Multi-Market Systemic Risk Spillover Dataset with Network and Machine Learning Features

This repository provides a comprehensive dataset for analyzing systemic risk spillovers across global equity markets using a network-based and machine learning framework. The dataset comprises 4,316 daily observations spanning from June 5, 2001 to November 14, 2024, covering twelve major stock market indices across North America, Europe, Asia, and Latin America.

The dataset is constructed to support the empirical framework developed in the study “Predicting Systemic Risk Spillovers: Evidence from a Multi-Market Network Analysis using Machine Learning” . It integrates market-level financial indicators with network-derived spillover measures based on the Diebold–Yilmaz methodology, enabling both structural analysis of financial contagion and predictive modeling of systemic risk.

Each observation includes a rich set of variables capturing market dynamics, including daily returns, conditional volatility, and directional spillover metrics such as spillover transmitted, spillover received, and net spillover positions. These variables are further expanded through lag structures to capture short-term temporal dependencies, resulting in a high-dimensional feature space suitable for machine learning applications.

The dataset is specifically designed for:
- Systemic risk measurement using variance decomposition and network topology
- Cross-market spillover and contagion analysis
- Identification of risk transmitters and receivers in global financial networks
- Forecasting systemic spillover intensity using supervised learning models such as Random Forest, XGBoost, and LightGBM
- Studying regime-dependent dynamics across crisis and non-crisis periods

By combining multi-market financial data with network-based indicators, this dataset facilitates advanced research in financial stability, risk propagation, and early warning systems. It is particularly suitable for researchers in financial econometrics, network analysis, and applied machine learning in finance.