Welcome to the @donutz.ai automated trading protocol! This bot leverages cutting-edge Reinforcement Learning ensembles, the SuperTrend indicator, Kalman Filter Forecasting, Simple & Exponential Moving Averages, Relative Strength Index (RSI), and other robust technical indicators to generate intelligent trade signals. It calculates and suggests profitable buy/sell orders for both Robinhood and Binance.US platforms using Python.
This project is an open-source playground for data scientists, quants, and trading enthusiasts alike. Fork, tweak, experiment, and enhance to your heart's content! The underlying decision science and forecasting principles here are widely applicable across predictive modeling, time-series analysis, and machine learning efforts well beyond trading and asset management. We only ask you to credit this repository and respect our community's guidelines.
For detailed insights into the bot's development and functionality, check out this Medium publication and this Kaggle Notebook, which showcases automated decision-making in action.
- Incorporate Social Sentiment Analysis: Bringing the pulse of the market to your trades.
- Add Advanced Indicators: Expanding our suite of technical tools.
- Pre-Built Trading Strategies: Allow users to select and deploy strategies tailored to their goals.
- Community Portfolio Mirroring: Users can create, share, and mirror peer portfolios for collective success.
- Autonomous Hyperparameter Tuning: Optimizing the bot’s decision-making as it learns and adapts over time.
We aim to optimize returns, minimize losses, and make trading as straightforward (and enjoyable) as possible. Join us in building a platform for traders of all levels—because making smarter trades shouldn’t be intimidating.
This unique feature lets users partially or fully mirror portfolio allocations of others in the community. Explore strategies, join public groups, or even create your own portfolios to share. When a group leader updates their allocations, all connected portfolios will automatically reflect the changes! It’s like being part of a live, dynamic investment team—designed for all risk appetites.
The bot is organized into three main components:
- Data Class: Handles live market data, including candlestick (OHLC) feeds and key indicators.
- Trade Strategy Class: A modular system allowing users to experiment with various trading strategies.
- Execution Class: Converts trade signals into live orders on Binance.US via the CCXT library.
Even though we’re working hard to release a fully user-friendly version, you can still test the bot locally. Fork the repo, give it a star, and follow our progress. If you have ideas to improve efficiency, scalability, or features—send them our way!
- Python (3.9.7+)
- Jupyter Notebook/Anaconda (Miniconda works too)
- Binance.US account (Sign up here)
- Clone this repo to your local machine (e.g.,
~/Desktop/GitHub).
Setting Up Binance.US API:
- Log in, navigate to API Management, and create a new API key.
- Safely store your
API_KEYandAPI_SECRETin aconfig.pyfile. Use this format for reference.
To execute the bot, navigate to its directory in your terminal and run:
$ python binance_bot.py
# For the Reinforcement Learning Mod, follow these instructions.
Components Overview
Data Class
Connects to Binance.US via CCXT WebSocket for live candlestick (OHLC) data. Includes rolling averages, Bollinger Bands, and binary indicators for trend detection.
Trade Strategy Class
A modular framework that supports a variety of trading strategies. Plug in your custom strategy—if it generates buy/sell signals, it’s good to go!
Execution Class
Executes buy/sell orders on Binance.US based on trade signals using CCXT integration.
Resources & Repositories
• Part Time Larry’s SuperTrend Crypto Bot
• Part Time Larry’s Binance Tutorials
• CCXT Binance.US Integration
• Binance Full History Dataset (2017-2020)
Stay hydrated, have fun, and happy trading! :chart_with_upwards_trend: