users because people want to see it — making this data accessible is critical to the evolution of the financial industry,” Carpenter said.

Register for free for GTC, online September 19-22, to attend sessions with NVIDIA and dozens of industry leaders. To see the financial services agenda for the conference.

Image credit: Luca Bravo from Unsplash

Comments are closed.