The Effects of Twitter Sentiment on Stock Price Returns (1506.02431v2)

Abstract: Social media are increasingly reflecting and influencing behavior of other complex systems. In this paper we investigate the relations between a well-know micro-blogging platform Twitter and financial markets. In particular, we consider, in a period of 15 months, the Twitter volume and sentiment about the 30 stock companies that form the Dow Jones Industrial Average (DJIA) index. We find a relatively low Pearson correlation and Granger causality between the corresponding time series over the entire time period. However, we find a significant dependence between the Twitter sentiment and abnormal returns during the peaks of Twitter volume. This is valid not only for the expected Twitter volume peaks (e.g., quarterly announcements), but also for peaks corresponding to less obvious events. We formalize the procedure by adapting the well-known "event study" from economics and finance to the analysis of Twitter data. The procedure allows to automatically identify events as Twitter volume peaks, to compute the prevailing sentiment (positive or negative) expressed in tweets at these peaks, and finally to apply the "event study" methodology to relate them to stock returns. We show that sentiment polarity of Twitter peaks implies the direction of cumulative abnormal returns. The amount of cumulative abnormal returns is relatively low (about 1-2%), but the dependence is statistically significant for several days after the events.

The Effects of Twitter Sentiment on Stock Price Returns

The intersection of social media and financial markets presents an intriguing area of research, particularly in understanding how sentiment expressed on platforms like Twitter correlates with stock price movements. The paper in question, "The Effects of Twitter Sentiment on Stock Price Returns," investigates the relationship between Twitter activity and financial markets, specifically focusing on the 30 companies comprising the Dow Jones Industrial Average (DJIA).

Summary of the Research

The paper begins by acknowledging the growing influence of social media on various complex systems, including financial markets. It posits that Twitter sentiment could potentially predict market behavior, particularly during periods of heightened tweet activity. Over a 15-month paper period, the authors focus on tweet volumes and sentiment about the companies listed in the DJIA, examining the correlation with stock returns.

Methodology

To uncover this relationship, the authors employ several analytical techniques:

1. Sentiment Classification: Tweets were classified into positive, neutral, and negative sentiments using a supervised machine learning approach. Over 100,000 tweets were manually annotated by financial experts, and a Support Vector Machine (SVM) model was trained to automatically assign sentiment to the broader dataset.
2. Correlation and Granger Causality: Initial analysis involved computing Pearson correlation coefficients between sentiment polarity and stock returns. Additionally, Granger causality tests were applied to detect potential predictive patterns.
3. Event Study Approach: The paper notably adapts event paper techniques from econometrics to identify Twitter activity peaks and assess their impact on stock returns. This approach is aimed at discerning whether certain tweets can indicate abnormal returns, using peaks in Twitter activity to denote potential market-moving events.

Key Findings

The paper finds a relatively low overall correlation between Twitter sentiment and stock returns across the entire paper period. However, during peaks in Twitter activity, particularly those that coincide with earnings announcements, sentiment exhibited a statistically significant impact on cumulative abnormal returns (CAR) of stocks. Noteworthy results include:

• Cumulative Abnormal Returns: Positive sentiment during identified peaks led to an increase in CAR, while negative sentiment corresponded with decreased CAR. The impact was sustained for several days post-event, with abnormalities in returns remaining statistically significant for up to ten days following positive sentiment peaks.
• Non-Earnings Announcements (Non-EA) Events: Even when excluding known earnings announcements, sentiment peaks still showed significant influence on stock returns, suggesting that Twitter can capture market-relevant information from other noteworthy but less apparent events.

Implications and Future Research

The findings underscore the complexity of the relationship between social media sentiment and financial markets, highlighting that the interaction is more pronounced during specific events rather than over an extended period. The research suggests a path forward in utilizing real-time social media sentiment as a supplementary tool for market analysis, potentially aiding in "now-casting" rather than forecasting market behavior.

For future research, the paper paves the way for more granular analyses, such as leveraging high-frequency trading data and examining intra-day movements correlated with Twitter sentiment changes. It also opens avenues for developing automated systems that can monitor social media activity for early detection of potential market shifts.

In conclusion, the research provides compelling evidence that Twitter sentiment does indeed influence stock returns, particularly around heightened periods of online activity. This insight could be invaluable for traders and policy-makers in understanding and reacting to market dynamics influenced by social media platforms.