hayatoy / ml-forex-prediction

Predicting Forex Future Price with Machine Learning

Predicting Financial Time Series Data with Machine Learning

This is an example that predicts future prices from past price movements. Here we implement it with EUR/USD rate as an example, and you can also predict stock prices by changing symbol.

Backtest example for EUR/USD

Using daily close prices from 2008 to 2016, first 95% for training and last 5% for testing. Green and red vertical lines represent winning trade and losing trade respectively.

Installation

To run this demo, you need following environment and libraries.

• Python 2.7 (not tested on 3.x)
• Jupyter Notebook
• Scikit-learn
• numpy
• pandas
• matplotlib
• seaborn

Note: You may need extra libraries to install above.

License

MIT License, Copyright (c) 2017

ml-forex-prediction's Issues

Unable to import data from yahoo finance anymore

You inported the header file below
from matplotlib.finance import quotes_historical_yahoo_ochl
is no longer applicable
anyway you would like to suggest to replace this?
preferbly method that can import data from online sources

Forex Prediction

How can i predict future price?

It is predicted and calculated to test data, how can i predict brand new day? I mean next day?

A future function

Hi,return_range = df['return'].max() - df['return'].min() is a future function

Are you passing to predict futures values?

Hello,
Quit an old repo, but it seems to me that when you are using Predict function you are passing values in the futur. The 60 features sample array you are building is using return from futur. In real life you can not pass thoses values to the Predict Method.

Did I miss something?

Regards

Accuracy of predictions