Personal automated trading libraries, fit with my favorite indicators calculated over DataFrames and sprinkled with a bit of ML
This project is a bit old, but could be easily refactored (minus the API wrappers, as the interfaces have changed)
The project abstracts a few different things, namely
- Quoter - A class for wrapping real-time websockets Bid/Ask from arbitrary exchanges, as well as historical quoters for OHLC data points
- Trader - Abstract class for executing orders, including autoconverting available balances into necessary asset for pair trading using a 'conversion graph'
- Indicators - A suite of my favorite indicators, including complex ones like Harmonic indicators and Demark indicators
- Risk Management - A position sizer for calculating safe margin sizing with fixed risk
- ML - machine learning model ensembler for optimizing classical trading strategies
Many indicators. Volatility, various oscillators, candlestick patterns, ZigZag, Harmonic Patterns, and including full implementations of all DeMark indicators.
Experiments with model ensembling using Boosted Trees, Random Forests, and simple neural networks to predict local inflection points for swing trading