📊 SentimentSignal - Social-Driven Market Intelligence

Professional sentiment analysis dashboard for retail investors

SentimentSignal is a sleek, AI-powered trading dashboard that analyzes market sentiment from news sources and generates actionable trading signals for NIFTY 50 stocks. Built with Flask, transformers (FinBERT & RoBERTa), and a premium dark-mode frontend.

---

✨ Features

🧠 AI-Powered Sentiment Analysis

• Dual-Model Approach: Combines FinBERT (financial news) and RoBERTa (social context) for accurate sentiment scoring
• Real-Time News Aggregation: Fetches historical news from Google News RSS
• Sentiment Scoring: Classifies sentiment as Positive, Neutral, or Negative with confidence metrics

📈 Interactive Visualizations

• Sentiment vs Price Chart: Overlay sentiment trends on stock price movements
• Buy/Sell Signal Markers: Visual indicators for trading signals
• Time Range Selection: Analyze 1M, 3M, or 6M periods

🔬 Backtesting Engine

• Strategy Comparison: AI Sentiment Strategy vs Buy & Hold
• Performance Metrics: Returns, Alpha, Max Drawdown, Win Rate
• Historical Validation: Correlate sentiment with actual price movements

💹 Top Signal Drivers

• News Impact Analysis: Identify headlines that moved the market
• Sentiment Attribution: See which news drove positive/negative sentiment
• Source Tracking: Know where the signal originated

🎨 Premium UI/UX

• Dark Mode First: Professional trading terminal aesthetic
• Glassmorphism Design: Modern, sleek panels with backdrop blur
• Neon Accents: Color-coded signals (Green=Buy, Red=Sell, Amber=Hold)
• Smooth Animations: Micro-interactions for premium feel
• Fully Responsive: Works on desktop, tablet, and mobile

---

🚀 Quick Start

Prerequisites

• Python 3.8+
• pip (Python package manager)
• Modern web browser (Chrome, Firefox, Edge)

Installation

1. Clone or navigate to the project directory

   cd "C:\Users\Mamta Sharma\Downloads\Sentimental"

2. Install Python dependencies

   pip install -r requirements.txt

   This will install:

   • Flask & Flask-CORS (API server)
   • transformers & torch (ML models)
   • pandas, numpy (data processing)
   • yfinance (stock data)
   • feedparser (news aggregation)

3. Start the Flask API server

   python app.py

   Note: First startup will take 10-30 seconds to load ML models (FinBERT & RoBERTa). You'll see:

   Loading sentiment models...
   Models loaded successfully!
   🚀 SentimentSignal API Server
   📍 Server running at: http://localhost:5000
   

4. Open the frontend

   • Open index.html in your web browser
   • Or use a local server:

     python -m http.server 8000

     Then visit: http://localhost:8000

---

📖 Usage

Analyzing a Stock

1. Enter Stock Ticker

   • Type a NIFTY 50 ticker in the search bar (e.g., TCS.NS, RELIANCE.NS, INFY.NS)
   • Use the autocomplete suggestions for common stocks

2. Select Time Range

   • Choose 1M, 3M, or 6M analysis period
   • Default is 3 months (90 days)

3. Click "Analyze"

   • The system will:
     • Fetch historical news articles
     • Analyze sentiment using AI models
     • Generate trading signals
     • Run backtesting simulation
     • Display comprehensive results

4. Review Results

   • Sentiment Overview: Current sentiment score and BUY/HOLD/SELL signal
   • Chart: Sentiment trend overlayed on price movement
   • Top Drivers: News headlines that influenced sentiment
   • Backtest Results: Strategy performance vs benchmark
   • Strategy Summary: Plain-English explanation

Supported Stock Tickers

NIFTY 50 Examples:

• TCS.NS - Tata Consultancy Services
• INFY.NS - Infosys
• RELIANCE.NS - Reliance Industries
• HDFCBANK.NS - HDFC Bank
• ICICIBANK.NS - ICICI Bank
• TATASTEEL.NS - Tata Steel
• WIPRO.NS - Wipro
• SBIN.NS - State Bank of India
• And more...

Format: SYMBOL.NS (for NSE) or SYMBOL.BO (for BSE)

---

🏗️ Architecture

Backend (app.py)

• Framework: Flask REST API
• ML Models:
  • FinBERT (ProsusAI/finbert) - Financial sentiment
  • RoBERTa (cardiffnlp/twitter-roberta-base-sentiment-latest) - Social nuance
• Data Sources:
  • Google News RSS (news aggregation)
  • Yahoo Finance (stock prices via yfinance)

Frontend

• HTML: Semantic structure with accessibility
• CSS: Custom design system with CSS variables
• JavaScript: Vanilla JS with Chart.js for visualizations
• No frameworks: Pure web technologies for maximum performance

ML Model (havanagila.py)

• Sentiment Scoring: Dual-model ensemble approach
• News Fetching: Historical news with date filtering
• Backtesting: Real sentiment-based trading simulation
• Signal Generation: Threshold-based BUY/HOLD/SELL logic

---

📡 API Endpoints

GET /api/health

Health check endpoint

Response:

{
  "status": "healthy",
  "models_loaded": true,
  "timestamp": "2026-01-23T13:00:00"
}

POST /api/analyze

Analyze stock sentiment and generate signals

Request:

{
  "ticker": "TCS.NS",
  "days": 90
}

Response:

{
  "ticker": "TCS.NS",
  "days": 90,
  "sentiment": {
    "score": 0.234,
    "signal": "BUY",
    "confidence": 85,
    "article_count": 42
  },
  "backtest": {
    "strategy_return": 12.5,
    "buyhold_return": 8.3,
    "alpha": 4.2
  },
  "news": {
    "top_positive": [...],
    "top_negative": [...]
  },
  "chart_data": {...}
}

---

🎯 SDG Alignment

This project addresses:

• SDG 8 - Decent Work and Economic Growth: Empowers retail investors with professional-grade tools
• SDG 10 - Reduced Inequalities: Democratizes access to sentiment analysis previously limited to institutional traders

---

🛠️ Troubleshooting

Models Not Loading

• Issue: API shows "Loading Models..." for too long
• Solution: Ensure you have stable internet (models download on first run) and sufficient RAM (2GB+)

No News Found

• Issue: "No news found for ticker"
• Solution:
  • Verify ticker format (e.g., TCS.NS not TCS)
  • Try a more popular stock
  • Check internet connection

CORS Errors

• Issue: Frontend can't connect to API
• Solution:
  • Ensure Flask server is running on http://localhost:5000
  • Check that Flask-CORS is installed
  • Try opening frontend via http:// not file://

Chart Not Rendering

• Issue: Blank chart area
• Solution:
  • Check browser console for errors
  • Ensure Chart.js CDN is accessible
  • Try refreshing the page

---

📊 Performance Notes

• First Analysis: 20-40 seconds (news fetching + sentiment analysis)
• Subsequent Analyses: 15-30 seconds
• Model Loading: 10-30 seconds (one-time on server start)
• Recommended: Analyze 3-month periods for balance of data and speed

---

🔮 Future Enhancements

• Real-time sentiment updates
• Reddit/Twitter integration
• Portfolio tracking
• Email alerts for signals
• Mobile app
• Multi-stock comparison
• Custom backtesting parameters

---

📄 License

This project is for educational and research purposes. Not financial advice.

---

👥 Credits

ML Model: FinBERT (ProsusAI), RoBERTa (Cardiff NLP)
Data Sources: Google News, Yahoo Finance
Framework: Flask, Chart.js
Design Inspiration: TradingView, Bloomberg Terminal, Zerodha Kite

---

📞 Support

For issues or questions:

1. Check the Troubleshooting section
2. Verify all dependencies are installed
3. Ensure API server is running
4. Check browser console for errors

---

Built with ❤️ for retail investors