Sentiment Analysis on Financial News (Assignment 2)

Overview

This project implements sentiment analysis models on financial news headlines and articles to estimate their potential impact on stock prices. It was built as part of a course assignment, combining data collection, preprocessing, machine learning, and trading simulation.

We:

• Collect historical stock price data and S&P 500 index values
• Scrape or process financial news data
• Compute volatility, abnormal returns, and impact scores
• Vectorize news text (bag-of-words, TF-IDF, curated sentiment words)
• Train a multi-layer perceptron (MLP) classifier in PyTorch
• Simulate trading strategies driven by predicted impact scores
• Evaluate results and reflect on improvements

---

Repository Structure

SentimentAnalysis/

│

├── Assignment_2/

│ ├── phase_1/ # Data collection & preprocessing

│ │ ├── 1_collect_data.py

│ │ ├── 2_calculate_volatility.py

│ │ ├── 3_estimate_impact_scores.py

│ │ ├── 4_identify_and_vectorize.py

│ │ └── testcollect.ipynb

│ │

│ ├── phase_2/ # Modeling

│ │ ├── 1_process.py

│ │ ├── 2_model.py

│ │ ├── 2_training.py

│ │ └── 4_eval.py

│ │

│ ├── phase_3/ # Trading simulation

│ │ ├── 1_trading_rules.py

│ │ └── 2_trading_sim_eval.py

│ │

│ ├── datasets/ # Generated CSV outputs

│ ├── dataset_schema.py

│ ├── test_dataset_schema.py

│ ├── utils.py

│ ├── webscraping.py

│ └── webscraping_headless_parsing.py

│

├── news_datasets/ # Provided input datasets

│ ├── analyst_ratings.csv

│ └── headlines.csv

│

└── README.md

---

Requirements

• Python 3.9+
• Libraries:
  • pandas
  • numpy
  • yfinance
  • scikit-learn
  • torch
  • beautifulsoup4, requests (if scraping full articles)

Install with:

pip install -r requirements.txt
---