Nicolas Liebaert
I spent two internships on structured products desks, at Natixis CIB in equity derivatives sales and at Kepler Cheuvreux in cross-asset structuring, pricing autocallables, reverse convertibles, shark notes and credit-linked notes. Afterwards I wrote my own implementations of those payoffs, to understand them from the inside rather than from a pricing screen.
This site runs that code. Move a slider and the note is repriced by the same Python that sits in the repositories, executing in your browser through WebAssembly. Nothing here is a precomputed table dressed up as a model.
01
Structured products
A Phoenix note pays a headline coupon of 7% and looks like income. Price it properly and the median investor collects 3.5 points, while 8.6% of paths end with capital down by more than half.
02
Vanilla options
Closed-form greeks, two binomial trees whose convergence you can watch, and an implied volatility solver that survives the wings where vega collapses and Newton-Raphson gives up.
03
Systematic strategy
Trend following on a 3x leveraged ETF over 16 years. It halves the drawdown and it loses to buy and hold on every risk-adjusted measure. Both halves of that sentence are the point.
Each page downloads a Python runtime compiled to WebAssembly, together
with NumPy and SciPy, and then fetches the project's own
.py files and imports them. That download is not small, so
it only starts once a model is on screen, and the progress is reported
as it happens.
Interactive figures use twenty thousand Monte Carlo paths so that the result follows the slider. Every panel also has a high precision setting that runs two hundred thousand paths, which reproduces the numbers published in the repository READMEs. Where a figure is uncertain, the confidence interval is shown next to it rather than hidden behind a rounded number.
The models here are flat Black-Scholes. That is enough to get the structure of a payoff right and to reason about its risk, and it is not what a desk quotes on. Each page says where its own model stops being trustworthy.