Dependencies :
-Python3.5 -Tensorflow -yahoo_finance -matplotlib -numpy -django
- python3.5 -m virtualenv YOUR_VIRTUAL_ENV_FOLDER
- cd to the folder 'YOUR_VIRTUAL_ENV_FOLDER'
- cd bin
- source activate (on windows)
Studied 'reinforcement learning' chapter in the book below, utilised the source code uploaded on the site and creating a new project. https://www.manning.com/books/machine-learning-with-tensorflow
This is a link to the original source code : https://github.com/BinRoot/TensorFlow-Book/blob/master/ch08_rl/rl.py
The original logic was to buy/sell a share rather than a number of shares. I've changed the algorithm and it now buys shares as many as possible from budget and sell every share in it's portfolio as to spend less money for stamp duty and other fees.