This session explores novel modelling approaches to speculative asset pricing, distinguishing between classic bubbles and more subtle "balloon-like" behaviour—characterized by rapid rises and gradual declines. Using high-frequency Bitcoin data, Andrew Harvey presents a series of score-driven and quasi score-driven models that integrate volatility, non-normality, and dynamic tail behaviour. The findings challenge traditional bubble narratives and offer practical tools for forecasting and risk modelling in speculative markets.

Overview objectives:

• Understand how score-driven models and time-varying tail indices improve the detection of bubble vs. balloon market dynamics.
• Learn how these models were applied to Bitcoin to identify and simulate locally explosive behaviour.
• Explore the implications for volatility modelling and density forecasting in financial markets with extreme price fluctuations.