Custom Data - is it possible to backtest on custom data without ingest?

Data
#1

Hi everyone,

I am new to Zipline library and wand te to check few things.

I have a “significant” amount of data in kdb. Now if my understanding is correct the only way to do backtesting via zipline is to ingest the data, so it cannot be custom? Is that correct? From my perspective ingesting every time seems unnecessary.

Any reason for such behavior? To avoid look-ahead bias?

Best,

-Alen

#2

I think the reason it has to be ingested is because zipline has a custom internal format called bcolz, and it stores the data in sqlite.

#3

Damn I was trying to build that just now using PostgreSQL and Polygon.io for my datasource.