ABSTRACT
This paper proposes a novel hybrid framework combining CAViaR with gradient boosting machines to forecast tail risk spillover, offering superior performance for real-time systemic risk monitoring.
PAPER · PDF
Loading PDF…
Key findings
The hybrid CAViaR-GBM framework outperforms traditional benchmarks in forecasting.
Violation rates are closer to nominal levels with reduced quantile loss.
Diebold-Yilmaz spillover analysis reveals significant tail risk transmission across sectors.
Limitations & open questions
The model's performance in markets with different levels of liquidity and volatility is not assessed.
The study does not address the potential impact of macroeconomic factors on spillover dynamics.