Stock Market companion: real-time data, hybrid analysis, and enhanced user interaction

Abstract

Stock price prediction has long been a major focus in financial sciences due to the need to anticipate movements in an unpredictable market. Traditional models often rely only on numerical data, overlooking external factors like news and social media. This project introduces a hybrid stock price prediction model that combines historical stock data with sentiment analysis of financial texts for improved accuracy. While initially demonstrated on eight major tech companies, the system is flexible enough to predict the stock prices of any publicly traded company, including popular names like Apple, Nvidia, Tesla, and Microsoft. Numerical data is processed using machine learning algorithms in Scikit-learn, and natural language processing (NLP) techniques assign sentiment scores to textual data. For real-time data handling and visualization, InfluxDB and Grafana are used alongside a detailed Power BI dashboard, which presents key insights, sentiment trends, and predictive analytics in an intuitive format. A Streamlit application allows users to interact with the model easily, and a Flask API backend supports scalable integration and future development. The project is open-source, available on GitHub with complete documentation, ensuring accessibility for collaboration and further innovation. By combining structured numerical analysis with unstructured sentiment data, the project offers a comprehensive and scalable approach to stock market forecasting.