LSTM Stock Price Prediction

Inspiration

Our project seeks to address the Stock Price Prediction challenged posed by the TAMU Datathon team. We were motivated to apply Long Short-Term Memory neural networks to time series analysis of stock indicators.

What it does

Our project receives data about the stock indicators of a particular stock, and predicts the opening price of the stock the next day, off of which it bases automated stock trading decisions.

How I built it

The project was built using Google Colaboratory notebooks, as well as Python and the Keras machine learning library.

Challenges I ran into

Getting everyone's colab and python development environments running smoothly was a major hurdle at the start, and afterwards we ran into issues with choosing the right architecture and design for our model.

Accomplishments that I'm proud of

We are proud that we created a LSTM model that is able to trade effectively and accumulate several hundred dollars over a month.

What I learned

We learned a lot about reinforcement learning, the first approach that we explored for stock price analysis. Afterwards we also learned a lot about applying LSTM architecture to time series forcasting problems.

What's next for LSTM Stock Price Prediction

Creating a more variable model that can model multiple stocks from different companies, to expand the reach of the model.

Built With

• google-colaboratory
• jupyter-notebooks
• keras
• python