As I do the back testing I happen to find out there is a very big price gap by SDS on 2017-07-18.
But this doesn't happen by Notebook.
As I do the back testing I happen to find out there is a very big price gap by SDS on 2017-07-18.
But this doesn't happen by Notebook.
There was a 4:1 reverse stock split for SDS effective on market open 7/17/2017. See http://www.proshares.com/news/2017-6-27.html
Prices (and associated volumes) are identical between an algorithm (the IDE) and research (the notebooks). The one difference however is the 'as of' split and dividend adjustment dates. This is what accounts for any apparent differences between the two. They are the same just maybe 'adjusted' to different 'as of' dates.
Algorithms (the IDE) always 'see' the actual data as of the current algorithm date with all previous dates split adjusted to that current date. The research environment (notebooks) prices, when using the the 'get_pricing' method, are adjusted as of the end date of 'get_pricing'. So, unless that end date is the same as the current date in an algorithm, the prices may appear different.
Take a look at this post https://www.quantopian.com/posts/help-notebook-vs-algorithm. @Jamie McCorriston has a good explanation in his post towards the end.