I'm attaching a notebook which displays a behaviour which I believe to be a bug (but correct me if this isn't so).
It's a pipeline result in form of a DataFrame. The index is a MultiIndex of (date, Equity) and the rows are fundamental data. Usual DataFrame behavior is that you can select a specific row by passing full MultiIndex "coordinates" into .loc, or you can slice one or other of the indexes by passing :
. See notebook for more details. The problem seems to be that grabbing an Equity by its sid fails in some cases.