Cognitive Trading (CogniTrade) is an AI-powered trading bot designed to trade autonomously using both traditional trading strategies and machine learning models.
- Automated trading using Alpaca API
- Backtesting framework with Python & YahooDataBacktesting
- Transformer model integration for trend prediction through sentiment analysis of current news
- Create a baseline bot using a framework like
lumibot. - Implement dynamic position sizing based on cash risk and set limits such as take profit and stop loss orders.
- Retrieve news articles using the
alpaca-trade-api-pythonand format them for sentiment analysis. - Leverage the FinBERT model to estimate the sentiment of news articles and generate a probability score.
- Incorporate the sentiment signal into the trading algorithm to trigger buy or sell orders based on strong positive or negative sentiment.
Follow these steps to set up your environment and run the bot:
-
Clone the Repository:
git clone https://github.com/yourusername/ai-trading-bot.git
-
Create a Virtual Environment: Use
venvto create a virtual environment for managing dependencies:python -m venv myenv myenv\Scripts\activate # for Windows
-
Install Required Libraries:
pip install lumibot timedelta alpaca-trade-api==3.1.1 pip install torch torchvision torchaudio transformers
-
Configure Environment Variables: Create a
.envfile with your API keys:API_KEY = YOUR_KEY API_SECRET = YOUR_KEY BASE_URL = "https://paper-api.alpaca.markets/v2" -
Run the Trading Bot:
python trading_bot.py
For detailed results, please refer to the results folder.
Contributions to the trading bot are welcome. Please fork the repository and submit a pull request with your features or fixes.
This bot is for educational purposes only. Use at your own risk. No financial advice. Always remember, live trading carries risk and it's important to test extensively in simulation environments before deploying any trading strategy with real capital.