Skip to content
Nas
02 / 05Solo · full-stack

FinQuantPro

Year

2026

Tech

Next.js · FastAPI · Python · Claude API

Links

Quant finance tools in the browser, with an AI Copilot that runs them for you.

  • 8

    quant tools

  • 100%

    grounded numbers in Copilot evals

  • $0.008

    cost per Copilot question

RETURNRISKMax Sharpe
Illustrative

What it is

It started as a Streamlit app and I rebuilt it as a full web app: a Next.js frontend with a Python API doing the maths. Ask the Copilot a question in plain English and it runs the right tool, then links you to the full result.

What's inside

A Black-Scholes pricer with Greeks and a 3D price surface. A Strategy Builder, plus a Strategy Picker that turns your market view into a strategy and tests it with Monte Carlo against market prices. An implied vol solver. Protect & Earn for protective puts, collars and covered calls, with a backtest. My Portfolio with what-if analysis and an efficient-frontier optimiser. A Stock Health Check covering default risk (Merton model), volatility regime, tail risk and sentiment. Market news with sentiment scoring.

The Copilot

Built on Claude Haiku with tool calling. The model never does maths itself: every number comes from the same code the pages use, and it only sees a short summary and a link. An eval suite checks it picks the right tool, passes the right inputs and only quotes numbers the tools returned.

How it's built

Next.js, TypeScript and ECharts on Vercel. FastAPI with NumPy, SciPy and statsmodels on Render, cached in Upstash Redis and rate-limited. GitHub Actions runs linting, tests and builds on every push.

Next project

Student Homeys