Company-level news scores include one master score and individual supporting scores. Each score is calculated using a different linguistic classifier. A classifier is an algorithm designed to assess the overall sentiment of a stock by detecting language within stories that is likely to drive prices upward or downwards over a given period of time.

Some classifiers specialize in deriving news sentiment according to the type of story (i.e. corporate actions like earnings announcements or press releases on product recalls and layoffs). Others emu-late how a group of sophisticated analysts would likely rate a news item in terms of potential impact on price or traded volume.

Finally, RavenPack classifiers take into account how markets tend to respond historically to particu-lar types of news by mining years of news archives.

News Sentiment in Milliseconds
By leveraging diverse news sources, sophisticated machine learning sentiment engines and intuitive delivery formats, RavenPack’s news analytics represent the new alpha-capturing inputs increasingly sought after by financial institutions. RavenPack’s news sentiment accuracy and analytics have been rigorously tested and are in use by some of the world’s largest financial institutions.

Highlights
Global coverage: More than 10,000 companies

• Multi-year historical archive for back-testing
  
• Utilizes multiple linguistic analysis techniques
  
• Low latency news event processing
  
• API access to news scores in real-time
  
• Fully hosted solution (Software as a Service)