S&P to Unveil New Data Platform, New Data Capabilities

S&P Global Market Intelligence is working on a new platform, dubbed S&P Global Marketplace, which will serve as a “storefront” for the data provider. The goal is to provide clients with an easier way to explore and discover datasets from across the organization, says Warren Breakstone, S&P Global’s chief product officer of data management solutions. The launch target is set for this spring.

Through Marketplace, clients will be able to access data documentation, sample data, data dictionaries, visualization tools and, for the first time, a suite of solutions from Kensho Technologies, a machine-learning specialist acquired by S&P in 2018. As part of the initiative, the vendor is also incorporating complementary third-party alternative data into the offering, such as court-case, patent, and health care data, and S&P is utilizing its quant team to curate proprietary alt data to sit alongside its traditional data.

“Very few decisions are ever made with discrete data; decisions are made when the data converges, which is a critical component of what we’re able to offer,” Breakstone says.

As the firm ramps up its data offerings, equal efforts are underway for new types of analytics. One major focus is environmental, social and governance (ESG), particularly climate analytics. Having already added Trucost, a company specializing in environmental data and risk analysis, to its Xpressfeed delivery channel, a new environmental set of data and analytics will be released in the next two weeks. Through Xpressfeed, S&P already covers the carbon and water footprints of about 15,000 companies and models companies’ physical and transition risks.

“Think about it as trying to understand the consequences of either market, governmental, or regulatory changes. For example, what would be the earnings at risk as it relates to the carbon tax in various scenarios?” Breakstone says. “The risk of both actions they can take in various scenarios, but also—increasingly importantly—their risk of inaction. And what are the costs associated with that?”

In addition to increased structured and unstructured data and analyses, the company is furthering its use of emerging technologies. In October, S&P launched Textual Data Analytics (TDA), which applies natural language processing (NLP) to generate about 40 sentiment and behavioral scores to transcripts from earnings calls, M&A calls, and teleconferences across more than 10,000 companies. In four weeks, that capability will extend to US filings, with global filings and credit research to follow. 

TDA uses supervised machine learning, and to create the models, they were fed with 10-K filings dating back 14 years. For analyzing sentiment found in calls, they were trained using the Loughran-McDonald Master Dictionary for business and finance. To quantify behavior, S&P’s quant team first wanted to gauge the level of transparency that firms provide during calls and how that relates to stock price, says Frank Zhao, senior director of quantamental research at S&P Global Market Intelligence. As examples, metrics include language complexity and numerical transparency, which determines how often executives use actual numbers and objective figures.

Last year, S&P Global Market Intelligence and Scripts Asia entered into an exclusive agreement to deliver machine-readable transcripts of certain invite-only Japanese earnings calls. The models require some reworking and retraining to understand the new language, but the company expects to reach new content sets and geographies, Breakstone says, and that work is underway.