Impact of Social Media Sentiments on Stock Price and Trading Volume : Evidence from Google Trends, Reddit Data and Tweets
Pysyvä osoite
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
This research investigates the influence of social media sentiment on stock price and stock trading volume, recognising the increasing role of online information dissemination in stock markets. The advent of social media like Twitter (X), Reddit and online financial news and opinion sites has led to an increase in the influence of real-time information on investor sentiment. This research is based on the theories of behavioral finance and the Efficient Market Hypothesis (EMH), which explore the impact of social media sentiment on stock price and stock trading volume.
The study uses a quantitative research approach, with sentiment data from various social media and news websites, and stock price and trading volume data of a list of listed companies over a certain time period. Sentiment analysis is applied to categorize the sentiment polarity of the text data, and statistical linear and nonlinear machine learning models such as Artificial Neural Networks (ANNs) and random forests are used to examine the relationships and predictability of the data.
The findings show that the relationship between social media sentiment and stock prices is statistically weak but positive using linear methods. But the predictive value of sentiment is greatly enhanced by the nonlinear model, revealing the high explanatory power of machine learning models. This study also concludes that sentiment polarity has a greater impact than the number of mentions, i.e., the message's tone is more important than its volume. On the other hand, this study found no correlation between social media sentiment and trading volume and volatility.
The research suggests that social media sentiment is a complementary variable in financial prediction, rather than a standalone variable. It suggests that investors and financial institutions should use social media sentiment in conjunction with traditional financial data and apply sophisticated analytical methods for better predictions. Future studies should increase sample sizes, apply more sophisticated sentiment analysis models and study inter-market interactions to improve the validity and generalizability of the results.