The use of ML with text data for algorithmic trading relies on the extraction of meaningful information in the form of features that directly or indirectly predict future price movements. Applications range from the exploitation of the short-term market impact of news to the long-term fundamental analysis of the drivers of asset valuation. Examples include the following:

• The evaluation of product review sentiment to assess a company's competitive position or industry trends
• The detection of anomalies in credit contracts to predict the probability or impact of a default
•  The prediction of news impact in terms of direction, magnitude, and affected entities

JP Morgan, for instance, developed a predictive model based on 250,000 analyst reports that outperformed several benchmark indices and produced uncorrelated signals relative to sentiment factors formed from consensus EPS and recommendation changes.