pyqstrat 0.1.1

fast / extensible library for backtesting quantitative strategies

Introduction

The pyqstrat package is designed for backtesting quantitative strategies. It was originally built for my own use as a quant trader / researcher, after I could not find a python based framework that was fast, extensible and transparent enough for use in my work.

This framework is designed for capable programmers who are comfortable with numpy and reasonably advanced Python techniques.

The goals are:

• Speed - Performance sensitive components are written at the numpy level which can lead to performance improvement of several orders of magnitude over python code. Where possible, we parrallelize work so you can take advantage of all the cores available on your machine.

• Transparency - If you are going to commit money to a strategy, you want to know exactly what assumptions you are making. The code is written and documented so these are as clear as possible.

• Extensibility - It would be impossible to think of all requirements for backtesting strategies that traders could come up with. In addition, traders will want to measure different metrics depending on the strategy being traded.

Using this framework, you can:

• Construct a portfolio containing multiple strategies that run concurrently

• Construct arbitrary indicators, trading signals and trading rules and add them to a strategy

• Add multiple symbols representing real or “virtual” instruments to a strategy

• Reuse existing market simulation or build your own to simulate how and when orders are filled

• Measure well known return metrics such as sharpe, calmar and also add your own metrics.

• Resample market data bars into lower frequencies

• Simulate futures rolling using “roll trades”

• Plot trades, market data, indicators and add custom subplots to give you insight into your strategy’s operation.

• Optimize your strategy’s parameters using all the CPU cores on your machine.

** NOTE: This is beta software and the API will change **

Installation

I would strongly recommend installing anaconda and creating an anaconda environment. See installation instructions at https://docs.anaconda.com/anaconda/install/

pyqstrat relies on numpy, scipy, matplotlib and pandas which in turn use Fortran and C code that needs to be compiled. Depending on your environment, using pip may not result in a smooth install of these packages

pip3 install -U pyqstrat

Requirements:

• Python version 3.6 or higher;

Documentation

The best way to get started is to go through this Jupyter notebook: Building Strategies

Jupyter Notebooks

API docs

Discussion

The pyqstrat user group is the group used for pyqstrat discussions.

Acknowledgements

Before building this, I looked at the following. Although I ended up not using them, I got some great ideas for them.

R quantstrat library

Python backtrader project

Some of the ideas I use in this framework come from the following books

Trading Systems: A New Approach to System Development and Portfolio Optimisation - Tomasini, Emilio and Jaekle, Urban

Machine Trading - Chan, Ernie

Algorithmic Trading: Winning Strategies and Their Rationale - Chan, Ernie

Disclaimer

The software is provided on the conditions of the simplified BSD license.

v0.1.0, 2018-07-31 – Initial release.