I try to create intraday trading system and load some custom indicators / external data by per-minute basis. I try to use fetch_csv function for that purpose but without success. Based on source code here it looks like Zipline / Quantopian logic prevent us from intraday decisions and allow just to wait until end of the day / next day for some reaction on reality :
# Since we know this df only contains a single sid, we can safely
# de-dupe by the index (dt). If minute granularity, will take the
# last data point on any given day
df = df.groupby(level=0).last()
But I hope I am wrong. Do you guys have any experience in usage fetch_csv to load and use by-minute data ? If so, do you remember some non-widely-mentioned tricks to make it works ?