This project is a collection of Python codes that aim to replicate the Matlab ones for Dr. Ernest Chan's Algorithmic Trading book.
The file names for scripts reflect AT Matlab script names, for example
the code for Ratio.m is in Ratio.py.
All feedback, comments, pull requests are welcome.
Python can be installed through the Anaconda distribution. This is the easiest way to do it, and on Windows, probably the only way.
Then run conda install or pip install for
numpy
scipy
statsmodels
pandas
arch
There are three big data files that are hard to share through Github, they can be downloaded through the links below:
The code will assume these files are under [HOME]/Dropbox/Public/data folder.
I prefer to work with CSV files, the Pandas library makes it a breeze
to access them plus I can view the contents of CSV files easily,
manipulate them with Unix based tools if necessary. For converting a
mat file into csv this is the method I followed: 1) Find the Matlab
script for AT that reads and prepares the data, i.e. TU_mom.m 2)
find the point where the data is all ready, i.e.
clear;
load('inputDataOHLCDaily_20120511', 'syms', 'tday', 'cl');
..
idx=strmatch('TU', syms, 'exact');
tday=tday(:, idx);
cl=cl(:, idx);
..
- After the last line above, the
tday,clvariables are prepared, with the same dimensions, with the right data, etc. We can create a data matrix out of these and write them to disk. Insert the following in the script,
A = [tday cl];
save('/tmp/out','A');
exit;
then run it. 4) Now from a seperate Python script,
from scipy import io as spio
import numpy as np
import pandas as pd
y = spio.loadmat('/tmp/out.mat')['A']
df = pd.DataFrame(y, columns=['Date','Close'])
df.to_csv('/tmp/out.csv',index=None)
This gives you a CSV file.
The [Algorithmic Trading] (http://amzn.to/1F4RTtT) Book
[My Blog] (http://sayilarvekuramlar.blogspot.com)
Dr. Chan announcing this project on his blog, see the bottom of the post.
The code is licensed under GPL v3. See COPYING for details.