MarketPulse AI

AI system that detects mismatches between public sentiment and stock performance to identify undervalued and overhyped opportunities.

Comment

🚀 Inspiration

Financial markets are often assumed to be efficient, but in reality, public sentiment and actual market movements don’t always align.

I wanted to explore: 👉 Can AI detect when the market is “wrong”?

This curiosity led to building MarketPulse AI — a system that identifies mispriced opportunities using sentiment vs price divergence.

💡 What it does

MarketPulse AI analyzes:

  • 📊 Stock price movements (NASDAQ data)
  • 📰 News sentiment (using VADER Sentiment Analyzer)
  • 📈 Trend signals (simulated Google Trends behavior)

Then it:

  • Compares sentiment vs actual price movement
  • Detects divergence (mispricing signals)
  • Ranks top opportunities
  • Visualizes insights with charts

👉 Output: Top 5 mispriced stocks

🛠️ How I built it

I built a complete data pipeline using:

  • Python
  • Pandas for data processing
  • NumPy for calculations
  • Matplotlib for charts
  • VADER Sentiment Analyzer for sentiment scoring

Workflow:

  1. Clean & process NASDAQ dataset
  2. Generate mock financial news
  3. Apply sentiment analysis
  4. Compute price vs sentiment divergence
  5. Rank top anomalies
  6. Visualize results
  7. Deploy using Streamlit

⚡ Challenges I ran into

  • Handling messy CSV format (tab-separated issues)
  • Aligning time-series sentiment with price data
  • Creating realistic synthetic sentiment signals
  • Avoiding misleading correlations
  • Deploying smoothly on limited resources

🏆 Accomplishments that I'm proud of

  • Built a fully working AI pipeline end-to-end
  • Identified real divergence patterns in stocks like TSLA & NVDA
  • Created clean visual dashboards
  • Successfully deployed a working web app
  • Made the system run without any API dependencies

📚 What I learned

  • Markets are not perfectly efficient
  • Sentiment ≠ price movement (huge insight 🔥)
  • Importance of clean data pipelines
  • How lightweight NLP (VADER) can still be powerful
  • End-to-end ML systems > just models

🔮 What's next for MarketPulse AI

  • 🔗 Integrate real-time APIs (NewsAPI, Yahoo Finance)
  • 🤖 Use advanced models like FinBERT for better sentiment
  • 📡 Build live trading signal alerts
  • 📊 Add interactive dashboards
  • 🧠 Improve prediction accuracy with ML models

👉 MarketPulse AI shows that sometimes, the biggest opportunities lie where perception and reality diverge.

Built With

Share this project:

Updates