/forex_py

[WIP] predicting forex.

Primary LanguagePythonMIT LicenseMIT

Status

https://travis-ci.org/0h-n0/forex_py.svg?branch=master Documentation Status Code Climate

CONCEPT

Frxpy is just a test tool to know whether DeepLearning predict a forex market. So, Frxpy doesn't provide the best prediction for a market. Of course, Frxpy may lead the completely wrong predictions. If you use this for your real trading and lose a lot of money, you have to take all the responsibility on yourself.

CONTENTS

  • predict FOREX with deep learning.
  • get the best trading strategy with reinforcement learning.

Requirements

  • pytorch >= 0.4.0
  • pandas >= 0.23.1

TODO

  • add tensorboard for visulaization.
  • add Deeplearning.

forex_py

predicting forex.

Download

git clone https://github.com/0h-n0/forex_py.git
# or
wget https://github.com/0h-n0/forex_py/archive/master.zip
unzip master.zip

Install(build&install)

build Envrionment

$ wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

Install frxpy

cd forex_py
python3 setup.py install

Docs

cd forex_py
cd docs
make html
firefox build/html/index.html

Where do I get data?

Histdata provides histrical forex data. The data is available for free.

suported data type

  • Tick Data
  • Generic .CSV File

gap treatment

There are gaps in data. So, I convert the gaps to the averages between the value before the gaps and the value after the gaps.

Examples

Command-line interface

frxpy --input --buy

API of python3

I assume that forex_py is used in python-scripts. So, Python3-APIs of forex_py is more poweful and frexible than command-line interface.

import frxpy

Supported Data Base & Serializer

In progress...

  • sqlite3
  • leveldb
  • mysql
  • hdf5
  • postgresql
  • tinydb
  • protobuf