A personal research repository exploring quantitative finance, model risk, and machine learning for derivatives pricing and hedging.
🔍 Note: This project is unrelated to QuantLib.org. It is a private learning and experimentation space by AlmostAnna.
This repo contains:
- Classical models: Black-Scholes, Heston, local volatility
- Exotic derivatives: Autocallables, Asian options, structured notes
- Hedging analysis: Error decomposition, stop-loss replication, discrete trading
- Machine learning: Buehler-style deep hedging with transaction costs
- Stress testing: Sensitivity to volatility misspecification, rebalancing frequency
All code is organized to support reuse, clarity, and diagnostics—not just one-off experiments.
quantlab/
├── notebooks/
│ ├── models/ # Stochastic volatility, Dupire, etc.
│ ├── model_risk/ # Hedging errors, replication failure, Greeks
│ └── ml/ # Deep hedging, training diagnostics
├── src/ # Reusable quant primitives (installable as 'quantlab')
| ├── quantlab/
│ ├── hedging/ # Greeks, naive strategies
| ├── instruments/
│ ├── market_data/
| ├── ml/ # Models, metrics
| ├── models/
| ├── pricing/
| ├── sim/ # MC simulations
│ └── utils/
├── ml/ # ML-specific training and evaluation
├── tests/ # Tests
├── pyproject.toml # For editable install
└── environment.yml
- Clone and install:
git clone https://github.com/AlmostAnna/quantlab.git cd quantlab pip install -e .[dev]
- Clarity over cleverness: Code should speak for itself.
- Model risk matters: Every assumption is surfaced and tested.
- ML as a tool, not a black box: Diagnostics, baselines, and stress tests are first-class citizens.
© 2025 AlmostAnna — For learning, reflection, and professional growth.