A Dockerized Jupyter notebook for back-testing in financial markets

Abstract

This is a docker container based on the data science notebook of Project Jupyter.

It includes additional components that are useful for financial market back-testing, as well as data acquisition through the APIs made available by Interactive Brokers.

Some of the components included:

• IB TWS
  • ib_insync
  • twsapi
• Backtesting
  • backtrader
• Plotting
  • bqplot
  • datashader
  • ggplot
  • holoviews
  • mpld3
  • plotly, with chart-studio and cufflinks
• Natural Language Processing
  • nltk
• Machine Learning
  • TensorFlow
• Misc
  • lxml, html5lib for pandas html_read
  • pandas-datareader

Components can added or removed by modifying the additional-requirements-conda.txt and additional-requirements-pip.txt files prior to building the image.

Due to licensing restrictions, this container is only made available as a Dockerfile to be built by the end user, and not as a ready-to-run pre-built container.

Basic usage

Begin by checking out the Dockerfile and building the container:

git clone https://github.com/wk/docker-trader-notebook.git
sudo docker build -t trader-notebook docker-trader-notebook

For runtime parameters, see Jupyter's docker-stacks.