Stock Price Predictor is a next-generation, enterprise-ready solution for predicting stock price movements using advanced AI. By combining insights from financial news with market data, it leverages state-of-the-art machine learning — specifically BERT for text analysis, LSTM for historical trend forecasting and uses the vector database for semantic search for the best prediction data retrieval (Weaviate) — to generate high-confidence predictions. Optimized for Apple’s latest hardware, the platform delivers fast, efficient performance, even in edge environments. A secure, user-friendly dashboard offers real-time analytics and intuitive visualizations. Integrated with a vector database, the system also supports intelligent information retrieval, making it valuable for investment research, risk assessment, and strategic decision-making.
- BERT Model – It is a language model which stands for which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations from unlabelled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks.
- LSTM Model - A Long short-term memory (LSTM) is a type of Recurrent Neural Network specially designed to prevent the neural network output for a given input from either decaying or exploding as it cycles through the feedback loops. The feedback loops are what allow recurrent networks to be better at pattern recognition than other neural networks. Memory of past input is critical for solving sequence learning tasks and Long short-term memory networks provide better performance compared to other RNN architectures by alleviating what is called the vanishing gradient problem.
LSTMs due to their ability to learn long term dependencies are applicable to a number of sequence learning problems including language modeling and translation, acoustic modeling of speech, speech synthesis, speech recognition, audio and video data analysis, handwriting recognition and generation, sequence prediction, and protein secondary structure prediction.
This research presents the development of an enterprise-grade stock price prediction system that integrates natural language processing (NLP) and time series (TS) analysis within a unified machine learning framework. The proposed architecture leverages a BERT-based transformer model for the semantic analysis of financial news and textual data, alongside a Long Short-Term Memory (LSTM) network trained on historical stock market data. A fusion mechanism combines the outputs of both models (BERT+LSTM) to enhance predictive performance. the complete application is optimized for execution or run on Apple Silicon using Metal Performance Shaders (MPS) and CUDA from Nvidia as well, enabling efficient on-device GPU acceleration. A secure and interactive dashboard, developed using Dash and Flask, facilitates real-time visualization and user interaction. Furthermore, a Weaviate vector database is employed to support semantic similarity search and contextual data retrieval, enhancing the interpretability and responsiveness of the system. The architecture is designed to be modular, scalable, and adaptable to production-grade environments, with built-in support for feedback monitoring and continuous learning.
- Combined BERT-LSTM model for advanced stock price prediction
- Apple Silicon (MPS) GPU acceleration support
- Secure dashboard with authentication
- Real-time stock price visualization
- Vector database (Weaviate) for efficient prediction storage
- Threading support for improved performance
- HDF5 file storage for model persistence
- Interactive charts and correlation analysis
- RESTful API endpoints
- Python 3.9+
- Docker and Docker Compose
- Apple Silicon Mac (for MPS support) or any machine with CUDA support
- 8GB+ RAM recommended
- Clone the repository:
git clone <repository-url>
cd LSTM-BERT-stock-predictor- Create a virtual environment (optional but recommended):
python -m venv venv
source venv/bin/activate # On Windows: .\venv\Scripts\activate- Install dependencies:
pip install -r requirements.txt- Start the Docker containers:
cd docker
docker-compose up -d- Access the dashboard:
- Open http://localhost:5000 in your browser
- Default login credentials:
- Username: admin
- Password: admin
The system consists of several components:
- BERT-LSTM Model: Combines BERT's language understanding with LSTM's sequential predic 9BBB tion
- Weaviate Database: Stores predictions and enables similarity search
- Dashboard: Secure web interface for visualization and analysis
- API: RESTful endpoints for model interaction
Edit config.yaml to customize:
- Model parameters
- Training settings
- Database connections
- Dashboard settings
- Security options
LSTM-BERT stock-predictor/
├── config.yaml # Configuration file
├── requirements.txt # Python dependencies
├── data/ # Data storage
├── docker/ # Docker configuration
├── models/ # Model artifacts
└── src/ # Source code
├── api/ # API endpoints
├── dashboard/ # Web interface
├── data/ # Data processing
├── models/ # ML models
└── utils/ # Utilities
python -m pytest tests/# Login
POST /api/auth/login
{
"username": "admin",
"password": "admin"
}
# Response
{
"token": "jwt-token"
}# Get predictions
GET /api/predictions/<stock_name>?days=30
# Make prediction
POST /api/predict
{
"stock_name": "AAPL"
}The system uses a combined BERT-LSTM architecture:
- BERT processes textual data and market sentiment
- LSTM handles time-series prediction
- Both models are combined for final prediction
To train the model:
python -m src.models.trainer- Real-time stock price visualization
- Model performance metrics
- Stock correlation analysis
- Prediction accuracy tracking
- User authentication and session management
- JWT-based authentication
- Password hashing with bcrypt
- Secure session management
- API endpoint protection
For production deployment:
- Update
config.yamlwith production settings - Set secure passwords and API keys
- Enable HTTPS
- Configure proper database backups
- Set up monitoring and logging
Copyright Protected, need permission from the researchers
If you want to contribute to this project, Let's connect. email - spand14@unh.newhaven.edu
BERT - https://doi.org/10.48550/arXiv.1810.04805 LSTM - https://doi.org/10.1162/neco.1997.9.8.1735 https://doi.org/10.1080/20430795.2024.2377551 Weaviate - https://weaviate.io/blog LSTM - https://developer.nvidia.com/discover/lstm#:~:text=A%20Long%20short%2Dterm%20memory,cycles%20through%20the%20feedback%20loops