Back to Projects
Stock Price Predictor
A machine learning model using LSTM networks to predict stock price movements based on historical data and sentiment analysis from financial news.
View on GitHub📈
Overview
This project combines time-series analysis with NLP-based sentiment scoring to predict short-term stock movements. The model achieved ~62% directional accuracy on held-out test data for large-cap US equities.
Methodology
- LSTM neural network trained on 5 years of historical price data
- Sentiment analysis on financial news using FinBERT
- Feature engineering: RSI, MACD, Bollinger Bands, volume indicators
- Ensemble model combining LSTM predictions with sentiment scores
- Backtested on S&P 500 constituents (2018–2023)