zipline-live / zipline

Zipline-Live, a Pythonic Algorithmic Trading Library
http://www.zipline-live.io/
Apache License 2.0
394 stars 65 forks source link

Proposal to add an out of the box support of some premium pricing data set from quandl #98

Open VitalyErmilov opened 6 years ago

VitalyErmilov commented 6 years ago

Dear Zipline Maintainers,

There is an open problem with WIKI prices dataset. The thing is that WIKI prices is not being updated for several days, the last pricing data there is for 28th of February, 2018.

I emailed the quandl's support and here is their answer: "Thanks for bringing this to our attention.

Your email has been forwarded to our data team so that this can be added to their development queue.

Please kindly note that as a free database, there is no guaranteed timeline by when our team will be able to investigate this issue. And, there is no guaranteed timeline by when they will be able to apply updates or corrections to the database.

If you need more reliable data, we recommend the following premium databases: EOD: https://www.quandl.com/data/EOD SEP: https://www.quandl.com/databases/SEP"

The presence of recent pricing data is not that important when backtesting strategies, but very essential when trading live. I think most of zipline-live users would consider to use payed data for live trading, so I propose to include a loader for one of the quandl's premium databases as an option, to make zipline-live to support reliable data out of the box

Sincerely, Vitaly

alphaville76 commented 6 years ago

One can take as basis the data ingestor for the ZEP quandl dataset: https://github.com/pbharrin/alpha-compiler/blob/master/alphacompiler/data/loaders/zacks_quandl.py

The SEP prices dataset is not only cheaper but also compatible with the SF1 fundamentals dataset from the same data vendor.

alphaville76 commented 6 years ago

I've just seen that @pbharrin implemented three days ago a loader for the SEP dataset: https://github.com/pbharrin/alpha-compiler/blob/master/alphacompiler/data/loaders/sep_quandl.py Very good, but there are some issues with the data as he pointed in his blog post: http://alphacompiler.com/blog/9/