LSEG Data & Analytics has released MarketPsych Transcript Analytics, a data and feeds solution that turns corporate transcripts into structured intelligence for institutional and quantitative investors. Built on advanced Natural Language Processing (NLP) and developed in partnership with MarketPsych, the new offering delivers granular analytics on over 16,000 global public companies.
The feed applies a proprietary NLP model, based on a RoBERTa architecture, to analyze spoken content from earnings calls, guidance updates, conference presentations, and other corporate events. It assigns sentiment and emotion scores to each sentence and tracks the frequency and context of over 1,000 topics and 4,000 events across business, finance, and legal domains.
“Fast, systematic way to integrate the voices of corporate leaders, institutional investors, and research analysts”
Eric Fischkin, Director at LSEG Data & Analytics, commented, “LSEG MarketPsych Transcript Analytics gives clients a fast, systematic way to integrate the voices of corporate leaders, institutional investors, and research analysts into their strategies. By applying the latest advances in Natural Language Processing and MarketPsych’s decades of experience transforming financial text into systematic inputs, the solution empowers users across investment management, ESG analysis, and risk management to uncover hidden signals and behavioral patterns at scale.”
The platform identifies more than 20 entity types, including companies, currencies, commodities, and countries. It quantifies financial sentiment and ESG-related tone at the sentence level, while also capturing emotional cues across 13 dimensions such as fear, optimism, and annoyance. This enables asset managers to detect shifts in corporate messaging and align trading strategies accordingly.
Available in both Standard and Premium versions, the analytics can be accessed via API in JSON or CSV formats. The Standard package includes document-level scores and summary aggregations, while the Premium version adds raw sentence-level outputs, complete with quoted text. Both formats support search by entity, topic, and time range, and are integrated into a web-based user interface for direct navigation and filtering.
The data is event-driven and begins with historical coverage from 2001 onwards. Each corporate transcript is broken down from sentence to document level, offering insight into how management sentiment correlates with stock performance. Internal analysis shows that companies ranked in the top 10 percent for positive sentiment in their earnings calls consistently outperformed peers over a 21-year backtest.
LSEG MarketPsych Transcript Analytics is the latest extension of LSEG’s broader quantitative data services. It builds on the company’s long-standing partnership with MarketPsych, which also includes ESG sentiment, news analytics, and predictive modeling. The transcript feed is intended for use in fundamental analysis, quantitative alpha generation, and proactive risk management.
By embedding sentiment and emotional intelligence into machine-readable data, LSEG aims to equip investors with a sharper lens on the qualitative aspects of corporate communication, at a time when language and tone play a growing role in investor decision-making.
Read more on The Industry Spread
