Effects of Event Sentiment Aggregation: Sum vs. Mean

RavenPack | May 15, 2018

In this paper, we introduce alternative ways to aggregate daily news sentiment and to incorporate event volume into a trading strategy.

Executive Summary:

In particular, we investigate how subtle differences in the way we use the same news information, such as different averaging, can lead to important differences in the characteristics of our portfolio. We recommend analyzing the Sum Excess Sentiment Indicator, SESI, which can be interpreted as a simple way to take into consideration event sentiment and event volume, simultaneously.

We find that:

  • SESI doubles Annualized Returns compared to using simple average sentiment (ESI [1-6]), while maintaining similar, slightly higher Information Ratios. This result is consistent across all holding periods and universes – except for EU Small-cap, where both indicators show similar performance.
  • Ranking by SESI’s magnitude not only provides value for selecting stronger trading signals, as it did abnormal news volume (Event Buzz) [6], but also does it with more balanced and diversified portfolios, characterized by more symmetric and uniform allocation distributions.
  • Driven by its better performance, SESI also provides better portfolio scalability than ESI, especially for Large/Mid-cap universes, where scalability is particularly strong. For US Large/Mid-cap, profits are still monotonically increasing for $1 billion AUM.

Sum vs Mean

Click here to continue reading the "Effects of Event Sentiment Aggregation: Sum vs. Mean" White Paper

Request White Paper

Request a Trial

Fill out the form below and see RavenPack in action.