End-to-End Python implementation of Regime-Weighted Conformal (RWC) prediction for sequential VaR control in nonstationary financial markets (Schmitt, 2026). Combines kernel-based regime similarity with exponential time decay to calibrate distribution-free risk bounds. CRSP data validation, GBDT quantile forecasting, and rigorous backtesting.
python machine-learning reproducible-research market-data kernel-methods quantitative-finance uncertainty-quantification value-at-risk gradient-boosting risk-management time-series-analysis backtesting quantile-regression financial-econometrics portfolio-management conformal-prediction sequential-prediction tail-risk icml-2026 crsp-data
-
Updated
Feb 8, 2026 - Jupyter Notebook