Macroeconomic Indicators and RavenPack Sentiment

To capture the overall investor mood in the market, we construct a sentiment index based on the Event Sentiment Score (ESS) developed by RavenPack.

The index is constructed using a simple, intuitive, yet robust approach capturing the sentiment momentum from news on the US market over a three month period.

We find that the RavenPack Sentiment Index is not only closely related to financial markets but also to widely accepted macroeconomic indicators.

The RavenPack Sentiment Index Moves Closely with Financial Markets

• From January 2000 to September 2011, the contemporaneous correlation between the RavenPack Sentiment Index and the S&P 500 is 79%.
• The RavenPack Sentiment index is consistently highly correlated with the S&P 500 across different market trends. Especially, we find an average correlation of almost 90% during bear markets.

The RavenPack Sentiment Index is Closely Related to Macroeconomic Indicators

• The RavenPack Sentiment Index is consistently positively correlated with widely used macroeconomic indicators, both at a monthly and quarterly horizon.
• The quarterly contemporaneous correlation between changes in the RavenPack Sentiment Index and in real GDP is 56%. Including a sentiment lag with up to three quarters improves the correlation to almost 70%.

Introduction

In this report, we take a closer look at the RavenPack Sentiment Index and its relationship to a set of widely used macro-economic indicators. In order to measure the overall market sentiment, we construct a sentiment index based on the Event Sentiment Score (ESS) developed by RavenPack (for more information on the index methodology see “Introducing the RavenPack Sentiment Index”).

The algorithm producing the ESS scores depends on the detected event category and other information it teases out of the news story like financial figures, analyst ratings, and directional or emotional language. ESS can take a value between 0 and 100, and is determined by systematically matching stories typically categorized by financial experts as having short-term positive or negative stock price impact.

The algorithm produces an ESS score for more than 330 types of news events –including product recalls, earnings announcements, layoffs, M&A activity, to name a few (see Appendix A for more details). An event with ESS less than 50 is usually considered negative, while an event with ESS greater than 50 is considered positive.

We take a simple and straightforward approach when constructing the RavenPack Sentiment Index. The index value on a given day is measured as the difference between the count of positive news (ESS>50) and negative news (ESS<50). Note that we exclude news from the groups “Insider-Trading” and “Order-Imbalance” as we find an abnormally large presence of events belonging to these two groups in the US relative to other countries.

Generally, these types of events receive a neutral sentiment score by the ESS algorithm. To control for news flow seasonality caused by the earnings season, we take a 90 day simple moving average of the difference between the count of positive and negative news.

To examine whether the RavenPack Sentiment Index is able to capture market sentiment, we check the contemporaneous correlation between the index and the S&P 500 from January 2000 to September 2011. As shown in Fig 1, the contemporaneous correlation over the entire period is 79%. To evaluate whether the sentiment index is able to capture the market mood consistently across different market trends, we break down the sample into 4 periods: January 2000 to August 2002 (Bearish Market), September 2002 to July 2007 (Bullish Market), August 2007 to February 2009 (Bearish Market), and March 2009 to September 2011 (Bullish Market). As shown in Fig 1, the sentiment index and the S&P 500 are consistently positively correlated across all 4 periods. During the two bearish markets, the average correlation reaches almost 90%.