TradeTester is a tool to test various trading strategies both on historic data and on live data streams. Tool uses backtrader to provide the comprehensive testing environment and tvDatafeed to easily (and without charge) retrieve both historic and live ticker data from TradingView platform.
TradeTester is provides controller through which user can create and manage testruns. Testruns are essentially separate threads executing backtrader (cerebro) with user specified strategy. Strategy is executed on live ticker data which is retrieved from TradingView. Live data is retrieved using modified tvDatafeed code base which, instead of retrieving historic data, monitors specified assets for new data and once it is released retrieves it and feeds into testruns (Cerebros running stratgies).
The data IS NOT LIVE in the sense that it has a delay. TradeTester is not intended to perform automated trading, but is visioned to perform automated stratgy testing and (in hoepfully in the future) ranking different stragies in real-time based on some metrics.
TradeTester is still in development as of now, but following features have been implmented (but not fully tested):
- Basic SQL database (on local machine) to save all the orders and testruns
- Retrieving live ticker data from TradingView using tvDatafeed
- Running separate backtrader strategies in thread concurrently (testruns)
- Simple controller module to start and stop running new strategies in a single TradeTester application (don't need to run multiple backtrader scripts to test multiple stratgies)
Features that are planned to add in the future:
- GUI through which user can add strategies that they want to test and also see live statistics and ticker/order data about running stratgies. Vision is to have some metrics which will quickly show which stratgies is most successful and there will be a ranking based on that
- Seamless transition from backtesting (historic data) to live testing (on data live streams). Currently the backtesting part must be done separtly and TradeTester only support live testing.
Currently the main.py acts as a testbench and an example code. In the future (once GUI is added) this will change and it will act as a boot up module.
The controller module is meant to act as a central controller through which user can add and manage testruns. As TradeTester includes an SQL database for saving orders then controller must be initialized by providing a path and a Sqlite3 database name. TradeTester will create a database in that location with that name. Each instance of controller will have its own database. In the future it is visioned that there will be only one instance of controller running 24/7 and user will use it to add and removed new testruns (strategies).
contr=controller(r"C:\Users\User\Documents\tradeTester\development_materials\testDB.db")
After creating a controller instance user can use that to add new testruns via method
start_testrun() and providing unique name for testrun, strategy class template,
asset, exchange/market, timeframe interval and starting account size.
tuid=contr.start_testrun("myTest1", MyStrategy, "ETHUSDT", "KUCOIN", Interval.in_1_minute, 10000)
This method will return an integer (TUID - Testrun Unique ID) which can be used later to uniquely reference that testrun.
TradeTester is designed with an idea of having GUI in the future. Because of that reason the logic is that testruns (strategies) will send all orders to SQLManager instance which takes care of saving them into database. The GUI will display data (orders, ticker data etc.) for one particular testrun at any given time. This is set by setting the streamer in SQLManager (via controller instance). The streamer setting means that SQLManager will propagate only that testrun related data to GUI. Propagating data is done via Queues.
contr.select_testrun(tuid1)