1/3/2016 Update: The original problem still exists, being it is not possible to use a pipeline to select stocks to trade inside of a research notebook. The reason is that, before an algo is run in the research notebook, one must request the stock data for the algo to run on. Limiting you to the stocks you request, making it impossible to programmatically choose stocks to trade inside the algo based on the pipeline.
12/24/2015 Update: I was having a problem creating a pipeline to construct a list of stocks given a filter.
My Solution: Use a random stock to get historical date data, then run the algo w/ that date data, and use pipeline to actually select the stocks to trade. This enables the pipeline to run, but trading is not allowed on any stocks not requested in data = ... before the algo is run
Post:
The problem I'm having is in order to get the candidates I need trading date information. ie What are the 500 largest mkt cap stocks on 1-1-1999...
I can't get trading information for every stock in the universe, so I need the pipeline, but I can't construct the pipeline without the trading date.
I think I'm missing something because the main purpose of the pipeline is to construct a stock universe that updates.