Thanks Jessica,

In looking over the code, I don't see a check for open orders. Additionally, in one of my algorithms, I check that all securities have real trade data for the current tic. A few lines of code handles both cases:

# skip tic if any orders are open or any stocks did not trade  
    for stock in context.stocks:  
        if bool(get_open_orders(stock)) or data[stock].datetime < get_datetime():  
            return  

For a discussion of the filling issue, see https://www.quantopian.com/posts/thinly-traded-stocks-why-no-gaps.

The first check ensures that no orders are pending (re-balancing with open orders could result in an imbalance). The second check ensures that all of the trade data are real and not interpolated by Quantopian.

Also, you might consider what happens if the market opens late (after 9:30). Will the code still execute?

Grant

Grant,

Thanks for that, I've added a check for open orders and thinly traded securities into the code. However, I'm not too familiar with late market open orders. Would these occur in cases of natural disasters and etc.?

-Seong

Hi Seong,

I'll have to look over your code a bit more to understand if there may be a problem with your assumption that the market will always open at 9:30.

Grant

A list of NYSE closings:

http://www.nyse.com/pdfs/closings.pdf

Hi Grant,

Timing rebalance is something I'd expect each algo author to want to modify on their own since trade execution can be a proprietary part of the algo. We put the 9:30am trigger here really just as a placeholder - and you are absolutely right that it is a brittle part of the code.

Probably more robust is to replace that with a wider range trigger that looks to rebalance anytime between say 10am and 3pm - in that case you just need to make sure to also add a condition that will keep your algo from rebalancing more than once in the trading day (unless that's what you intend).

So in this example you might modify the method openofday_check to be intradingwindow_check and do something like this:

def intradingwindow_check(context):
# Converts all time-zones into US EST to avoid confusion
loc_dt = get_datetime().astimezone(timezone('US/Eastern'))
if loc_dt.hour > 10 and loc_dt.hour < 15:
return True
else:
return False

Best regards,
Jess

Thanks Jess,

With the code you and Seong posted, is there a way to go to 100% cash (completely out of the market)? I don't have time to try it now, but it seems that by setting all of the weights to zero, the rebalance function should go to cash right? Does it work?

Grant

Grant,

Yes, setting the weights of all securities to 0 will essentially exit all of your positions, leaving you in cash.

An example would be

context.weights = 0  
rebalance(context, data, 0.0, context.weights, 0.0)  

Hello Jesse and Seong,

I added variable weighting in the attached example (with random weights applied to test the code). Perhaps of interest to you or others.

By the way, any discussion at Quantopian to integrate github? It sure would be nice to track changes and collaborate on code.

Grant