Smart Stock Charts

Smart Stock Charts is an investment blog where I also automatically publish short trading signals for various stocks, indices, and cryptocurrencies.  These signals are generated by a neural network model that looks at a 20 day chart of the investment and predicts if the investment will break out of its previous 5, 10, 15, and 20 day prices ranges in the next 5, 10, 15, and 20 days respectively.  Although past performance is not indicative of future performance, it has generally predicted breakout behavior with about a 70% accuracy historically.

Skills Used

 

  • Python
  • Machine Learning
  • Pytorch
  • WordPress
  • HTML
  • Financial Data APIs

 

Resources Used

 

  • UC Berkeley’s Open Computing Facility, which hosts the Pytorch model
  • Hostgator, which hosts the blog and the model outputs
  • Spyder IDE for writing the code
  • Github for tracking code changes (repository is private)

 

Completion Date

 

Ongoing Project