#ChinaMeltDown: an Agent-based Simulation model

This thesis investigates the question on whether Twitter data can provide interesting and thought-provoking insights on financial markets and if so under which circumstance. It begins with a research of an appropriate dataset of ¿tweets¿ concerning the actual Chinese financial crisis. Consequently, after a theoretical overview, a sentiment analysis has been made in order to obtain precious information. Then an Agent-Based simulation model of an artificial stock market is constructed in NetLogo. The combination of real data collected and elaborated previously with manufactured financial dynamics represents the crucial part of the analysis; thus, various experiments are undertaken within it. For instance, changing how agents reacts to different ¿tweets¿ in terms of quality and quantity. Finally, the results are discussed and some possible extensions put forward.