Behavioral economics in actions: How biases and heuristics shape stock market investment decisions

This bachelor’s thesis investigates how digital herding and the spread of investment narratives through social media disrupt the functioning of classical econometric models and lead to persistent market inefficiencies during the 2020–2022 period. By analyzing assets such as GameStop, Tesla, Bitcoin, and Dogecoin, the study demonstrates that standard ARMA models fail to capture price dynamics when markets are driven not by fundamentals but by coordinated retail sentiment. Empirical findings reveal prolonged shock persistence, structurally correlated residuals, and recurring volatility clusters that align with identifiable social media events. These phenomena represent systematic violations of the assumptions required for ARMA model stability and predictive accuracy. Cases such as GameStop’s extreme surge during the short squeeze and Dogecoin’s volatility following Elon Musk’s tweets illustrate that emotionally synchronized behavior and narrative-driven speculation often dominate investor decisions. The thesis thus provides quantitative evidence of the structural limits of market efficiency and shows that traditional financial models cannot accurately describe markets influenced by digital coordination, behavioral biases, and viral narratives. The research highlights the need for new valuation frameworks that integrate behavioral finance and the dynamics of modern digitally mediated markets.