ETD

• agent based model
• complex systems
• liquidity
• market impact
• market microstructure
• order flow
• quantitative finance
• transient impact model

This thesis studies how financial markets react to firm-specific public information, distinguishing changes in the statistical state of order flow from changes in the mechanical market-impact mechanism. Using high-frequency data for twelve liquid US stocks in 2023–2024 and RavenPack news, daily response functions, order-flow cross-moments, and transient-impact propagators are estimated and analyzed with PCA and ticker-stratified statistical tests. News days are systematically associated with atypical response and order-flow curves, while no comparable systematic change is detected in the propagator. The order-flow effect is mainly linked to a larger fluctuation scale, with comparatively stable normalized temporal dependence. A microscopic framework based on heterogeneous liquidity takers and providers is then developed, showing how persistent order flow and transient impact can emerge from heterogeneous execution and relaxation timescales.