Machine learning for asset wealth management

Abstract

The fields of Machine Learning and Artificial Intelligence have made significant advances in recent decades and have been increasingly integrated into people’s daily lives. These days, it is common to see various Machine Learning models being incorporated in order to help solve real life challenges faced by not only individuals but also big organizations.

This paper proposes a stock trading web application that helps to empower the user with knowledge about the stock market and tries to reduce the fear of a common person from entering the market by giving recommendations about various stocks (buy or sell). The latest techniques for financial time-series prediction have been studied and implemented. Web scraping techniques have been implemented to gather and display real-time stock data while three machine learning models: Linear Regression, Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA) have been trained and evaluated for this application. A sentiment analysis model has also been incorporated in the application to consider external factors such as news-based factors that are not directly covered by the Machine Learning models. After testing, it is found out that adding the market sentiment analysis of a particular company leads to more accurate recommendations about a stock. The frontend of the application proposed by this project is created using HTML, CSS, JavaScript, and PHP. The web scraping techniques along with various machine learning and sentiment analysis models have been implemented using Python. A MySQL database is used to store the details of the user including their credentials and portfolio details while Flask is used to connect the frontend to the backend.