Stock Prediction System — News + Market Analysis (FastAPI + Streamlit)

An AI-powered stock prediction system that analyzes news sentiment, market momentum, and volatility to generate a Buy / Hold / Sell recommendation for any company.

This system uses:

• FastAPI (backend)
• Streamlit (frontend)
• FinBERT (sentiment analysis)
• yfinance (market data)
• Newspaper3k + RSS (news ingestion)

---

Features

✔ News Ingestion

• Scrapes recent news articles from Google News RSS & APIs
• Extracts full article text using Newspaper3k

✔ Sentiment Analysis (FinBERT)

• Predicts positive / neutral / negative sentiment
• Computes average sentiment of all news

✔ Market Data Processing

• Fetches price history using yfinance
• Computes:
  • 5-day momentum
  • 20-day momentum
  • Volatility
  • Current price

✔ Recommendation Engine

Combines:

0.6 * sentiment + 0.3 * momentum - 0.1 * volatility

Outputs a final decision:

• BUY
• HOLD
• SELL

✔ Frontend Dashboard (Streamlit)

• Enter company + ticker
• Displays decision, features, signals, sentiment
• Clean, interactive interface

---

Project Architecture

User → Streamlit UI → FastAPI Backend →  
   → News Ingestion  
   → Sentiment Analysis (FinBERT)  
   → Market Signals (yfinance)  
   → Recommendation Engine  
→ Response → Streamlit Dashboard

---

Folder Structure

Stock-prediction/
│
├── backend/
│   ├── main.py
│   ├── src/
│   │   ├── data_ingestion/
│   │   ├── sentiment_alz/
│   │   ├── market_data/
│   │   └── recommendation/
│   └── venv/
│
├── frontend/
│   └── streamlit_app.py
│
└── README.md

---

Backend Setup (FastAPI)

Create virtual environment:

python3 -m venv venv
source venv/bin/activate

Install dependencies:

pip install fastapi uvicorn transformers torch yfinance feedparser newspaper3k

Run FastAPI:

uvicorn main:app --reload

Backend runs on:

http://127.0.0.1:8000

---

Frontend Setup (Streamlit)

Install Streamlit:

pip install streamlit plotly pandas

Run the Streamlit UI:

streamlit run streamlit_app.py

Opens at:

http://localhost:8501

---

API Example

POST /predict

Request:

{
  "company": "Apple",
  "ticker": "AAPL"
}

Response:

{
  "company": "Apple",
  "ticker": "AAPL",
  "decision": "HOLD",
  "features": {
    "score": 0.0116,
    "avg_sentiment": 0.035,
    "momentum_5d": -0.028,
    "volatility": 0.009
  },
  "market_signals": {
    "price": 267.44,
    "momentum_5d": -0.028,
    "momentum_20d": 0.018,
    "volatility": 0.009
  },
  "articles_used": 46
}

---

Future Improvements

• ML prediction using Scikit-Learn or XGBoost
• Backtesting engine (historical accuracy)
• Article preview section in Streamlit
• Stock price charts
• Redis caching
• Cloud deployment (AWS / Render / Railway)

---

License

This project is licensed under the MIT License.

---

Contributing

Pull requests are welcome! Improve ingestion, sentiment accuracy, or UI dashboards.