SignalStreet

Small price changes in stocks can mislead prediction models into overestimating trends, causing investors to make poor decisions. SignalStreet is a full-stack web application that filters out market noise and identifies genuine signals for buy or sell actions, providing investors with reliable insights.

Demo

Devpost Link

Features

• Noise Filtering: Advanced algorithms to distinguish between market noise and true signals.
• Real-time Analysis: Fetches live stock data and applies machine learning models for predictions.
• Interactive Dashboard: User-friendly React-based frontend for inputting stock tickers and viewing analysis results.
• Robust Backend: Flask API with data processing, ML inference, and observability.
• Comprehensive Testing: Automated testing suite using pytest and hypothesis for reliability.

Architecture

SignalStreet follows a client-server architecture with a clear separation of concerns:

• Frontend: React application built with Vite, handling user interactions and displaying results.
• Backend: Flask REST API serving analysis requests, integrating data fetching, ML predictions, and evaluation.
• Data Layer: Handles stock data retrieval using yfinance.
• ML Model: Pre-trained model loaded on-demand for predictions.
• Testing: Separate testing suite for backend components.
• [The Model]

Architecture Diagram

┌─────────────────┐
│   Frontend      │
│   (React/Vite)  │
└─────────────────┘
         │
         ▼
┌─────────────────┐
│   Backend       │
│   (Flask API)   │
└─────────────────┘
         │
    ┌────┼────┐
    │    │    │
    ▼    ▼    ▼
┌─────┐ ┌─────┐ ┌─────┐
│Data │ │ ML  │ │Eval │
│Layer│ │Model│ │uator│
└─────┘ └─────┘ └─────┘

Technologies Used

Backend

• Python 3.11.9: Core programming language.
• Flask 3.1.3: Web framework for the REST API.
• pandas & numpy: Data manipulation and numerical computing.
• yfinance: Library for fetching stock market data.
• joblib: Model serialization and loading.
• pytest & hypothesis: Testing frameworks for unit and property-based tests.

Frontend

• React 19.2.5: JavaScript library for building user interfaces.
• Vite: Build tool and development server for fast development.

Development Tools

• Git: Version control.
• npm: Package manager for frontend dependencies.

Setup and Installation

Prerequisites

• Python 3.11.9 or later
• Node.js and npm
• Git

Backend Setup

1. Navigate to the backend directory:

   cd signal-street/backend

2. Install Python dependencies:

   pip install flask pandas numpy yfinance joblib pytest hypothesis

3. (Optional) Install Flask-CORS if not included:

   pip install flask-cors

Frontend Setup

1. Navigate to the frontend directory:

   cd signal-street/frontend

2. Install dependencies:

   npm install

Running the Application

Backend

1. From the backend directory, run the API server:

   python api.py

   The server will start on port 5000 (or the next available port).

Frontend

1. From the frontend directory, start the development server:

   npm run dev

   The app will be available at http://localhost:5173 (or similar).

2. Open your browser and navigate to the provided URL to access the dashboard.

Running Tests

1. From the backend directory, run the test suite:

   python -m pytest Testing/

Usage

1. Open the frontend in your browser.
2. Enter a valid stock ticker (e.g., "AAPL").
3. Click "Analyze" to fetch data and get predictions.
4. View the results on the dashboard, which will indicate buy/sell signals based on filtered analysis.

Key Design Decisions

• Lazy Loading for ML Model: The ML model is loaded on-demand to reduce startup time and memory usage, improving application responsiveness.
• Input Validation and Security: Implemented regex-based validation for stock tickers and security headers (CSP, XSS protection) to prevent injection attacks and ensure data integrity.
• CORS Restrictions: Limited to localhost origins during development to enhance security.
• Error Handling and Logging: Comprehensive try-except blocks with logging to handle failures gracefully and aid debugging without exposing sensitive information.
• Timeout Handling: Frontend API calls include timeouts to prevent hanging requests and improve user experience.
• Separation of Concerns: Modular backend components (data layer, ML model, evaluator) for maintainability and scalability.
• Testing Strategy: Property-based testing with hypothesis to ensure robustness against edge cases.

Contributing

Contributions are welcome! Please fork the repository and submit a pull request with your changes. Ensure all tests pass before submitting.

License

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