I'm pretty frustrated with the organization of the documentation.

For instance, look at this page:

https://www.backtrader.com/docu/indautoref.html

• The left column is a list of pages of varying topics and depths in a seemingly random order.

• The search is ineffective in finding reference documentation, e.g. "where is the strategy reference docs".

• There is no index to for the functions (!WTF?!). The only way to find a function documentation is to use in-page search, which is ineffective for common terms like "Average".

It seems like great effort has been put into the documentation and there is so much there, it's a shame it isn't organized in a usable way, particularly for developers just needing API reference documentation.

Am I missing something?

I've been developing on backtrader for a bit to test out some trading theories and learn the ropes, however I'm considering testing the waters with QuantConnect. While I love the flexibility of running all my own code and the lack of IP risk, I have had to create a rough tool chain to process historical data acquired from various sources, particularly in crypto. That data seems to be included for free with QuantConnect. Not having to deal with my own systems and scaling is also appealing. QuantConnect offers plethora of tick level historical data and live feeds, which seem like they would be both expensive and time consuming to acquire and process.

What are people's experience with acquiring historical and live data feeds versus an integrated online system like QuantConnect? The appeal of focusing only on writing algorithms is a bit hard to resist.

I have tick data and would like to include the money flow information I can gather in resampled data, e.g. +/- for buy/sell side ticks and accumulate to a money flow total for the resample period.

In order to do this I have done something very ugly and modified dataseries.py to += instead of = the openinterest in the bupdate method. This works but is not sustainable, and I may want to do similar things in the future.

Is there a better way to get this behavior? I'm not clear on the implications of adding a new field.

Thanks for the response. I apologize for my mistake - I had added a _name to my observers as well, which seem to not accept that parameter.

When trying to add an analyzer with a _name parameter, it says: init() got an unexpected keyword argument '_name'

I have written a custom data feed which resamples large amounts of tick data. That is of course quite slow. When tick data is not necessary I would like to start with less granular data.

I know that I can add a writer to the resampler, however the output is not immediately useable as a data source. Is there some simple way to resample and output a usable data file?