From digital leaders to market volatility in cryptocurrency markets: Volatility modeling based on social network and sentiment analysis

Investment decisions in traditional financial markets have long been shaped by financial statements, balance sheet analyses, and expert reports. However, the emergence of cryptocurrency markets has fundamentally transformed this paradigm. For cryptocurrencies that operate independently of any central authority, trade continuously, and lack any underlying asset, traditional information sources have proven inadequate, and social media has begun to assume a decisive role in financial decision-making processes. The billions of dollars in value lost in the global cryptocurrency market following a single tweet by Elon Musk represents one of the most striking indicators of this transformation. This study aims to examine the impact of sentiments expressed on social media platforms on volatility in cryptocurrency markets within a multi-layered analytical framework. The research was conducted on a unique dataset comprising millions of tweets shared about Ethereum on the Twitter platform over a six-year period. The study adopts a three-stage methodological approach that integrates social network analysis, sentiment analysis, and econometric modeling. In the first stage of the research, social network analysis was conducted to identify the network leaders who direct information flow within the Ethereum ecosystem. The primary rationale for selecting the in-degree centrality metric is that this measure answers the question of "whose opinion is being sought" and enables the identification of actors who have genuine influence on financial decisions. The findings from the social network analysis revealed that the leadership structure within the Ethereum ecosystem has undergone a fundamental transformation over the years. While technical founders and project developers occupied the center of the network in the early periods, cryptocurrency exchanges came to prominence in subsequent years. In recent periods, the rise of influencer impact has been notably observed, and the NFT ecosystem has transformed the leadership structure. This evolution demonstrates that the crypto community has shifted from a tightly-knit structure where few individuals engaged in intensive interaction to a loosely connected structure where many individuals engage in superficial interaction. In the second stage of the research, the interactions of identified network leaders were examined through sentiment analysis. Using the NRC sentiment analysis library, the emotional tone of each tweet was determined, and positive, negative, and total tweet counts were calculated on a daily basis. The distinctive aspect of this approach is that it analyzes the interactions of leaders who have genuine market impact rather than randomly selected tweets. Thus, a meaningful and high-impact sentiment measurement was obtained from within the big data. In the third stage of the research, sentiment analysis variables were integrated into the GARCH model as exogenous variables. While the classical GARCH model explains volatility solely based on the series' own history, the model developed in this study enables the measurement of the direct impact of social media-driven information flow on volatility. The econometric analysis results revealed that all sentiment variables have a statistically significant effect on volatility. One of the most critical findings of the research is that the impact of negative sentiment on volatility is much stronger compared to positive sentiment. This result directly aligns with the loss aversion hypothesis and the concept of negativity bias in behavioral finance literature. It has been proven that investors react more harshly to negative information and that this reaction significantly increases market volatility. Another important finding is that tweet volume increases volatility regardless of content. This result demonstrates that the intensity of information flow alone constitutes a source of uncertainty and that investor attention directly affects market dynamics. News Impact Curve analyses visually confirmed the asymmetric effect of shocks on volatility. While positive and negative information create similar effects during small shocks, it has been demonstrated that negative information causes much more severe market fluctuations during large shocks. This thesis offers significant theoretical and methodological contributions to the literature. It has developed an original approach that integrates social network analysis, sentiment analysis, and GARCH modeling within a single framework. While most studies in the existing literature focus on superficial correlations between social media and financial markets, this study systematically answers the questions of who influences, how they influence, and how much they influence. Furthermore, by challenging the Efficient Market Hypothesis, it has shown that information is not equally distributed to the market and spreads asymmetrically through leader actors. By adapting Agenda-Setting Theory to financial markets, it has demonstrated that social media is not merely an information channel but a direct determinant of financial decisions. The research findings offer important practical implications for different stakeholders. For investors, the development of hybrid strategies that consider social media sentiment alongside traditional analysis methods has become inevitable. The strong effect of negative sentiment should be considered as a critical parameter in risk management and portfolio optimization. For regulatory bodies, monitoring the risks of social media-driven manipulation and information pollution, as well as expanding market surveillance mechanisms, is of importance. For platform providers, the development of transparency mechanisms for financial content is necessary. In conclusion, this thesis has strongly demonstrated that cryptocurrency markets are shaped not only by price and trading volume dynamics but also by social media sentiment. In the digital age, information no longer originates solely from traditional financial institutions but from social media posts that instantly reach millions, directly influencing markets. Social media has become the digital laboratory of global finance, and cryptocurrency markets have emerged as a new field where behavioral finance theories can be observed in their purest form. This study lays the methodological foundation for a new interdisciplinary research field that could be termed social finance and demonstrates that the new dynamics of finance are shaped by the social interactions of the digital age.