Installing on an M1 machine can be a real pain in the rear end. Below are some notes that will hopefully help if you are trying to do this. Note that pip means pip3 (I have pip aliased to pip3)
Also, there is a script install_packages.sh in user_data/strategies/scripts, that will do all of this for you
WARNING: If you are running on an Apple processor (M1 etc) then there will likely be some dependency issues. The Mac M1 versions of libraries are generally only available via conda, and they do not keep up wit the pace of other packages such as python and tensorflow. In partuclar, be careful about upgrading python. At the time of writing, the latest version of python is 3.12, but the M1-based version of tensorflow-deps and tensorflow-macos only support python 3.10
You should only need to do this once
curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh
sh Miniconda3-latest-MacOSX-arm64.sh
Run pip(3) install on the following packages (these are not Mac-specific):
- finta
- prettytable
- PyWavelets
- simdkalman
- pykalman
- scipy
- scikit-learn
Of course, there may be package conflicts. Right now, there is a conflict with numba and numpy, so I have to do this:
conda install numba; pip3 uninstall numba
pip3 install numpy<1.24 # obviously check version if packages update
conda install -c apple tensorflow-deps
pip install tensorflow-macos
pip install tensorflow-metal
conda install -c conda-forge -y pandas jupyter
conda install pytorch torchvision -c pytorch
pip install darts
For some reason, conda does not work. Probably a mac package incompatibility
##talib Often, updating freqtrade fails with an error building talib. If that happens, try this:
pip3 uninstall talib
brew install talib
pip3 install talib
(it may be ta-lib)