NiftyLLaMA: An open information extraction system integrated with sentiment analysis for Nifty-50 companies using LLaMa-2

The paper examines the use of large language models (LLaMa-2) in analyzing one of the fastest emerging stock markets, Indian financial exchange, National Stock Exchange. The main contribution of the paper is the development of a model specifically for the Indian market, targeting the NIFTY50 blue chip companies. This involved creating a unique dataset that includes stock prices, Wikipedia and news articles. Further the paper conducts sentiment analysis using FinBERT and fine-tunes the LLaMa-2 model with advanced techniques like Parameter Efficient Fine-Tuning (PEFT) and Low-Rank Adaptation (LoRA). The methodology encompassed data collection, integration, and model optimization. The findings demonstrate the effectiveness of the model in summarizing data and addressing new questions, showing its potential for financial investment strategies. The paper concludes by suggesting future enhancements and positions the NiftyLLaMa model as a notable step in applying LLMs to financial analysis in emerging markets, indicating a direction for future data-driven decision-making in various sectors.