Research repository for the mathematics of financial markets, centered on options. The thesis:
An option chain encodes the market's entire risk-neutral probability distribution for a stock's future price. Options carry more information about a stock than its own price history does.
Goal: build, in C++, a multi-model predictive-distribution engine for a stock's near-future price — an expected price with honest uncertainty bands, not a single guessed number, by combining the option-implied distribution with calibrated structural models and showing where they agree and disagree.
This is personal, educational quantitative-finance research. Not investment advice. Mathematics-first; chart indicators (RSI, MA crossovers, etc.) are explicitly out of scope.
Theory + roadmap phase. No C++ written yet. The mathematical path and the
C++ build order are specified in docs/curriculum.md.
| File | Purpose |
|---|---|
CLAUDE.md |
Authoritative project reference, conventions, division of labor, OptionWatch data source |
AGENTS.md |
Agent execution protocol and scope |
docs/curriculum.md |
8-part options-centric math path + 11-step C++ roadmap |
| File | Topic |
|---|---|
materials/perfect-portfolio.md |
Markowitz mean-variance / efficient frontier (video summary) |
materials/black-scholes-history.md |
Bachelier → Brownian motion → Black-Scholes-Merton (video summary) |
materials/quant-next-playlist.md |
Rigorous breakdown of the Quant Next "Option Pricing and Risk Management" playlist (10 topics, stochastic calculus → Breeden-Litzenberger) |
materials/american-options-binomial-tree.md |
Pricing American options on a CRR tree (early-exercise backward induction) |
Option chains are fetched in C++ from the public OptionWatch API — the
same source used by the operator's separate options_analysis Python library.
Endpoints and rules are documented in CLAUDE.md → "Data source". No API key;
no synthetic data ever substituted for a failed fetch.