See if this does it for you. Trick is to use:

context.portfolio.positions.keys()  

Presumably if the number of shares in ticker XYZ drops to 0, XYZ is dropped from the dictionary. I would confirm this.

Thank you very much for the quick reply! When I use context.portfolio.positions.keys() I get the same error and crazily enough when I copy in your entire line of code with data.history into my algo I still get the exact same error! (TypeError: Expected assets argument to be of type or iterable of type Asset, str). Even though when I use it in your example algo everything works fine.

When I print context.portfolio.positions.keys() it still looks like i get the SID's just fine but for some reason doesnt work when i use it in the Order function or History function.

Thank you for the input. Seems like this problem is specific to my code then and I'm messing up my portfolio object somewhere along the way.

Robert,

Well, feel free to strip out anything you don't want to share, and post your code. Sounds like a mystery.

Grant

Ok just found the problem!

When entering my positions I was using this loop:

for Stock in context.Stock_List.index:  
    if context.portfolio.positions[Stock.sid].amount == 0:

This loop worked on Quantopian V1 however, for Quantopian V2 the Stock.sid somehow seems to be messing up the portfolio.positions object for any time I call it afterwards. Its now working correctly by switching my loop to this:

 for Stock in context.Stock_List.index:  
    if context.portfolio.positions[Stock].amount == 0:

Grant, thank you for your help. You definitely shortened my debugging time here big time.

Robert,

Glad you found the error. That may be a bug in the Quantopian software. I don't see why this code should modify your the positions:

for Stock in context.Stock_List.index:  
    if context.portfolio.positions[Stock.sid].amount == 0  

I'll send an e-mail to Quantopian support asking them to have a look.

Grant

Hi Robert,

To answer your original post, data.history() expects either an Equity object (the type returned by sid()/symbol()), or a list of Equity objects. Similarly, context.portfolio.positions is a dictionary keyed by Equity objects. As well, the context.portfolio.positions object behaves like a defaultdict. If a key doesn't exist in the dict when it's referenced, a default (zero) entry will be made. Combining these two facts, using Stock.sid, which returns the integer value of the security ID will never get you a non-zero amount.

On top of this, there's a bug with context.portfolio.positions. It's supposed to get rid of all zero-amount entries in the dict at the end of each day. This is something that we were recently made aware of and our engineers are looking into it.

It's supposed to get rid of all zero-amount entries in the dict at the end of each day.

It doesn't say that in the help. In fact, it implies that only current open positions should be found ("The position object represents a current open position..."). It would seem to be a lot less confusing if you immediately dropped entries when they drop to zero. I suppose one can always loop over the keys and collect only the ones that are non-zero, but you could do that under the hood.

Also, I don't understand the discussion about being able to modify the keys. Your code shouldn't allow this, right?