The moving average crossover strategy is a popular algorithmic trading approach. This moving average crossover strategy analyzes when fast and slow moving averages intersect to generate trading signals with Interactive Brokers and Python.
Understanding Moving Average Crossover Strategy
Moving average crossover strategy is one of the popular strategies that lots of traders have paid attention to. The details of the strategy is described at Wikipedia. https://en.wikipedia.org/wiki/Moving_average_crossover
In a short summary of moving average crossover strategy, the trend of the security is going up when the fast moving average line cross over the slow moving average line from the lower area and it is a signal about long positions.
In the following example, the code calculates the moving average of 5 (fast moving average line) and 15 (slow moving average line) at 15:59:00 US Eastern time, 1 min before the market closes, on every trading day. It places an order of SPY, ETF tracking S&P 500, when the fast moving average line starts to cross over the slow moving average line and exits the positions when crossover happens in the opposite direction.
The code can be backtested at Quantopian.com because IBridgePy can run most of the strategies posted at Quantopian without any changes.
If you need any help to live trader your codes from Quantopian, please contact with us at IBridgePy@gmail.com
# -*- coding: utf-8 -*- ''' There is a risk of loss in stocks, futures, forex and options trading. Please trade with capital you can afford to lose. Past performance is not necessarily indicative of future results. Nothing in this computer program/code is intended to be a recommendation to buy or sell any stocks or futures or options or any tradable securities. All information and computer programs provided is for education and entertainment purpose only; accuracy and thoroughness cannot be guaranteed. Readers/users are solely responsible for how they use the information and for their results. If you have any questions, please send email to IBridgePy@gmail.com ''' # This code can be backtested at Quantopian import pandas as pd def initialize(context): context.security=symbol('SPY') # Define a security, SP500 ETF # schedule_function is an IBridgePy function, also supported by Quantopian # date_rules.every_day : the dailyFunc will be run on every business day # time_rules.market_close(minutes=1) : the time to run dailyFunc is # 1 mintue before U.S. market close (15:59:00 US/Eastern time) schedule_function(dailyFunc, date_rule=date_rules.every_day(), time_rule=time_rules.market_close(minutes=1)) def dailyFunc(context, data): # dailyFunc is scheduled by schedule_function # It will run at 15:59:00 US/Eastern time on every business day hist = data.history(context.security, 'close', 20, '1d') # Calculate 5 days and 15 days moving average MA5 = hist.rolling(window=5).mean()[-1] MA15 = hist.rolling(window=15).mean()[-1] current_price = hist[-1] # Get current position current_position = context.portfolio.positions[context.security].amount print('current_position=', current_position) print('current_price=', current_price, 'MA5=', MA5, 'MA15=', MA15) # Buy signal: MA5 crosses above MA15 if MA5 > MA15 and current_position == 0: order_target_percent(context.security, 1.0) print('Buy', context.security) # Sell signal: MA5 crosses below MA15 elif MA5 < MA15 and current_position > 0: order_target_percent(context.security, 0.0) print('Sell', context.security)
Implementation Tips
When implementing technical indicators, consider using longer lookback periods during volatile markets. Always backtest your parameters thoroughly before live deployment.
For more trading strategies and examples, check our tutorials or visit our documentation.
Hi!
The following message
AttributeError: module ‘pandas’ has no attribute ‘rolling_mean’
is displayed when running the code. Any ideas how to fix it? Thank you.
Try just importing pandas.rolling_mean at the top of whatever script your running
Hi, I can show the real time price of ‘CASH,EUR,USD’ without any problem, but I got an error when I try to show the real time price of CL light sweet crude oil futures(WTI) Nov 2018 with the command line below. please let me know what the right format is for Futures, thanks a lot.
print (“CL 201811 ask_price=”,show_real_time_price(symbol(‘FUT,CL,USD,201811′),’ask_price’))
https://ibridgepy.com//ibridgepy-documentation/#superSymbol
Hi,
Tried importing pandas.rolling_mean but got an error below.
ModuleNotFoundError: No module named ‘pandas.rolling_mean’
Any idea to fix it? thanks
Can I use only IBridgePy & TWS without Quantopian?
Your code, as presented, above “runs” error-free but no results.
Is there a complete example (single script) that does say MA crossover in Spyder(3.2.8) with Anaconda36_64 including connection to kernel?
thanks
I have the same issue “Your code, as presented, above “runs” error-free but no results.”
The trading decisions are made once a day. If you run the code for a few minute, it is very likely you won’t see any meaningful results.
I had a question what should I do if my algo gets disconnected due to some issue so all my information saved in context variables would lost what should I do if my algo gets disconnect and I reconnect all varible info is loaded again ?
If you hard-code information in context variables and run your algo again, they will show up again. If you save information to context variables during your code, you may consider to write to a csv file and read them in as needed.
what logic should one put to detect the algo has disconnected and when
what is in my mind keeping the iteration count of handle data and one the algo gets disconnected iteration count and then i start algo say 5 minutes later say algo got disconnected at 10:15 and i restarted algo at say 10:30 . and lets handle data is running every minute so in my variable where i was recording the handle data iterration count it would have 60 assuming market started at 9:15 and when i restarted the algo at 10:30 iteration count of handle data should have been 75 which is not so program would know the algo was disconnected and load information in context variables through CSV where I have recorded them do u think this is a good logic to loading a previous information in context variables when algo gets disconnected or can u give some suggestions