Why `observer.array` so many `nan`?

[asdasda]

Seems the nan count same to the non nan count, in other words, the nan is a half of the observer.array.

[vladisld]

@asdasda Observer's array member actually refers to observers's first line (LineBuffer object) array member. This array member is used internally for LineBuffer data storage and usually not used/accessed/manipulated directly unless one knows exactly what she/he is doing. In some cases it may be of larger size than loaded/filled data - in which case the extra size is filled wit nans. LineBuffer actually defines its own interface to access the underlying array buffer ( for example it defines the currently active index, access to the past and future values, and many more)

It is better to use a regular documented ways to access the line data. Please see the following docs:
https://www.backtrader.com/docu/concepts/#lines

Also the following post may be useful as well:
https://community.backtrader.com/topic/3921/zero-records-while-loading-data-from-csv/

In case you really would like to access the internal buffer directly, please take a lool at the LineBuffer's implementation first:
https://github.com/mementum/backtrader/blob/master/backtrader/linebuffer.py

[asdasda]

@vladisld How can I access the full line data of observer after cerebro.run()?

[vladisld]

@asdasda Using slicing could be an option. Docs here