This repository implements a full pipeline to recover risk-neutral one-year inflation distributions and moments from option prices.
The objective is identification and stability, not forecasting.
We compare multiple methods and analyze when higher moments are weakly identified due to limited strike coverage and boundary effects.
- Prices → Density → Moments
- Density recovered via second derivative of call prices: [ f(k) = \frac{1}{B(0,1)} \frac{\partial^2 C(k)}{\partial k^2} ]
- Sensitive to numerical noise (finite differentiation)
- Operational version of BL
- Local polynomial smoothing before differentiation
- Improves numerical stability
- Prices → Moments directly
- No density recovery
- Highly sensitive to truncation and tail coverage
- Prices → Density → Moments
- Discrete probability grid
- Regularized convex optimization
- Used as a robustness benchmark
Information window: pi in [-1%, 5%]
option_implied_inflation_probability_density_functions/
├─ README.md
├─ requirements.txt
├─ LICENSE
├─ .gitignore
├─ options_implied_inflation_pdf.py
├─ results/
│ ├─ fig/
│ └─ updated_results.csv
├─ report/
│ └─ main.tex
└─ data/
└─ README_DATA.md
Raw data files are sourced from Bloomberg and are therefore not included in this repository.
To reproduce the pipeline, place your own licensed data files in data/.
python -m venv .venv
source .venv/bin/activate # macOS/Linux
# .venv\Scripts\activate # Windowspip install -r requirements.txtpython src\options_implied_inflation_pdf.pyNam Khanh Nguyen, Rodrigue Mieuzet, Melany Gipsy Moreno, Khrystyna Kateryna Valenia, Katarzyna Pastuszka
Master 2 Finance Technology Data (FTD) Université Paris 1 Panthéon-Sorbonne