Predict NVDA close using Machine Learning

description

1. The project prepares historical market data.
2. Uses pytorch to train a model on that data.
3. Makes a prediction at a user defined number of days in the future as to what the closing price of NVDA will be on that day.

use

1. clone the repo and move into the directory

git clone https://github.com/austinp0573/nvda_close_predictor.git

cd nvda_close_predictor

2. if you're using uv

uv venv

uv sync

# (optionally)
source .venv/bin/activate

uv run src/main.py

3. using python3

python3 -m venv .venv

source .venv/bin/activate

pip install .

python3 ./src/main.py

recreation

• the 4 .csv files that are used to begin the project came from stooq.com
• one could use the most recent available data to recreate the project by acquiring NVDA.US (nvda_us.csv), ^SPX (sp500.csv), VI.C (sp500_vix.csv), and ^NDX (nasdaq_100.csv)
• and replacing the .csv files that are there with more recent versions
  • using the exact name that I used is crucial
  • when collecting data for ^SPX, VI.C, ^NDX only go back to January 22nd 1999 (the day of NVDA's IPO)
• (my data includes up to October 9th 2025.)

 

466f724a616e6574