Simple event based backtester for Forex markets
$ cd $HOME/projects
$ git clone https://github.com/Oxylo/fxengine
$ workon env
$ cd fxengine
$ pip install -r requirements.txt
$ export PYTHONPATH=$HOME/projects # export Python path
$ python trading/trading.py
fxengine is a project for quantitative forex trading at OANDA.
It is based on code from quantstart. and QuantFxEngine by H.Kopp.
CAUTION: ALGORITHMIC TRADING IS A WAY TO LOSE LOTS OF MONEY VERY FAST. IF YOU DO NOT KNOW; WHAT YOU ARE DOING, DO NOT TRADE ON LIVE ACCOUNTS. Practice accounts are okay.
You can start this stuff with python trading/trading.py
Also you probably want to have a look at settings.py. If you want to trade with OANDA with this programm, you need to have your account id in the environment variable OANDA_API_ACCOUNT_ID and your access token in OANDA_API_ACCESS_TOKEN.
Backtesting is also possible if you have a suitable csv-file.
If you want to adjust the logging, look at logging.conf.
The main function is in trade/trading.py. We open two threads, one who streams prices from a file or from a broker in streaming/streaming.py. This adds TickEvents to the queue which the two threads use to communicate.
The second thread handles the Events in the queue. TickEvents get handled by the strategy module, which can then trigger SignalEvents. The portfolio also looks at TickEvents to track the price of open positions.
The portfolio class looks at the SignalEvents and if it fits in the risk assessment, it throws OrderEvents.
OrderEvents get executed by the execution module. That means that we buy or sell. The execution class then throws FillEvents, which include the price for which the Order was filled. This can be different from the price at which the decision to invest was made.
FillEvents get handled by the portfolio to keep track of the current positions.
Tests are done with nosetests and are bundled per object in a separate folder.
Please have a look at the docs section if you need some background on the trading process.