deep-hedging

Implementation of the vanilla Deep Hedging engine

A curated list of resources dedicated to Deep Hedging

Open-source quantitative finance models in Python: option pricing and Greeks, volatility surfaces, deep hedging, portfolio construction and Monte Carlo. Every model runs on free data, with the full write-up on davidariasfinance.com

Coding assignments of the "Machine Learning in Finance & Insurance" course at ETH Zürich (Fall 2024).

Deep Hedging neural network for dynamically pricing and managing derivatives risk under realistic market frictions

Interpretable volatility-surface prototype hedger that matches delta-vega on tail risk while staying auditable (ICAIF 2026)

A package to learn optimal hedges by a deep feed forward neural network, to minimise the terminal error

Code and reproducibility package for “What Does Deep Hedging Actually Learn? Delta Corrections, Regime Fragility, and Symbolic Distillation.”

Deep-learning option pricing and hedging: a neural surrogate for Asian options benchmarked against Monte Carlo, a rough Bergomi model for 0DTE, a CVaR deep hedging policy, live calibration, and an interactive dashboard.

LightGBM-based hedging strategy under Merton's jump diffusion with custom loss and delta approximation

Deep Hedging under market frictions.

Option-pricing and hedging models built from scratch and verified on real market data, with honest caveats. Black-Scholes to Heston to deep hedging.

Reinforcement Learning derivative hedging (Deep Hedging) vs Black-Scholes baseline.

Neural hedging policies (GRU, deep-set, no-trade-band) minimising CVaR, entropic and spectral risk under transaction costs, on fused CUDA kernels at billions of paths per second, with a deep BSDE pricer.

Deep‑Hedging in PyTorch (MCPG): europäische & amerikanische Optionen mit RSQP‑Risiko, GJR‑GARCH‑Pfade, IV‑Features und Chebyshev‑Pricing inkl. Baselines.

Deep-hedging pipeline: trains a from-scratch PPO agent (JAX) to hedge options under discrete rebalancing + transaction costs, then compares it to Black-Scholes delta hedges across GBM/Heston regimes and two cost levels (5 bps vs 20 bps) on 3,000 paired paths. At low costs BS wins (as theory predicts). At higher costs the same PPO trades 30–45% less

Numpy-only replication of Deep Hedging (Buehler et al. 2019): a neural hedging policy trained via hand-written backprop-through-time to minimize entropic risk under transaction costs, benchmarked against Black-Scholes delta hedging.

A Master Project implementing a Deep Reinforcement Learning (DDPG) agent for transaction-cost-aware option hedging. Features Behavioral Cloning for a "warm start" and is backtested on real-world SPY ETF data.

Neural ODEs, CDEs, SDEs and deep hedging — building continuous-depth models from the maths up

Deep Bellman Hedging in Hedging Gym: actor-critic value iteration under monetary utilities, entropic objective, Lean proofs — thesis fork of 0xC000005/hedging-gym