Here is the daily backtest:

Hi Mandeep,

There are a few reasons why your minutely backtest differs greatly from the daily backtest. The biggest is that in minutely mode, your code in handle_data is executing every minute versus in daily mode where your code is executing everyday. So in daily mode, you're ordering once per day, but in minutely mode (potentially) you're ordering every minute.

The other reason why this can differ is that daily mode executes your orders, at the fastest, at the close price of the next day. I would suggest a read through this thread: https://www.quantopian.com/posts/trade-at-the-open-slippage-model for more information.

Anyways, a good way to go about modeling a daily algo in minutely mode is to use our schedule_function (https://www.quantopian.com/posts/announcement-new-schedule-function-method-allows-you-to-specify-when-a-function-runs) method to run your code at the end of day, everyday.

You can see an example attached to this post.

Hi Seong Lee,

Thanks for the response. I think you're right in that I might have to introduce the schedule function into my code. What confuses me is that with the shares = int(cash / currentprice) line, shouldn't this stop orders from being placed every minute in a minute backtest? In my minute backtest, it looks like orders are being placed regardless of how much cash the portfolio contains. I imagine if I were to set a schedule to trade at certain times, the minute backtest would behave the same way. Does the minute backtest behave different than I think it does?

In the original, this would have prevented the runaway as it was ordering again before the previous minute's order was filled.

def handle_data(context, data):  
    if get_open_orders():  
        return  

Thanks Gary, that was exactly what was missing. Hopefully this helps others in the future with the same problem.