Research Paper#Financial Modeling, Machine Learning, Systemic Risk🔬 ResearchAnalyzed: Jan 3, 2026 16:38
CRBMs for Systemic Risk Regime Detection
Published:Dec 26, 2025 01:23
•1 min read
•ArXiv
Analysis
This paper explores the application of Conditional Restricted Boltzmann Machines (CRBMs) for analyzing financial time series and detecting systemic risk regimes. It extends the traditional use of RBMs by incorporating autoregressive conditioning and Persistent Contrastive Divergence (PCD) to model temporal dependencies. The study compares different CRBM architectures and finds that free energy serves as a robust metric for regime stability, offering an interpretable tool for monitoring systemic risk.
Key Takeaways
- •CRBMs are used to model high-dimensional financial time series.
- •Autoregressive conditioning and PCD are used to capture temporal dependencies.
- •Free energy is a useful metric for regime stability.
- •The model can distinguish between magnitude shocks and market regimes.
- •CRBMs offer an interpretable tool for monitoring systemic risk.
Reference
“The model's free energy serves as a robust, regime stability metric.”