Hi Karen,
The use case I have in mind is to be able to do something like Simon illustrates here:
https://www.quantopian.com/posts/equity-long-short
However, rather than doing some/all of the universe selection in
before_trading_start
the idea would be to have a more logical flow, setting up the universe within pipeline, using the fundamentals database, and then applying additional filters/optimizations within pipeline to the subset of securities. I think this would imply having access to all fundamentals fields within pipeline. Is that the goal?
Then before_trading_start
could be used to further refine the security selection/weighting/signal generation, using minute data. This is sorta clunky as things stand today, as I show on https://www.quantopian.com/posts/code-for-getting-minute-data-into-before-trading-start. Since history
is not supported in before_trading_start
(and I don't think update_universe
takes effect within before_trading_start
), one has to use the output of pipeline from the prior day, and the corresponding history
output from the end of the prior day, tacking on data from the opening minute of the current day.
For example, say I wanted to set up a proxy universe for an index (e.g. Russell 2000 or NASDAQ 100). Within pipeline, I'd make the selection, and then do further refinements/filtering within pipeline, using daily data. These data would flow into before_trading_start
where minute data could be applied. Finally, if any minute-by-minute decisions needed to be made, either handle_data
or a scheduled function would be used.
Grant