I am new to Alpaca but experienced in algo trading. I am interested to learn about your current and continued support for Zipline integration.
I see two Alpaca GitHub projects related to Zipline
-
pylivetrader
at https://github.com/alpacahq/pylivetrader -
alpaca-zipline
at https://github.com/alpacahq/alpaca-zipline
It looks like the first is under continued development and the second is not. I am interested in using Alpaca to both backtest and live trade a Zipline-flavored strategy. Can you provide guidance and best practices given your two packages for how to operate a research-to-production workflow:
- backtest a Zipline-flavored strategy locally versus a local data bundle
- transition to live-trading the same strategy without changing any of the strategy code (i.e., just flipping command line flags)
In order for that to work properly, the historical data in both cases must be the same. Meaning a data.history()
call for multiple days of history should draw on the same source for historical data, but append the live price in the case of live trading. Is the alpaca-bundle
still supported and would that serve as a proper historical data source in both cases? In the case of the backtest, the daily state is carried over from T-1 whereas in live trading it comes from the start-of-day update.
Thank you, and fantastic work to-date on Alpaca!