cristian-bicheru / fast-ta

Python Technical Analysis Library For Big Data
https://fast-ta.readthedocs.io/en/latest/index.html
MIT License
10 stars 4 forks source link
c finance numpy python3 technical-analysis technical-analysis-library

Build Status codecov

Cloning Source Code:

git clone https://github.com/cristian-bicheru/fast-ta.git && cd fast-ta && git submodule update --init

Building:

python3.x setup.py build_ext --inplace

OR using CMake

mkdir test_build && cd test_build
cmake -D<arch>=1 ..
make -j
``
For debugging purposes, you can use `cmake -DCMAKE_BUILD_TYPE=Debug -D<arch>=1 ..` (where `<arch>` is either SSE2, SSE41, AVX, AVX2, or AVX512, or you can omit the `-D<arch>=1` entirely for a SIMD-free build.) 
NOTE: without `-DCMAKE_BUILD_TYPE=Debug` the compiler may introduce SIMD optimizations.

Building with MSVC:
```bash
mkdir test_build && cd test_build
cmake ..
msbuild fast-ta.sln

NOTE: This requires msbuild to be in PATH, also make sure the selected Python distribution was installed with debug binaries. If not, re-run the installer and tick the option.

Testing:

To run CI tests:

./test.sh

All of these must pass for any code to be added to the repo.

For general, eyeball testing you can generate plots of the indicators with this script.

python3.x tests/tests.py --show-plots --save-plots

Benchmarks:

REQUIRES: ta

python3.x benchmarks/<indicator>.py

OUTPUTS: SVG plotting times speedup

Useful Resources:

https://docs.python.org/3/c-api/index.html

https://docs.scipy.org/doc/numpy/reference/c-api.html

https://db.in.tum.de/~finis/x86-intrin-cheatsheet-v2.1.pdf

TODO: