Available Datasets
The following LSEG MarketPsych feeds are available on the Databricks Marketplace. All datasets are point-in-time accurate and suitable for both backtesting and live signal generation.
Structured sentiment and topical indicators derived from global news and social media, covering 130,000+ entities across equities, fixed income, commodities, currencies, cryptocurrencies, and country-level markets. Data begins in 1998 and updates every 60 seconds.
- 50+ sentiment scores per asset at minute, hourly, and daily intervals
- 4,000+ curated sources including wire services, social media, and blogs
- Fundamental themes (earnings, inflation, interest rates) and emotional indicators (optimism, fear, uncertainty)
- Spam detection and source deduplication applied prior to scoring
Update Frequency
Every 60 seconds
Asset Classes
Equities, FX, Commodities, Crypto, Macro
ESG sentiment and controversy scores derived from external media coverage rather than corporate self-reporting. Provides an independent, real-time perspective for detecting greenwashing, controversies, and ESG risks not visible in company disclosures.
- 120,000+ public and private companies and 252 countries
- 4,000+ ESG dimensions scored daily and at minute-level
- Environmental, social, governance, controversy, and buzz metrics
- Derived from 2 million+ articles and posts daily in 12 languages
- History from 1998
Update Frequency
Daily and minute-level
Coverage
120,000+ companies, 252 countries
NLP-scored earnings conference call transcripts providing sentence-level, speaker-level, and document-level sentiment and thematic data for 18,000+ global public companies. History from 2001.
- Sentence, speaker, and document granularity
- 13 emotional tone dimensions per utterance
- Financial and ESG sentiment polarity
- 20+ entity types, 1,000+ topics, and 4,000+ events tagged per transcript
- Document types: earnings calls, guidance calls, investor presentations
- Standard (aggregated) and Premium (raw NLP output) tiers available
Company Coverage
18,000+ global public companies
A daily 1-100 ranking score for 16,000+ global equities built entirely from media sentiment data, co-developed with the LSEG StarMine team. Forecasts relative stock performance over a 30-day horizon. Live since January 2020.
- Three sentiment dimensions: Equity (price action and investor sentiment), Business (fundamentals and earnings), Management (leadership and corporate events)
- Scores are uncorrelated with traditional factors (size, value, momentum) and other StarMine models
- US model: market-neutral annualised return of 10.9% and Sharpe ratio greater than 1 since live launch
- Geographic coverage: U.S., Europe, Asia, Emerging Markets, Japan, Canada
Universe
16,000+ global equities
Unified access to all MarketPsych data feeds in a single subscription. Data is mapped point-in-time to organizational PermIDs for consistent entity linkage across products. Intended for backtesting and signal research requiring 20+ years of history across multiple data types. Note that data is subject to a 3-month embargo and reflects a rolling window ending 3 months prior to the present date.
- MarketPsych Analytics
- MarketPsych ESG Analytics
- MarketPsych Transcript Analytics
- StarMine MarketPsych Media Sentiment Model
Entity Mapping
Point-in-time PermID
Data Embargo
3 months behind present