Abstractive summary of Earning Call

Earning call analysis (Public info)

Paper: Template aware earning call summarization (Prashant paper published)

Problem

• Earning calls are an important resource for investors and analysts for updating the price target of stocks.

• Earning call transcripts are too long and do not have predefined structure. It is nearly impossible for analysts to read thousands of earnings calls for decision making.

Data Product developed

• Model : Using machine learning generates abstract earning call reports. Our machine learning model is published here - Template aware earning call summarization .

• Operationalization: As soon as the earning call is published, within 1 minute the pipeline fetches a transcript and generates a hierarchy summary.

• Interface: Enable users to read summary/analysis and full transcript.

• Other: Use a variety of data augmentation approaches. Factual analysis ensures integrity of facts presented.

Outcome : new dataset that summarize a earning call.

Expected benefits

• Instead of spending a day, analyst can gain insight in few minutes

• Scale to thousands of earnings call

• Competitive advantage