/backtrader

Python Backtesting library for trading strategies

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

backtrader

PyPi Version License Travis-ci Build Status Python versions

Yahoo API Note:

Don't use it. Yahoo removed the original API and the alternative methods are prone to failure due to the glitches introduced by Yahoo (column swapping and others)

Tickets

If it's NOT an issue (i.e.: bug), don't post it as an issue. It will be automatically closed.

For feedback/questions/... use the Community

Here a snippet of a Simple Moving Average CrossOver. It can be done in several different ways. Use the docs (and examples) Luke!

from datetime import datetime
import backtrader as bt

class SmaCross(bt.SignalStrategy):
    def __init__(self):
        sma1, sma2 = bt.ind.SMA(period=10), bt.ind.SMA(period=30)
        crossover = bt.ind.CrossOver(sma1, sma2)
        self.signal_add(bt.SIGNAL_LONG, crossover)

cerebro = bt.Cerebro()
cerebro.addstrategy(SmaCross)

# **NOTE**: Read the note about the Yahoo API above. This sample is kept for
# historical reasons. Use any other data feed.

data0 = bt.feeds.YahooFinanceData(dataname='YHOO', fromdate=datetime(2011, 1, 1),
                                  todate=datetime(2012, 12, 31))
cerebro.adddata(data0)

cerebro.run()
cerebro.plot()

Including a full featured chart. Give it a try! This is included in the samples as sigsmacross/sigsmacross2.py. Along it is sigsmacross.py which can be parametrized from the command line.

Features:

Live Trading and backtesting platform written in Python.

  • Live Data Feed and Trading with
    • Interactive Brokers (needs IbPy and benefits greatly from an installed pytz)
    • Visual Chart (needs a fork of comtypes until a pull request is integrated in the release and benefits from pytz)
    • Oanda (needs oandapy) (REST API Only - v20 did not support streaming when implemented)
  • Data feeds from csv/files, online sources or from pandas and blaze
  • Filters for datas, like breaking a daily bar into chunks to simulate intraday or working with Renko bricks
  • Multiple data feeds and multiple strategies supported
  • Multiple timeframes at once
  • Integrated Resampling and Replaying
  • Step by Step backtesting or at once (except in the evaluation of the Strategy)
  • Integrated battery of indicators
  • TA-Lib indicator support (needs python ta-lib / check the docs)
  • Easy development of custom indicators
  • Analyzers (for example: TimeReturn, Sharpe Ratio, SQN) and pyfolio integration
  • Flexible definition of commission schemes
  • Integrated broker simulation with Market, Close, Limit, Stop, StopLimit, StopTrail, StopTrailLimit*and *OCO orders, bracket order, slippage, volume filling strategies and continuous cash adjustmet for future-like instruments
  • Sizers for automated staking
  • Cheat-on-Close and Cheat-on-Open modes
  • Schedulers
  • Trading Calendars
  • Plotting (requires matplotlib)

Documentation

The blog:

Read the full documentation at:

List of built-in Indicators (122)

Python 2/3 Support

  • Python 2.7
  • Python 3.2 / 3.3/ 3.4 / 3.5 / 3.6
  • It also works with pypy and pypy3 (no plotting - matplotlib is not supported under pypy)

Compatibility is tested during development with 2.7 and 3.5

The other versions are tested automatically with Travis.

Installation

backtrader is self-contained with no external dependencies (except if you want to plot)

From pypi:

  • pip install backtrader

  • pip install backtrader[plotting]

    If matplotlib is not installed and you wish to do some plotting

Note

The minimum matplotlib version is 1.4.1

An example for IB Data Feeds/Trading:

  • IbPy doesn't seem to be in PyPi. Do either:

    pip install git+https://github.com/blampe/IbPy.git

    or (if git is not available in your system):

    pip install https://github.com/blampe/IbPy/archive/master.zip

For other functionalities like: Visual Chart, Oanda, TA-Lib, check the dependencies in the documentation.

From source:

  • Place the backtrader directory found in the sources inside your project

Version numbering

X.Y.Z.I

  • X: Major version number. Should stay stable unless something big is changed like an overhaul to use numpy
  • Y: Minor version number. To be changed upon adding a complete new feature or (god forbids) an incompatible API change.
  • Z: Revision version number. To be changed for documentation updates, small changes, small bug fixes
  • I: Number of Indicators already built into the platform

Alternatives

If after seeing the docs and some samples (see the blog also) you feel this is not your cup of tea, you can always have a look at similar Python platforms: