Benjamin Kraft

Regime detection and volatility modeling for European energy and carbon markets. Trading desks and risk teams typically estimate volatility persistence on full historical windows, which treats stable and crisis regimes as one process. The result: VaR numbers, hedge ratios, and scenario sensitivities that misstate risk during regime transitions in EUA, gas, and Nordic power. I detect structural breaks, estimate regime-conditional persistence, and quantify the misstatement.

Regime detection and structural break analysis. Where conditional volatility or correlation has shifted, regime probabilities under different specifications, implications for model assumptions. 5 to 10 days.

Volatility model build and validation. Implementing or validating GARCH-family models (DCC, GJR, MS-GARCH, HAR-RV-X) for risk, hedging, or scenario analysis. 10 to 15 days.

Cross-asset conditional correlation. Time-varying dependence across EUA, TTF, Nordic power, and adjacent assets, including correlation breakdown under stress. 5 to 15 days.

Multivariate GARCH (DCC, BEKK), GJR-GARCH-X, Markov-switching GARCH, HAR-RV-X, Bai-Perron structural break detection, QML estimation, Diebold-Mariano testing, Model Confidence Set, Monte Carlo simulation, rolling-window backtesting.

R (rmgarch, rugarch, strucchange), Python (statsmodels, arch, NumPy, Pandas, SciPy), SQL, Git.

Nord Pool day-ahead and system price, Nordic bidding zones (SE3, NO2, FI), ICE Futures Europe EUA and TTF, clean spark spread economics. ENTSO-E (API), Refinitiv Eikon, Bloomberg. EU ETS Phase IV mechanics, Nordic market coupling, REPowerEU.