How News Events Impact Market Sentiment

Previously, I presented an idea on how to construct a Market-level Sentiment Index based on counts of positive and negative sentiment news about companies belonging to a target equity index. I found that considering a company relevance metric is an important element in constructing sentiment-based strategies as using relevant news improved the out-of-sample return correlation by a factor of 3.

In my latest study, I take relevance filtering a step further and include only news that is "contextually" relevant to the companies in the S&P500. That is, where a company is involved in an event such as a lawsuit or acquisition and the role it plays is also detected (e.g. the company is the defendant, plaintiff, acquirer, acquire, etc.).

To construct a market-level news sentiment index I use RavenPack's Event Sentiment Score, which indicates how company-related events are typically rated by financial experts as having positive or negative share price impact. To capture news events specifically related to S&P500 companies, I use the RavenPack Company Relevance Score set to 100. This metric provides a way to capture stories that are 100% relevant to the S&P500 constituents and not mere mentions in the text. The numerical score indicates "how" relevant the story is to the company and assigns higher values based on the context of the news using semantic analysis. Finally, I use an Event Novelty Score, which represents how "new" or novel a news story is over a given time window. The first story revealing the company-related event is considered to be the most novel and receives a score of 100.

Figure 1 below depicts the US Market-Level Sentiment Index I constructed vs. the S&P500 cumulative index log-returns covering the period March 2005 through December 2009.

Figure 1: Market-Level Sentiment Index (red-line) - primary-axis; vs. the S&P500 cumulative log-returns (black-line) - secondary-axis. The sentiment index has been constructed based on the average Event Sentiment Score of all S&P500 companies over a 90 day trailing window covering the period January 2005 through December 2009. In addition, events have been filtered only to include the most novel stories.


Under the assumption that market returns are likely to move in the same direction as market sentiment, I base my trading decisions on the sentiment index delta. Focusing on the index delta will capture the sentiment of the most recent period (i.e. one month), but also include the sentiment change from the previous period which may have similar characteristics (i.e. as captured by the "same" month in the earnings season cycle).

Figure 2 below shows the cumulative return of a trading strategy based on the US Market Sentiment Index with and without applying an Event Novelty Score filter, benchmarked against a one-month price-momentum strategy.

Figure 2: Cumulative strategy returns covering the out-of-sample period May 2005 through December 2009 for the Market-Level Sentiment Index (blue-line) with an Event Novelty Score filter, the Market-Level Sentiment Index (green-line) without an Event Novelty Score filter, and the one-month price-momentum strategy based on the S&P500 (red-line). The sentiment indexes have been constructed based on a 90 day trailing window, and a trading decision was made based on the monthly index delta. A positive delta resulted in a long position and a negative delta in short position.


Not only do both sentiment-based strategies outperform one-month price-momentum, but filtering based on event novelty seems to add significant value in predicting the future price direction of the S&P500. The Event Novelty filtered strategy would have obtained an overall Information Ratio of 1.75 over the period with the values 1.02 and 2.47 pre- and post the market high of the test period in October 2007, respectively. Overall, the event novelty filtered strategy would have realized an annualized return of 26.5% with a Hit Ratio of almost 70%. Interestingly, both sentiment indexes not only deliver significantly better returns than price momentum, but do so with lower volatilities. Also, a significant improvement can be observed in the Hit Ratios.