Institution: Sungkyunkwan University
Period: Spring 2025
Project: Persistent Homology for Financial Crisis Detection
Overview
This project applied topological data analysis (TDA) to the study of financial crises, focusing on the use of persistent homology as a framework for detecting early warning signals in market dynamics.
Contributions
- Implemented a persistent homology-based framework for early warning signals of financial crises in R
- Developed an independent implementation of the analysis code
- Extended the framework, originally applied to US, Singapore, and Malaysia, to include the Korean stock market
- Conducted comparative evaluation with traditional statistical approaches
- Found that the Mean Power Spectrum (MPS) of the residual time series achieved the best performance across all four stock markets US, Singapore, Malaysia, and Korea