System is composed by 4 folders:
- Strategies (where I am only going to upload 2 basic examples)
- Utilities
- GUI
- DB
Data is retrieved using ib_isyinc and ccxt which can be stored in a local SQLite3 database for further usage.
Current state: Compute +80 TA indicators and ~10 other features (NLP [[Twitter, StockTwits, wallstreetbets,Google Trends, yfinance news] using Blob, Vader and a Transformer model], vol, spread, order book imbalance, etc) from OHLCV data queried from SQLite DB --> Principal Component Analysis --> Ensemble[AutoGluon (ML ensemble: 52% accuracy OOS), Facebook Prophet, more to be added]
Furthermore, a GUI is also being configured (built using PyQT5) for visual assistance where you can see a live P&L graph, models training and performance and order log. Tick interval for data to be downloaded can also be configured here
first install https://sqlitebrowser.org/dl/ to assess DB and create account in respective exchanges in order to get the API key.
create_DBs.py
populate_DBs.py
get_prices.py # Use Windows Task Scheduler, Crontab or Docker to run this script every x interval
Main.py
python mainwindow.py # initialize GUI in order to start the system
=======
main.py
This project is also my Master thesis which I will present in January and as such, until then, regular updates will be made to this repo.