TEAM 7 - Predicting Gold Price Movement in India Using Financial Sentiment Analysis

Indian Institute of Science Summer 2025 DA225o Submission7 Authors

Keywords: Gold Price Movement Forecasting, Financial Markets, India, Language Models, Fine-tuning, Custom LLM, Encoder-based Models

Abstract:

This project presents a language model-based automation workflow for forecasting gold price movements in India, leveraging sentiment signals extracted from financial news, stock market data, and technical indicators such as moving averages and RSI. While sentiment-driven models are widely used in equity markets, their application in gold price movement prediction is underexplored.

The proposed methodology orchestrates a system capable of performing data collection, web scraping, preprocessing, sentiment analysis, LLM encoder-based sentiment extraction, and gold price movement assertion. The system includes a custom fine-tuned language model trained locally on domain-specific data, integrated into an orchestration framework for real-time inference. The performance of the model is also benchmarked against standard baseline models.

Submission Number: 7