Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Offline Development

Hello,
I am having trouble implementing some machine learning in my trading strategy. I understand the errors, its just a slow process developing in the web IDE, I tend to code very iteratively, I like my own IDE, basically the experience is just a bit slow.

I understand the data is protected, and that there exists an open-source implementation of the system which can be run locally, but it is only useful with your own data.

Would it be possible to make available simulated day interval data available for download? I am not looking to benchmark my algorithms outside of the system, but more to work out the bugs and develop the "meat" of it on my own machine. I would then transfer the code to your web-system and fine-tune it.

Maybe you could ask your data provider if shuffling the labels and/or intervals of trading would be sufficient randomization?

Thanks!
James

2 responses

It's pretty easy to use zipline with yahoo data, but I agree -- offline debugging is much faster and less tedious. I wish I had more time, I would be inclined to contribute to zipline. :)

@James, you can get 5 minute data from Stooq. That should be enough to prototype with.

http://stooq.com/db/