Stock Market Predictor

Ship was denied, I was informed watchlist didn't work, made API route in flask realized I had forgot to make that, fixed logic in js and voila it works now

Fixed again by reducing epoches that reduced training time significantly, news section properly added doesn't give parsing error when you click the tab

Added much more functions like a watchlist, macd and rsi and a news section

finally figured out how to run flask on a server and actually deploy it gunicorn app:app -b localhost:42313 --timeout 120 &
I had to add a timeout to my command

tried running it on nest server, caddyfile configuration I give up

found out that hackclub has free servers, will deploy properly shortly

Have a problem, I dont want to pay for any service that makes me deploy an app as I'm broke, cant find anywhere to deploy free python application, python anywhere doesn't work, I deployed front end, back end instructions can be found in github. Designed and implemented a responsive HTML/CSS interface for the Stock Price Predictor app. The UI includes input fields for user-defined stock parameters and renders prediction results using Chart.js.. Styled with a dark theme, custom fonts, and hover effects for enhanced usability. Layout adapts for desktop and mobile screens. Connected to Flask backend for full integration.

Converted the LSTM model to .h5 and saved the scaler as .pkl. Used Flask for the backend and built a virtual environment due to library install issues. The app takes a ticker and timeframe, fetches data, scales it, runs predictions, and returns results.

Trained a two-layer LSTM neural network using closing price data for AAPL from 2010–2024. The model used a 60-day lookback window and was trained for 20 epochs with normalized inputs. Libraries included yFinance, NumPy, Pandas, scikit-learn, TensorFlow, and Matplotlib. Loss dropped consistently during training, and predictions showed good alignment with actual prices.