Currently backtrader
has a built-in integration for Interactive Brokers (IB) Trader Workstation API, but it only works for older versions of the API.
This project re-worked the orginal integration and created a pluggable version allowing backtrader
to use the latest IB API.
To start trading follow the steps below:
- Open an account with IB
- Download the IB Python TWS API
- Download the IB client (TWS or IB Gateway) Trader Workstation Platform
- Test for connectivity – Check code sample below
Create a local python virtual environment:
python3 -m venv ./env
Activate the environment, all the following packages will be install in the local ./env directory
source ./env/bin/activate
Install backtrader locally (see Installing Backtrader for more options)
pip install backtrader
Download Interactive Brokers TWS API (from Trader Workstation API)
Select the latest copy, currently it is (twsapi_macunix.1016.01.zip) for Linux
unzip twsapi_macunix.1016.01.zip
The file will unzip to the directoty IBJts
cd IBJts/source/pythonclient
Run the setup.py to install the TWS API.
python setup.py install
Download a Atreyu Backtrader API, released version:
wget https://github.com/atreyuxtrading/atreyu-backtrader-api/archive/refs/tags/v1.0.zip
Unzip file, and install the Atreyu Backtrader API.
unzip v1.0.zip
cd atreyu-backtrader-api-1.0 ; python setup.py install
import backtrader as bt
from atreyu_backtrader_api import IBData
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="GOOG", # Data name
dataname='GOOG', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART',# Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
what='BID_ASK', # Get data fields (see note below)
rtbar=True, # Request Realtime bars
_debug=True # Set to True to print out debug messagess from IB TWS API
)
cerebro.adddata(data)
cerebro.run()
Note that this is created as a stratgey and will print all the bars that it receives
import backtrader as bt
class TestPrinter(bt.Strategy):
def log(self, txt, dt=None):
''' Logging function for this strategy'''
dt = dt or self.datas[0].datetime.datetime(0)
print(f'{dt}, {txt}')
def __init__(self):
self.open = self.datas[0].open
self.high = self.datas[0].high
self.low = self.datas[0].low
self.close = self.datas[0].close
self.volume = self.datas[0].volume
self.openinterest = self.datas[0].openinterest
def next(self):
self.log(f'Open:{self.open[0]:.2f}, \
High:{self.high[0]:.2f}, \
Low:{self.low[0]:.2f}, \
Close:{self.close[0]:.2f}, \
Volume:{self.volume[0]:.2f}, \
OpenInterest:{self.volume[0]:.2f}' )
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_printer import TestPrinter
import datetime as dt
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="GOOG", # Data name
dataname='GOOG', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART',# Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
historical=True,
what='BID_ASK', # Update this parameter to select data type
)
cerebro.adddata(data)
# Add the printer as a strategy
cerebro.addstrategy(TestPrinter)
cerebro.run()
2021-08-09 23:59:59.999986, Open:137.24, High:144.44, Low:136.25, Close:137.55, Volume:-1.00
2021-08-10 23:59:59.999986, Open:138.02, High:139.84, Low:125.00, Close:138.26, Volume:-1.00
2021-08-11 23:59:59.999986, Open:137.54, High:138.95, Low:130.66, Close:137.89, Volume:-1.00
2021-08-12 23:59:59.999986, Open:137.82, High:139.07, Low:130.00, Close:138.12, Volume:-1.00
2021-08-13 23:59:59.999986, Open:138.23, High:139.09, Low:137.78, Close:138.52, Volume:-1.00
2021-08-16 23:59:59.999986, Open:138.04, High:139.90, Low:125.00, Close:138.34, Volume:-1.00
....
2022-08-05 23:59:59.999986, Open:118.06, High:128.00, Low:111.06, Close:118.19, Volume:-1.00
2022-08-07 20:00:00, Open:118.93, High:120.88, Low:113.00, Close:119.02, Volume:-1.00
Historical data is returned in the form of candlesticks, and accessed using the “what=” parameter when requesting data. (see Interactive Brokers Data Types)
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_printer import TestPrinter
import datetime as dt
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="GOOG", # Data name
dataname='GOOG', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART',# Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
fromdate = dt.datetime(2016, 1, 1),
todate = dt.datetime(2018, 1, 1),
historical=True,
what='TRADES',
)
cerebro.adddata(data)
# Add the printer as a strategy
cerebro.addstrategy(TestPrinter)
cerebro.run()
2016-01-05 00:00:00, Open:37.38, High:37.38, Low:36.56, Close:37.10, Volume:460493.60
2016-01-06 00:00:00, Open:37.00, High:37.60, Low:36.93, Close:37.15, Volume:272008.00
2016-01-07 00:00:00, Open:36.87, High:37.36, Low:36.25, Close:37.30, Volume:276044.20
2016-01-08 00:00:00, Open:36.17, High:36.92, Low:35.95, Close:36.50, Volume:425276.00
...
2017-12-27 00:00:00, Open:52.86, High:53.00, Low:52.51, Close:52.70, Volume:70263.00
2017-12-28 00:00:00, Open:52.90, High:52.92, Low:52.40, Close:52.46, Volume:151108.40
2017-12-29 00:00:00, Open:52.66, High:52.74, Low:52.24, Close:52.36, Volume:105796.60
2017-12-30 00:00:00, Open:52.42, High:52.55, Low:52.13, Close:52.24, Volume:75590.60
The data retrieved from IB is presented in the strategy as the variable self.datas[0].
The latest close price is available at index 0, and progressively earlier prices are stored using a negative index. (See diagram below)
import backtrader as bt
# Create a Stratey
class TestStrategy(bt.Strategy):
def log(self, txt, ts=None):
''' Logging function for this strategy'''
ts = ts or self.datas[0].datetime.datetime(0)
print(f'{ts}, {txt}')
def __init__(self):
self.close = self.datas[0].close
def next(self):
# Current close
self.log(f'Close:{self.close[0]:.2f}' )
if self.close[0] < self.close[-1]:
# current close less than previous close, think about buying
if self.close[-1] < self.close[-2]:
# previous close less than previous close, so buy
self.log('BUY CREATE, %.2f' % self.close[0])
self.buy()
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_strategy import TestStrategy
import datetime as dt
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="GOOG", # Data name
dataname='GOOG', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART',# Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
fromdate = dt.datetime(2016, 1, 1),
todate = dt.datetime(2018, 1, 1),
historical=True,
what='MIDPOINT',
)
cerebro.adddata(data)
# Add the test strategy
cerebro.addstrategy(TestStrategy)
# Set our desired cash start
cerebro.broker.setcash(100000.0)
cerebro.run()
print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue())
Datasources can be given logical datanames (i.e. based on the "what" parameter), the logical name can then be accessed using the _name variable. In the strategy below the logical name is stored in the self.name0 and self.name1 variables, and can be used to identify the buy for each symbol.
import backtrader as bt
# Create a Stratey
class TestStrategy(bt.Strategy):
def log(self, txt, ts=None):
''' Logging function for this strategy'''
ts = ts or self.datas[0].datetime.datetime(0)
print(f'{ts}, {txt}')
def __init__(self):
self.close0 = self.datas[0].close
self.name0 = self.datas[0]._name
self.close1 = self.datas[1].close
self.name1 = self.datas[1]._name
def next(self):
# Current close dataset0
self.log(f'{self.name0} Close:{self.close0[0]:.2f}' )
if self.close0[0] < self.close0[-1]:
# current close less than previous close, think about buying
if self.close0[-1] < self.close0[-2]:
# previous close less than previous close, so buy
self.log(f"BUY {self.name0} @ {self.close0[0]:.2f}")
self.buy()
# Current close dataset1
self.log(f'{self.name1} Close:{self.close1[0]:.2f}' )
if self.close1[0] < self.close1[-1]:
# current close less than previous close, think about buying
if self.close1[-1] < self.close1[-2]:
# previous close less than previous close, so buy
self.log(f"BUY {self.name1} @ {self.close1[0]:.2f}")
self.buy()
See the name parameter being used to tag each datasource in the example below:
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_strategy import TestStrategy
import datetime as dt
cerebro = bt.Cerebro()
goog_data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="GOOG_TRADES", # Data name
dataname='GOOG', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
fromdate = dt.datetime(2016, 1, 1),
todate = dt.datetime(2018, 1, 1),
historical=True,
what='TRADES',
)
cerebro.adddata(goog_data)
apple_data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="AAPL_MIDPOINT",# Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
fromdate = dt.datetime(2016, 1, 1),
todate = dt.datetime(2018, 1, 1),
historical=True,
what='MIDPOINT',
)
cerebro.adddata(apple_data)
# Add the test strategy
cerebro.addstrategy(TestStrategy)
# Set our desired cash start
cerebro.broker.setcash(100000.0)
cerebro.run()
print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue())
2016-01-05 00:00:00, AAPL_MIDPOINT Close:26.43
2016-01-05 00:00:00, BUY AAPL_MIDPOINT @ 26.43
2016-01-05 00:00:00, GOOG_TRADES Close:37.10
2016-01-05 00:00:00, BUY GOOG_TRADES @ 37.10
...
2017-12-29 00:00:00, AAPL_MIDPOINT Close:42.71
2017-12-30 00:00:00, GOOG_TRADES Close:52.24
2017-12-30 00:00:00, BUY GOOG_TRADES @ 52.24
2017-12-30 00:00:00, AAPL_MIDPOINT Close:42.27
Final Portfolio Value: 102168.92
Real-time bars represent a price performance for a specific period. These periods could be as long as a day or as short as a second, depending on the purpose for which the bar is to be used. Daily bars are usually the most popular for analysis whereas shorter duration bars can be used for trading.
In the case of IB the TWS API can be used to fetch 5-second duration bar. The example below creates an active subscription that will return a single bar in real time every five seconds that has the OHLC values over that period. Additionally we are switching off the backfill of data from initial start to reconnect in case of connection disruption.
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_printer import TestPrinter
import datetime as dt
from datetime import datetime, date, time
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
backfill_start=False,
backfill=False,
what='TRADES', # TRADES or MIDPOINT
rtbar=True
)
cerebro.adddata(data)
# Add the test strategy
cerebro.addstrategy(TestPrinter)
cerebro.run()
2022-08-11 15:22:20, Symbol: AAPL Open:169.29, High:169.29, Low:169.28, Close:169.29, Volume:101.13
2022-08-11 15:22:25, Symbol: AAPL Open:169.29, High:169.29, Low:169.25, Close:169.26, Volume:79.5
2022-08-11 15:22:30, Symbol: AAPL Open:169.27, High:169.30, Low:169.23, Close:169.23, Volume:57.5
2022-08-11 15:22:35, Symbol: AAPL Open:169.23, High:169.30, Low:169.22, Close:169.27, Volume:89.72
Using the TWS API, real time market data can also be requested for trading and analysis. This data is not tick-by-tick but consists of aggregated snapshots taken at intra-second intervals which differ depending on the type of instrument:
We select non-bar data by setting rtbar=False, note that the data will still be presented in the OHLCV format for use in the strategy.
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_printer import TestPrinter
import datetime as dt
from datetime import datetime, date, time
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
backfill_start=False,
backfill=False,
rtbar=False
)
cerebro.adddata(data)
# Add the test strategy
cerebro.addstrategy(TestPrinter)
cerebro.run()
2022-08-11 16:36:11.410065, Symbol: AAPL Open:169.70, High:169.70, Low:169.70, Close:169.70, Volume:2200.0
2022-08-11 16:36:11.410105, Symbol: AAPL Open:169.70, High:169.70, Low:169.70, Close:169.70, Volume:100.0
2022-08-11 16:36:11.410156, Symbol: AAPL Open:169.70, High:169.70, Low:169.70, Close:169.70, Volume:100.0
2022-08-11 16:36:11.410196, Symbol: AAPL Open:169.70, High:169.70, Low:169.70, Close:169.70, Volume:253852.0
2022-08-11 16:36:11.411061, Symbol: AAPL Open:169.69, High:169.69, Low:169.69, Close:169.69, Volume:253852.0
2022-08-11 16:36:11.411081, Symbol: AAPL Open:169.69, High:169.69, Low:169.69, Close:169.69, Volume:1900.0
2022-08-11 16:36:11.411141, Symbol: AAPL Open:169.71, High:169.71, Low:169.71, Close:169.71, Volume:1900.0
2022-08-11 16:36:11.411161, Symbol: AAPL Open:169.71, High:169.71, Low:169.71, Close:169.71, Volume:2900.0
In addition to the aggregated tick snapshots, IB also has true tick-by-tick data, but it comes with some limitations. Tick-by-tick data corresponds to the data shown in the TWS Time & Sales. The maximum number of simultaneous tick-by-tick subscriptions allowed for a user is determined by the limitations below.
Limitations - Additional symbol request can be purchased through a quote booster pack, each quote booster pack provides a 100 market data lines. There is a limit of 10 quote boosters packs per account and rest of the market data lines are allocated using equity and commissions.
The “what=” must be set to the following BID_ASK (BidAsk), TRADES (Last), TRADES_ALL (AllLast), MIDPOINT (MidPoint), the default is TRADES if “what” is not set. TRADES_ALL has additional trade types such as combos, derivatives, and average price trades that are not included in TRADES. Note in the example below “timeframe=bt.TimeFrame.Ticks” to select the tick-by-tick IB function.
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_printer import TestPrinter
import datetime as dt
from datetime import datetime, date, time
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
timeframe=bt.TimeFrame.Ticks,
what='BID_ASK', # TRADES, TRADES_ALL, MID_POINT
rtbar=False
)
cerebro.adddata(data)
# Add the test strategy
cerebro.addstrategy(TestPrinter)
cerebro.run()
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:260.0
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:100.0
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:100.0
...
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:100.0
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:100.0
2022-08-11 17:29:22, Symbol: AAPL Open:nan, High:nan, Low:nan, Close:169.41, Volume:100.0
In this section we look at the steps in developing a simple strategy and back testing it with historical data from IB. Then modifying it to Paper Trader in the TWS frontend.
An Example of a Simple Buy/Sell Strategy - Not that in this strategy we generate Buy and Sell market orders related to a simple moving average calculation. This strategy has been adapted from (https://www.backtrader.com/docu/quickstart/quickstart) with minor updates.
import backtrader as bt
# Create a Stratey
class TestStrategy(bt.Strategy):
params = (
('ma_period', 15),
)
def log(self, txt, ts=None):
''' Logging function for this strategy'''
ts = ts or self.datas[0].datetime.datetime(0)
print(f'{ts}, {txt}')
def __init__(self):
# Keep a reference to the "close" line in the data[0] dataseries
self.dataclose = self.datas[0].close
# To keep track of pending orders and buy price/commission
self.order = None
self.buyprice = None
self.buycomm = None
# Add a MovingAverageSimple indicator
self.sma = bt.indicators.SimpleMovingAverage(self.datas[0], period=self.params.ma_period)
def notify_order(self, order):
if order.status in [order.Submitted, order.Accepted]:
# Buy/Sell order submitted/accepted to/by broker - Nothing to do
return
# Check if an order has been completed
# Attention: broker could reject order if not enough cash
if order.status in [order.Completed]:
if order.isbuy():
self.log(
f'BUY EXECUTED, Price: {order.executed.price:.2f}, Cost: {order.executed.value:.2f}, Comm: {order.executed.comm:.2f}')
self.buyprice = order.executed.price
self.buycomm = order.executed.comm
else: # Sell
self.log(f'SELL EXECUTED, Price: {order.executed.price:.2f}, Cost: {order.executed.value:.2f}, Comm: {order.executed.comm:.2f}')
self.bar_executed = len(self)
elif order.status in [order.Canceled, order.Margin, order.Rejected]:
self.log('Order Canceled/Margin/Rejected')
self.order = None
def notify_trade(self, trade):
if not trade.isclosed:
return
self.log(f'OPERATION PROFIT, GROSS: {trade.pnl:.2f}, NET: {trade.pnlcomm:.2f}')
def next(self):
# Simply log the closing price of the series from the reference
self.log(f'Close, {self.dataclose[0]:.2f}')
# Check if an order is pending ... if yes, we cannot send a 2nd one
if self.order:
return
# Check if we are in the market
if not self.position:
# Not yet ... we MIGHT BUY if ...
if self.dataclose[0] > self.sma[0]:
# BUY, BUY, BUY!!! (with all possible default parameters)
self.log(f'BUY CREATE @MKT: {self.dataclose[0]:.2f}')
# Keep track of the created order to avoid a 2nd order
self.order = self.buy()
self.log(f'BUY CREATED Size: {self.order.size} @ MKT')
else:
if self.dataclose[0] < self.sma[0]:
# SELL, SELL, SELL!!! (with all possible default parameters)
self.log(f'SELL CREATE @ MKT: {self.dataclose[0]:.2f}')
# Keep track of the created order to avoid a 2nd order
self.order = self.sell()
self.log(f'SELL CREATED Size: {self.order.size} @ MKT')
import backtrader as bt
from atreyu_backtrader_api import IBData
from test_strategy import TestStrategy
import datetime as dt
from datetime import datetime, date, time
cerebro = bt.Cerebro()
data = IBData(host='127.0.0.1', port=7497, clientId=35,
name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
historical=True
)
cerebro.adddata(data)
# Set our desired cash start
cerebro.broker.setcash(100000.0)
# Add the test strategy
cerebro.addstrategy(TestStrategy)
# Add a FixedSize sizer according to the stake
cerebro.addsizer(bt.sizers.FixedSize, stake=10)
cerebro.run()
# Print out the final result
print(f'Final Portfolio Value: {cerebro.broker.getvalue():.2f}')
2021-09-01 23:59:59.999989, Close, 153.16
2021-09-01 23:59:59.999989, BUY CREATE @MKT: 153.16
2021-09-01 23:59:59.999989, BUY CREATED Size: 10 @ MKT
2021-09-02 23:59:59.999989, BUY EXECUTED, Price: 152.86, Cost: 1528.60, Comm: 0.00
2021-09-02 23:59:59.999989, Close, 153.59
2021-09-03 23:59:59.999989, Close, 154.29
2021-09-07 23:59:59.999989, Close, 156.76
2021-09-08 23:59:59.999989, Close, 155.00
2021-09-09 23:59:59.999989, Close, 154.27
2021-09-10 23:59:59.999989, Close, 148.67
2021-09-10 23:59:59.999989, SELL CREATE @ MKT: 148.67
2021-09-10 23:59:59.999989, SELL CREATED Size: -10 @ MKT
2021-09-13 23:59:59.999989, SELL EXECUTED, Price: 149.70, Cost: 1528.60, Comm: 0.00
2021-09-13 23:59:59.999989, OPERATION PROFIT, GROSS: -31.60, NET: -31.60
2021-09-13 23:59:59.999989, Close, 149.53
2021-09-14 23:59:59.999989, Close, 148.19
...
2022-08-08 23:59:59.999989, Close, 165.27
2022-08-09 23:59:59.999989, Close, 165.01
2022-08-10 23:59:59.999989, Close, 169.10
2022-08-11 00:00:00, Close, 168.89
Final Portfolio Value: 100409.00
Once a strategy has been back tested for basic operation it should be paper traded i.e. where simulated trades are generated and marked using live data to gain confidence in your overall trading strategy and goal.
Be careful to login to TWS using the Paper Trading option (see arrow):
In addition to selecting “Paper Trading” the execution of the strategy also needs to be changed to use IB as the broker to place orders and update positions.
import backtrader as bt
from atreyu_backtrader_api import IBData
from atreyu_backtrader_api import IBStore
from test_strategy import TestStrategy
import datetime as dt
from datetime import datetime, date, time
cerebro = bt.Cerebro()
ibstore = IBStore(host='127.0.0.1',
port=7497,
clientId=35)
data = ibstore.getdata(name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD' # Currency of SecurityType
)
cerebro.adddata(data)
broker = ibstore.getbroker()
# Set the broker
cerebro.setbroker(broker)
# Add the test strategy
cerebro.addstrategy(TestStrategy)
# Add a FixedSize sizer according to the stake
cerebro.addsizer(bt.sizers.FixedSize, stake=10)
cerebro.run()
...
2022-08-17 19:06:33.352691, Close, 174.84
2022-08-17 19:06:33.352741, Close, 174.84
2022-08-17 19:06:33.352781, Close, 174.84
2022-08-17 19:06:33.601432, SELL EXECUTED, Price: 174.83, Cost: 1749.40, Comm: 1.04
2022-08-17 19:06:33.601432, OPERATION PROFIT, GROSS: -0.10, NET: -2.14
2022-08-17 19:06:33.601432, Close, 174.83
2022-08-17 19:06:33.601462, Close, 174.83
2022-08-17 19:06:33.601522, Close, 174.85
2022-08-17 19:06:33.601522, BUY CREATE @MKT: 174.85
2022-08-17 19:06:33.601522, BUY CREATED Size: 10 @ MKT
2022-08-17 19:06:33.601533, Close, 174.85
...
Paper trading can help to validate the efficacy of a strategy, so it is prudent to continue paper trading until a stable performance and profitability profile has been achieved.
Sometimes successful paper trading strategies may not translate to the same performance in real trading. This can be for a myriad of reasons from external factors (market impact, price slippage, market microstructure etc.) to simple programming issues.
To help isolate and identify these issues it is important to have a robust monitoring and debugging facilities built into the trading process.
The simplest and most effective function is having accurate logging that records not only the trades but identifies the assets, opening and closing prices and trigger conditions and parameters pertinent to the strategy. Additionally the Backtrader and TWS API also has debugging built and can be accessed by setting the appropriate configuration parameters.
import backtrader as bt
import logging
# Generate file level logger
logger = logging.getLogger(__name__)
# Create a Stratey
class TestStrategy(bt.Strategy):
params = (
('ma_period', 15),
)
def log(self, txt, ts=None):
''' Logging function for this strategy'''
# Add log statement
logger.info(f'{txt}')
...
There is a example custom logger supplied in the atreyu_backtrader_api module, and can be referenced in the code as follows:
import logging
from atreyu_backtrader_api.custom_logger import setup_custom_logger
The logger has the following properties:
- It can write a named log file into a named directory thereby keeping all the logs together.
- It has log rotation, e.g. it does not overwrite logs on repeated restarts but backs each log keeping the specified logfile as the latest log.
- Each log line can be echoed to the terminal as well as written to the logfile
The logger can be setup as follows:
LOGGING_DEBUG = False
logger = setup_custom_logger(global_name = 'root',
debug_level = logging.DEBUG if LOGGING_DEBUG else logging.INFO,
filename = "sma_strategy.log", # Set the log file name
logdirname = './logs', # and the directory for the logs
console = True) # Echo the log line to the terminal
Configuring each sub-logger for the module and source file that is of interest, can control the logfile output.
LOGGING_DEBUG = False
logger = setup_custom_logger(global_name = 'root',
debug_level = logging.DEBUG if LOGGING_DEBUG else logging.INFO,
filename = "sma_strategy.log", # Set the log file name
logdirname = './logs', # and the directory for the logs
console = True) # Echo the log line to the terminal
# Log configuration for Atreyu Back Trader API module
logging.getLogger('atreyu_backtrader_api').setLevel(logging.ERROR)
# logging.getLogger('atreyu_backtrader_api.ibbroker').setLevel(logging.INFO)
# logging.getLogger('atreyu_backtrader_api.ibdata').setLevel(logging.INFO)
# logging.getLogger('atreyu_backtrader_api.ibstore').setLevel(logging.INFO)
# Log configuration for IB API module
logging.getLogger('ibapi').setLevel(logging.ERROR)
# logging.getLogger('ibapi.decoder').setLevel(logging.ERROR)
# logging.getLogger('ibapi.utils').setLevel(logging.ERROR)
# logging.getLogger('ibapi.client').setLevel(logging.ERROR)
#logging.getLogger('ibapi.decoder').setLevel(logging.DEBUG)
Enable Logging in Backtrader To enable logging in the Backtrader framework the _debug = True variable can be passed when setting up the IBStore and IBData classes.
PAPER_TRADING_PORT = 7497
LIVE_TRADING_PORT = 7496
cerebro = bt.Cerebro()
HOST = '127.0.0.1'
PORT = PAPER_TRADING_PORT
logger.info(f"Starting host: {HOST} port: {PORT}")
ibstore = IBStore(host=HOST,
port=PORT,
clientId=35,
_debug = True)
data = ibstore.getdata(name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
_debug = True
)
cerebro.adddata(data)
broker = ibstore.getbroker()
# Set the broker
cerebro.setbroker(broker)
Calling updateAccountTime('20:10')
Calling tickSize(16777217, 8, Decimal('361198'))
Calling tickString(16777217, 45, '1661454621')
Calling tickString(16777217, 84, 'Z')
Calling tickPrice(16777217, 4, 169.19, 4361192208: CanAutoExecute: 0, PastLimit: 0, PreOpen: 0)
Calling tickSize(16777217, 5, Decimal('100'))
Calling tickSize(16777217, 5, Decimal('100'))
Calling tickPrice(16777217, 1, 169.18, 4361192208: CanAutoExecute: 1, PastLimit: 0, PreOpen: 0)
Calling tickSize(16777217, 0, Decimal('1600'))
Calling get_acc_cash()
Calling tickPrice(16777217, 2, 169.19, 4361192448: CanAutoExecute: 1, PastLimit: 0, PreOpen: 0)
Calling get_acc_value()
Calling tickSize(16777217, 3, Decimal('1500'))
Calling get_acc_value()
Calling tickSize(16777217, 0, Decimal('1600'))
Calling tickSize(16777217, 3, Decimal('1500'))
Calling tickString(16777217, 32, 'PQXZUH')
Calling tickString(16777217, 33, 'KPQZNUH')
Now that you have implemented logging, and have and good understanding of how to debug your strategy you are ready to go live.
It is suggested that trading starts small, and that as confidence is gained trading sizes/frequency is increased within your prescribed risk appetite. At each stage it is recommended that you review the logs for anomalies that you may not notice from looking at the TWS screens.
Switching the strategy to live trading is as simple as logging in with TWS for trading and changing the port designator in the code below:
PAPER_TRADING_PORT = 7497
LIVE_TRADING_PORT = 7496
cerebro = bt.Cerebro()
HOST = '127.0.0.1'
PORT = LIVE_TRADING_PORT
logger.info(f"Starting host: {HOST} port: {PORT}")
ibstore = IBStore(host=HOST,
port=PORT,
clientId=35,
_debug = True)
data = ibstore.getdata(name="AAPL", # Data name
dataname='AAPL', # Symbol name
secType='STK', # SecurityType is STOCK
exchange='SMART', # Trading exchange IB's SMART exchange
currency='USD', # Currency of SecurityType
_debug = True
)
cerebro.adddata(data)
broker = ibstore.getbroker()
# Set the broker
cerebro.setbroker(broker)
The software is provided on the conditions of the simplified BSD license.
This project is not affiliated with Interactive Brokers Group, Inc.