- Sergey Chigrinov - [email protected]
- Dhruv Agrawal - [email protected]
- Man Sing Ho - [email protected]
This repository contains the code for the Capstone project for WQU university https://wqu.org Thanks a lot, Igor Tulchinsky, for this amazing opportunity!
This repository contains the code (can be used as a package) and notebooks implementing a trading strategy based on machine learning algorithms. Major part of the code was put into a package "WQUcapstoneCode". Notebooks contain visual representation of analysis steps and results. The idea and code are inspired by Marcos Lopez de Prado:"Advances in Financial Machine Learning" and https://www.kaggle.com .
Required libraries:
- most standard libraries - pandas v1.0.5, numpy v1.18.1, matplotlib v3.1.3, seaborn v0.10.0, sklearn v0.22.1
- fxcmpy - the main data source. You can skip it and utilize the csv files from the "input_data" folder instead
- scipy v1.4.1 - statistics
- tqdm v4.43.0 - progress bar - can be completely removed from the code if you wish
- lightgbm v2.3.0, xgboost v0.90 - ML algos - didn't make it into the final ensemble - they're too good, so prone to overfitting
- pyfolio v0.9.2+73.gcfdf82a - performance analytics
The high-level diagram is below:
