Stock Prediction Web App

Wanted to learn the MERN stack end to end and ML deployment at the same time. A finance dashboard gave both: API ingestion, persistent storage, model inference, and a charts-heavy UI.

Build a web app that ingests daily NASDAQ closes for a chosen ticker, runs a custom regression model server-side, and visualises the predicted next-7-day path against the historical close.

Express API as the backend, MongoDB to cache daily closes, Python service behind it for the model. React on the front with a chart library for the visualisation.

• Daily ingest job pulled closes from a free API and cached them in MongoDB.
• Python regression service exposed an HTTP endpoint the Node server called.
• React frontend let users pick a ticker, see the historical line, and overlay the prediction.
• Deployed both services to a single VPS behind nginx.

Working deployment. Used as my MERN learning vehicle through 2023.

The model was naive — a moving-average regression that flattered itself on sideways markets. I would replace it with a proper baseline (ARIMA or a small LSTM) before claiming any predictive value.