StockLens - Analysis and News Sentiment for Smart Investing

This research presents an intra-day stock prediction model based on financial news analysis and sentiment aiming for intelligent investing. Nowadays, the intra-day market is constantly fluctuating, and there is a difficulty in recognizing the market trend. Addressing this, our research aims to develop a predictive framework that extracts actionable insights from news data. We process financial news articles, headlines, and video data from different platforms through techniques such as YOLO for the text detection and Tesseract OCR for text extraction. We then apply sentiment analysis on it using FinBERT and add a custom credibility assessment to further refine predictions for better results. The innovation in our approach is to not only consider traditional textual data but also include video data, and besides this, we take into consideration a few other very critical factors in the model: publisher credibility, market analysis-volatility index, sector-wise analysis, and many more. The developed stock prediction model offers a strategic advantage in navigating the complexities of the stock market, enabling investors to capitalize on emerging opportunities and mitigate risks effectively.