markowitz-api

FastAPI service that computes mean-variance-optimal portfolios using the Markowitz model. Point it at a list of tickers and a date range; it returns the efficient frontier and the max-Sharpe allocation.

Endpoint

POST /markowitz_model/

{
  "stocks": ["SPY", "TLT", "GLD", "DBC"],
  "start_date": "2020-01-01",
  "end_date": "2024-12-31"
}

Returns optimal weights, expected portfolio return, expected volatility, Sharpe ratio, and the efficient-frontier scatter.

How it works

• Downloads daily close prices via yfinance
• Computes log returns and annualized covariance matrix (252 trading days)
• Generates 10,000 random long-only portfolios to sketch the efficient frontier
• Solves for max-Sharpe allocation via SciPy SLSQP under sum-to-1 and non-negativity constraints

The math lives in app/routers/helpers/markowitz_helper.py.

Running locally

pip install -r app/requirements.txt
uvicorn app.main:app --reload
# → http://localhost:8000/docs for the OpenAPI UI

Stack

Python 3.12 · FastAPI · yfinance · NumPy · SciPy · Pandas · Plotly. Deployable on Render via render.yml.