This app uses the historical data of different crypto currencies to train deep learning models for each currency and predict the currency direction price.
The app is mainly based on the Trader class which inheritates from the backtrader.client.Client adding more functionalities to make much easier the training, prediction, buy and sell of the currencies.
To start, we need to create the docker image
git clone https://github.com/JonathanElejalde/playing_with_btc.git
cd playing_with_btc
docker build -f Dockerfile -t ml_strategy .Before we run the container, make sure to add your own keys for the binance API. You can do this by adding keys.py as follows:
api_key = "your api key"
secret_key = "your secret key"or you can just add them directly in the app.py file when calling:
trader = Trader('your api key', 'your secret key')After this, we can run the app using the container
docker run -v %cd%:/strategy -ti ml_strategy /bin/bash -c "cd strategy && python -W ignore app.py"Change the %cd% with the absolute path where the playing_with_btc folder is.
It will run in test mode, if you want to make real trading you will need to change TEST in the config.py file to True and use the trader.create_order function instead of trader.create_test_order. - For this, you just need to comment and uncomment the lines in the app.py file-
This project is licensed under the MIT License
I've been learning trading and AI in parallel, so I wanted to create something that will involve both concepts just as a way to practice and not thinking too much in making money.
The app.py is an example and not a reliable way of making money. Thus, I recommend not to use it as a real trading strategy. However, you can use the concepts in this file to build your own reliable strategy.