/simple-trading

Codebase to show how to do ML for trading, data analysis, algo trading etc from ground up

Primary LanguageDockerfileMIT LicenseMIT

simple-trading

Codebase to show how to do ML for trading, data analysis, algo trading etc from ground up

To setup

  • Install Docker
  • If you have GPU and want to access the same from the container then follow this steps
  • Build the container. Like so - docker build -t my-dl-img .
  • Run the container. Like so - ./bin/run_container.sh <app|notebook> (This script assumes you have GPU, if not use this command docker run -p 127.0.0.1:8888:8888 --rm -it -v $PWD:/tmp my-dl-img <app|notebook>)
  • There are a few utility scripts to be run from the host machine. Like the following. They will help to clean up sometimes.
    • ./bin/clean_containers.sh
    • ./bin/scan_8888.sh

Warning: The total image size is pretty big, about 7GB because of the base image (pytorch)

What is included -

  • Pytorch (Modelling)
  • XGBoost (Modelling)
  • Numpy (Numerical computations)
  • scipy (Scientific computations)
  • plotly (Plotting lib)
  • pandas (DataFrame)
  • nsepy (Specific package for fetching NSE related data)
  • pandas-datareader (Remote data access for pandas)

What is in the roadmap

  • Create tutorials for introduction and basic visualization
  • Basic intro to Algorithmic trading and Quant (??)
  • Start building an app to load data and play around with it
  • Train basic models
  • Create an English like language to to backtest based on price-action or similar things
  • Add Deep RL capabilities (Specifically FinRL)
  • More??

Some addition