To bridge the gap between the quantitative community and discretionary investors, RavenPack launched the latest edition of its self-service data and visualization platform back in Fall 2017 - making it easier to create custom daily indicators on top of RavenPack’s granular data. Indicators which can be used as signals to inform trading decisions.
As part of our most recent research, we presented a compelling application of such indicators. Thanks to the increasing availability of news analytics services like RavenPack’s, markets nowadays are highly sensitive to news buzz. For this reason, it becomes of paramount importance to identify which news events are most significant to trading. Doing this allows investors to focus on stocks that are more likely to experience extreme price changes, and therefore secure higher returns. A step towards this goal is to focus on news spikes.
News Buzz Study
In our study “Abnormal Media Attention Impacts Stock Returns”, we test the assumption that more important events will get higher media coverage, attract the attention of investors, and result in greater price impact. For this purpose, we introduce a way of measuring abnormal news buzz volume on publicly traded companies (a.k.a. Event Buzz). We show that this measure of abnormal media attention and news buzz can assist in the formation of more targeted portfolios, resulting in an improvement in both Annualized Returns (see Figure 1) and Information Ratio (see Table 1 and Table 2).
As a convenient shortcut to a factor-neutral portfolio implementation, we evaluate the performance in specific returns, using the Barra US Total Market Equity Trading Model (USFAST).
Consistent with our initial hypothesis, we find that companies, with high Event Buzz are characterized by more extreme price changes which translates into higher volatilities than companies with low Event Buzz. Moreover, focusing on companies with extreme sentiment and high Event Buzz, provides greater average per-trade returns, with improvements of up to 2x for both large/mid-cap and small-cap companies.
Filtering News Buzz Provides Significant Improvements
In Table 1 and Table 2, we demonstrate that filtering on Event Buzz provides significant improvements in portfolio annualized returns with an increase from 9.6% to 15.5% and from 25.2% to 45.5% for large/mid-cap and small cap companies, respectively. Information Ratios increase from 1.74 to 2.04, across large/mid-cap companies; whilst for small-cap companies, the IR improves from 2.73 to 3.68.
Finally, using Monte Carlo techniques to produce resampled benchmark portfolios, provides a non-parametric tool to check the statistical significance of our findings. Looking at the IRs (or Annualized Returns) of the proposed strategies, we establish that overall results are found to be robust in both large/mid-cap and small-cap universes: portfolios focused on the highest quintile of Event Buzz yield performance that belong to the most extreme percentiles of the empirical distribution of IRs (or annualized returns) for the resampled portfolios, hence providing strong confidence in these strategies when it comes to signal construction.
To learn more about this study, news buzz, how to construct these indicators, or how strategies are formed and evaluated, request the full white paper titled “Abnormal Media Attention Impacts Stock Returns”.
Peter Hafez (Chief Data Scientist, RavenPack) is speaking at the upcoming RavenPack Research Symposium: “The Big Data & Machine Learning Revolution Comes to London” about applications of news analytics in financial markets. Other speakers include John "Morgan" Slade (CEO, CloudQuant), Andrej Rusakov (Co-founding Partner, Data Capital Management), Asger Lunde (Director/Prof of Economics, Copenhagen Economics/Aarhus University).