🌿 ESG Pulse - Sentiment & Stock Impact Tracker

A quantitative analysis system that tracks ESG-related news sentiment and analyzes its impact on short-term stock returns.

ESG Pulse is an automated agent that:

• Tracks ESG news for user-defined stock portfolios
• Analyzes sentiment using FinBERT (financial sentiment analysis)
• Calculates abnormal returns using market model regression
• Quantifies impact through multi-factor regression analysis
• Provides insights on ESG sentiment's effect on stock performance

Portfolio Input → News Ingestion → Sentiment Analysis → Market Data → Regression → Insights

• Frontend: Streamlit dashboard for portfolio input and results visualization
• Backend: Python scripts for data processing and analysis
• NLP: FinBERT for ESG sentiment scoring
• Analytics: Multi-factor regression (sentiment + momentum + market factors)
• Automation: n8n workflows for end-to-end orchestration

Our analysis of 118 observations shows:

• Momentum Impact: Highly significant (p = 0.001) - strong positive effect on returns
• ESG Sentiment: Not statistically significant in current sample
• Model Performance: R-squared = 9.2%, F-statistic = 5.827 (p = 0.00389)

• Python 3.8+
• Git

# Clone the repository
git clone https://github.com/yourusername/esg-sentiment-agent.git
cd esg-sentiment-agent

# Install dependencies
pip install -r requirements.txt

# Set up environment variables
cp .env.example .env
# Edit .env with your API keys

• GNews API: For ESG news fetching
• Yahoo Finance: For stock price data (free)

1. Run the Streamlit Dashboard:

   streamlit run dashboards/app.py

2. Input Your Portfolio:

   • Enter stock tickers (e.g., AAPL, MSFT, TSLA)
   • Click "Save Portfolio"

3. Run Analysis:

   # Fetch ESG news
   python src/ingestion.py
   
   # Score sentiment
   python models/sentiment_model.py
   
   # Generate market features
   python src/ingestion.py  # (switch to market features mode)
   
   # Run regression analysis
   python models/regression_model.py

esg-sentiment-agent/
├── dashboards/          # Streamlit dashboard
├── data/               # CSV files (portfolio, news, features)
├── models/             # ML models (sentiment, regression)
├── src/                # Core processing scripts
├── n8n/               # Automation workflows
├── notebooks/         # Jupyter notebooks for analysis
└── requirements.txt   # Python dependencies

GNEWS_API_KEY=your_gnews_api_key

• Edit data/user_portfolio.csv or use the Streamlit interface
• Supported tickers: Any stock available on Yahoo Finance

• ✅ Portfolio management
• ✅ ESG news ingestion
• ✅ Sentiment analysis (FinBERT)
• ✅ Market data collection
• ✅ Abnormal return calculation
• ✅ Multi-factor regression
• ✅ Results visualization

• 🔄 Real-time alerts
• 📊 Interactive dashboards
• 🤖 Automated scheduling
• 📱 Mobile notifications
• 🌐 Web API endpoints

• Estimation Window: T-80 to T-6 days
• Event Window: T-3 to T+3 days
• Benchmark: S&P 500 returns
• Model: OLS regression with CAPM-style market model

• Model: FinBERT (ProsusAI/finbert)
• Categories: Positive, Neutral, Negative
• Features: Sentiment score, sentiment label

abnormal_return ~ sentiment_score + momentum + vix + sector_dummies

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

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

• FinBERT: For financial sentiment analysis
• Yahoo Finance: For market data
• GNews: For ESG news aggregation
• Streamlit: For the interactive dashboard

• Project Link: https://github.com/yourusername/esg-sentiment-agent
• Issues: https://github.com/yourusername/esg-sentiment-agent/issues

Built with ❤️ for ESG-conscious investors

import os
import pandas as pd
import streamlit as st

if os.path.exists('data/market_features.csv'):
    df = pd.read_csv('data/market_features.csv')
    st.header('Abnormal Returns by Sentiment Category')
    df['sentiment_category'] = pd.cut(df['sentiment_score'], bins=[-float('inf'), 0.3, 0.7, float('inf')], labels=['Negative', 'Neutral', 'Positive'])
    st.write(df.groupby('sentiment_category')['abnormal_return'].agg(['mean', 'std', 'count']))

    st.header('Distribution of Abnormal Returns')
    st.hist = st.pyplot()
    df['abnormal_return'].hist(bins=30)
    st.pyplot(st.hist)