Stock Sentiment AI Trading Bot

This is a stock sentiment trading bot that uses news articles and GPT-3.5-turbo from OpenAI to analyze stock sentiment and make buy/sell decisions based on the sentiment score. It currently supports backtesting for Google, Apple, and Tesla.

Requirements

• Python 3.7 or higher
• pip for installing dependencies
• OpenAI API key for GPT-3.5-turbo
• NYTimes API key for fetching news articles
• Yahoo Finance for fetching stock data

Installation

Clone this repository:

git clone https://github.com/yourusername/stock-sentiment-trading-bot.git
cd stock-sentiment-trading-bot

Install the required dependencies:

pip install -r requirements.txt

Set up environment variables for your OpenAI API key and NYTimes API key:

export OPENAI_API_KEY="your_openai_api_key"
export NYT_API_KEY="your_nyt_api_key"

Usage

Open the bot.py file and ensure the stock symbols and company names are defined in the stocks dictionary.

stocks = {
"Google": "GOOG",
"Apple": "AAPL",
"Tesla": "TSLA"
}

You can add or remove stock symbols and company names as needed.

Run the script:

python bot.py

The script will fetch historical news data and stock data, perform backtesting, and display the results. It will also cache the fetched data to speed up future runs.

Customization

You can customize the bot's behavior by modifying the following parameters:

• sentiment_threshold: Adjust this value in the SentimentStrategy class to change the threshold for buy/sell decisions based on sentiment scores. A higher value will make the bot more selective in its buy/sell decisions, while a lower value will make it more responsive to sentiment changes.

class SentimentStrategy(Strategy):
params = {'sentiment_threshold': 0.2} # ...

• cash: Adjust the initial cash available for the backtest in the Backtest instantiation.

bt = Backtest(combined_df, strategy_class, cash=10000, commission=.002, exclusive_orders=True)

• commission: Adjust the commission per trade in the Backtest instantiation.

bt = Backtest(combined_df, strategy_class, cash=10000, commission=.002, exclusive_orders=True)

Troubleshooting

If you encounter any issues or errors, please double-check your API keys and ensure you have the latest versions of the required dependencies. If the issue persists, create an issue on this repository with a detailed description of the problem, and we will assist you.

This project is licensed under the MIT License. See the LICENSE file for details.