https://github.com/devmaan707/stock.ai
Stock.ai is a comprehensive platform for predicting stock market movements using machine learning models, real-time data analysis, and sentiment analysis of financial news. The system continuously learns from its predictions to improve accuracy over time.
https://github.com/devmaan707/stock.ai
python stock-market stock-price-prediction tensorflow
Last synced: about 2 months ago
JSON representation
Stock.ai is a comprehensive platform for predicting stock market movements using machine learning models, real-time data analysis, and sentiment analysis of financial news. The system continuously learns from its predictions to improve accuracy over time.
- Host: GitHub
- URL: https://github.com/devmaan707/stock.ai
- Owner: DevMaan707
- Created: 2025-03-15T04:16:01.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-15T09:28:23.000Z (over 1 year ago)
- Last Synced: 2025-03-15T10:26:27.273Z (over 1 year ago)
- Topics: python, stock-market, stock-price-prediction, tensorflow
- Language: Python
- Homepage:
- Size: 66.4 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Stock.ai - Advanced Stock Market Prediction System
Stock.ai is a comprehensive platform for predicting stock market movements using machine learning models, real-time data analysis, and sentiment analysis of financial news. The system continuously learns from its predictions to improve accuracy over time.
## Features
- **Real-time Stock Data Analysis**: Fetch and process stock data from Yahoo Finance
- **Advanced ML Predictions**: LSTM neural networks for time series forecasting
- **Technical Indicators**: RSI, MACD, Bollinger Bands, and more
- **Sentiment Analysis**: Analyze financial news for market sentiment
- **Recommendation Engine**: Generate actionable buy/sell recommendations
- **Performance Tracking**: Track and validate prediction accuracy
- **Continuous Learning**: Models that improve with each prediction
## System Architecture
```
├── core/ # Core system components
│ ├── data_manager.py # Fetches and manages stock data
│ ├── database.py # Handles data storage and retrieval
│ ├── feature_engineer.py # Creates features for ML models
│ ├── model_manager.py # Manages ML model lifecycle
│ ├── recommendation_system.py # Generates actionable recommendations
│ └── stock_prediction_system.py # Main system orchestration
├── models/ # ML model definitions
├── services/ # External service integrations
│ ├── news_service.py # Fetches financial news
│ ├── sentiment_service.py # Analyzes news sentiment
│ └── llm_service.py # Large language model integration
├── ui/ # User interface components
└── utils/ # Utility functions and helpers
```
## How It Works
### Workflow
1. **Data Collection**:
- Stock price data is fetched from Yahoo Finance
- Financial news is gathered from News API
- Data is processed and stored in the local database
2. **Feature Engineering**:
- Technical indicators are calculated (RSI, MACD, etc.)
- News sentiment is analyzed and integrated with price data
- Features are normalized for model input
3. **Prediction Generation**:
- LSTM models process feature data to predict future prices
- Confidence scores are calculated for each prediction
- Buy/sell/hold recommendations are determined
4. **Result Validation**:
- Previous predictions are compared with actual outcomes
- Model performance metrics are updated
- Rewards/penalties are assigned to improve future predictions
5. **Continuous Learning**:
- Models are retrained periodically (default: every 6 hours)
- New market data and prediction results enhance model accuracy
- System performance improves over time
## Installation
### Prerequisites
- Python 3.8 or higher
- pip package manager
- TA-Lib installed on your system (for technical indicators)
### Setup
1. Clone the repository:
```bash
git clone https://github.com/yourusername/stock.ai.git
cd stock.ai
```
2. Run the setup script:
```bash
bash scripts/setup.sh
```
This will:
- Create a virtual environment
- Install required packages
- Set up the database and directories
3. Configure your `.env` file:
```
NEWS_API_KEY=your_news_api_key
LLM_ENDPOINT=http://localhost:8080/generate
DATABASE_PATH=data/stock_predictor.db
MODEL_PATH=models/
LOG_PATH=logs/
```
## Usage
### Basic Run
To start the system with default settings:
```bash
bash scripts/run.sh
```
Or run directly with Python:
```bash
python src/main.py --symbols AAPL GOOGL MSFT AMZN --interval 60
```
### Command Line Arguments
- `--symbols`: List of stock symbols to monitor (default: AAPL GOOGL MSFT AMZN)
- `--interval`: Prediction interval in minutes (default: 60)
### Changing Run Frequency
To adjust how often the system makes predictions:
```bash
python src/main.py --symbols AAPL GOOGL MSFT --interval 30
```
This will run prediction cycles every 30 minutes instead of the default 60.
### Training Schedule
By default, models are retrained every 6 hours. To change this, modify the `run` method in `src/core/stock_prediction_system.py`:
```python
if datetime.now().hour % 12 == 0: # Change from 6 to 12 for 12-hour retraining
await self.train_models()
```
## Monitoring
- Predictions and system status are displayed in the console
- Detailed logs are saved to `logs/stock_predictor.log`
- Database stores all predictions and performance metrics
## Data Storage
- **Raw Data**: Stored in `data/raw/`
- **Processed Data**: Stored in `data/processed/`
- **Database**: SQLite database at `data/stock_predictor.db`
- **Models**: Saved to `models/` directory
## Advanced Configuration
Edit `src/utils/config.py` to change system-wide settings:
- `LOOKBACK_DAYS`: Days of historical data to use (default: 500)
- `TRAINING_EPOCHS`: Number of epochs for model training (default: 10)
- `BATCH_SIZE`: Batch size for training (default: 32)
- `TECHNICAL_INDICATORS`: List of technical indicators to use
## Extending the System
### Adding New Stock Symbols
Simply add them to the command line arguments:
```bash
python src/main.py --symbols AAPL GOOGL MSFT AMZN TSLA NFLX
```
### Adding Custom Technical Indicators
1. Extend the `add_technical_indicators` method in `src/core/feature_engineer.py`
2. Update the `feature_list` in the `FeatureEngineer` class
3. Retrain models to incorporate the new features
### Customizing Recommendation Logic
Modify the `analyze_stock` method in `src/core/recommendation_system.py` to adjust how recommendations are generated.
## Troubleshooting
- **Missing Data**: Ensure your internet connection is stable for API access
- **Model Training Errors**: Check if you have enough historical data for selected symbols
- **TA-Lib Errors**: Verify TA-Lib is properly installed on your system
## License
This project is licensed under the MIT License - see the LICENSE file for details.
## Acknowledgments
- Yahoo Finance for stock data
- NewsAPI for financial news
- TensorFlow and Keras for machine learning capabilities
---