Tensorflow with Backtrader
first tests with Backtrader works fine, and a sample-strategie can be backtested via IB-API.
It is installed by GitHub-Download and "python setup.py install" on top of Python=3.8.5
Now Tensorflow needs to be added and requirements need to be exactly fullfilled.
Due to my admin-experiences I am very exact with versions, just to avoid hellfire caused by version-inaccuracies even when it worked fine in first place.
With Backtrader and Tensorflow there is now a conflict in requirement that is a not solvable in a satisfying way:
"Backtrader-requirements: Python 3.2 / 3.3/ 3.4 / 3.5"
Tensorflow-requirements: Python 3.5–3.8
Sounds ok with 3.5
...but when installing Tensorflow ontop of Python=3.5 "conda" can install only Tensorflow 1.10.
And when explicitly installing Tensorflow=2 it says, it needs Python 3.6
Therefore 3.6 is essential.
But is this ok for Backtrader?
How strict are the requirements for Backtrader?
Just read about https://github.com/backtrader2/backtrader
1.) There is sais even 3.7 is ok for backtrader2, but this GitHub-Repository is just a copy and not yet vaild?
2.) Can I install Backtrader on Python=3.6? Anyone has experiences with this?
run-out last edited by
Backtrader is tested for up to 3.7.