Make the most of every analyst call

A scalable approach to extracting alpha from the sentiment analysis of earnings call transcripts.

Earnings disclosures are critical for investors, but company executives tend to carefully choose their words on analyst calls, so it's harder to identify actionable insights. The RavenPack EI Transcripts factors streamline this process using the quantitative analysis of earnings call transcripts to produce:

As well as the Transcript Factor, a combination of all the previous factors.

Use Cases

Discretionary and quantitative investors can leverage these factors to:

Factor Performance

The performance of the Transcripts Factor is less focused on short-term market impact but fairly robust over time, with a less pronounced decay as the holding period increases compared to other components. Its performance on cumulative returns is particularly strong for Europe and APAC.

'
UNDER THE HOOD

Computing Earnings News Sentiment Factors

The Factors are the outcome of a multi-step process including:

Content Collection

RavenPack analyzes earnings call transcripts from 20,000+ publicly-listed companies, including 2000+ in Japan, China, Hong Kong, Taiwan, and South Korea sourced by SCRIPTS Asia, and global coverage from FactSet transcripts.

Analytics Enrichment

For each event identified in transcripts, RavenPack uses sentiment analysis to compute multiple scores by matching stories usually categorized by financial experts as having a positive or negative financial or economic impact.

Signal Construction

Sentiment scores and event volumes are aggregated and smoothed over time with various decays and presented as daily factors, computed using a published methodology.

Company Factors Facts
Purpose
  • Extract alpha
  • Mitigate portfolio risk
FrequencyDaily
HistorySince 2007
Universe20,000+ Listed Companies worldwide
Content
  • When the factor was produced
  • The company, as name and RavenPack entity identifier, and ISO country code.
  • Exponential smoothing decay — the number of days used for exponential smoothing, with 1, 7, and 30 days provided.
  • Moving average window for the delta calculation: 90 or 180 days.
  • Transcripts event sentiment score, averaging the sentiment of key events identified during the earnings call.
  • Transcripts document sentiment score aggregating the sentiment of each sentence transcribed from the call.
  • Transcripts transparency score which quantifies the level of transparency observed in transcripts, based on the number of market-moving events disclosed in the call.
  • Transcripts, the numerical score value for the combination of all the previous signals.
Delivery RavenPack, Snowflake

Access the Earnings Intelligence Transcripts Factors


By submitting this form, you agree to RavenPack's terms of service and privacy policy.