Research trading

summary:

• Description
• Install
• Notebook
• Documentation

Description

This repository regroups sources about Medium article(How to make profits in cryptocurrency trading with machine learning).

Python3.5 environment for datasciences.

Install

Requirements:

• Docker
• Docker-compose

Run :

make

more informations about Makefile commands in the documentation section

Open your browser and visit the link localhost:8888.

The notebook used is src/Financial_presentation.ipynb.

Notebook

You have to know 3 tips for this notebook:

1. The dataset is not in the repository.

The size of this file is too big to be uploaded on Github.

You can create your own dataset with this 'data crawler'

2. The function get_data() is not implemented.

The object construct by this function must have this format:
  [
    {
      "name": "Your market name (ex. BTC-NEO)",
      "data": "Pandas dataframe of the market(with a chronological order)"
    },
  ]
example:

  [ {"name":"BTC-ETH", "data":<class 'pandas.core.frame.DataFrame'>} ]

3. Data format:

index	ask	base_volume	bid	date	high	id	id_bot	last	low	marketName	moy_prev_day	open_buy_orders	open_sell_orders	volume
int	int	float	int	str	int	int	int	int	int	str	int	int	int	float
0	72	0.61017	70	2018-04-04 04:30:30	72	1597	4242	72	68	BTC-2GIVE	69	113	979	869022.0

This dataset was collected with this software : Data crawler

Only features listed in the "labelisation_features_name" (settings dictionary) are used to your model.

References

Makefile commands available:

commands name	description
make	1. down each service
	2. build basic project
	3. run project
make build_up	1. build basic project
	2. run project
make build	build basic project.
make up	run project
make down	down project
make ps	list container
make logs	display all platform logs

Documentation:

• Crawler Data repository
• Jupyter
• Jupyter Lab
• Python3.5
• Docker
• Docker-compose
• Bittrex
• CryptoCompare
• scikit-learn