FastTest VS Lean

Wrote my own in C++ with python wrapper: https://github.com/ntorm1/FastTest. It's not stable so do with that you will, but no problems handling thousands of symbols at whatever frequency.

Not sure exactly, depends a lot on your existing implementation. I implemented it as a C++ class that overloads the () operator to access the underlying data. For reference here is my implementation of it: https://github.com/ntorm1/FastTest/blob/main/include/Asset.h

6.) You cant maximize position size of spread options strategies, LEAN always assumes naked margin first and ON THE NEXT DATA FRAME - you get your reg-t margin for spreads. https://github.com/QuantConnect/Lean/issues/5693 We're running into 2 years of this issue being reported.

Thanks. I will check it out. Can you point me to the right part of the codebase I should be looking at? I know it must be somewhere here - https://github.com/QuantConnect/Lean

They have nice documentation on their website for how to download/import external data, though. They even have GitHub examples that pull data from Dropbox.

Here is an example https://github.com/QuantConnect/Lean/blob/master/Algorithm.Python/DropboxUniverseSelectionAlgorithm.py

There's also the open-source lean.io stuff by QuantConnect which looks decent and comes with data. It's pretty inexpensive too.

I can plot OHCL lines, I can plot OHCL scatter points on the chart, why can't I plot candles? It's just those scatter points put together. This thing has been going on for 8 YEARS.

Here's a link to one of the strategy examples from the QC team that does this -- loading a universe from an external source. https://github.com/QuantConnect/Lean/blob/master/Algorithm.Python/DropboxCoarseFineAlgorithm.py

+1 see https://github.com/QuantConnect/Lean