Construction of Market Sentiment Indices
Using News Sentiment
Published Date: Aug 28, 2009
Author: Peter Ager Hafez
The study provides insights into how RavenPack News Scores can be used to construct sentiment indices
for broader equity portfolios. The results can be used as input to multi-factor models or as part of the
formulation of triggering events for high-frequency trading or longer-term investing.
The purpose of this paper is to present a methodology for constructing market sentiment indices based on
news sentiment. The market sentiment indices are to be considered a first step in creating a range of
company sentiment indices, which take into consideration not only company specific news events, but also
broader market sentiment. One way of constructing market sentiment indices is to consider ratios of positive
and negative sentiment counts over well-defined backward-looking news sentiment aggregation windows. The
resulting indices can be used either as quant factors in multi-factor models or as part of creating triggering
events applied in high frequency trading or longer-term investing. Specifically, applying the "optimal"
news sentiment aggregation window as part of a DJIA sentiment index, we arrive at correlations between the
sentiment index values and the two week forward-looking returns that range between 20 – 27%, while for
a Eurostoxx50 sentiment index the corresponding range is 9 – 15%. Furthermore, we find Hit ratios ranging
between 59 – 65% and 55 – 67% for the DJIA and Eurostoxx sentiment indices, respectively; this with
Profit/Loss ratios ranging between 1.01 – 1.27 and 0.86 – 1.18.