Forecast of Stock Market using Machine Learning Strategies

Abstract

Accurately predicting the stock market is one of the things that investors are most interested in since it may help them make money in the economy. Given that such markets are significantly influenced by volatility and news, it is difficult to predict stock prices, which are solely dependent on market timing. Owing to this difficulty and volatility, it is vital to evaluate stock forecasting using historical data as well as external variables such as investor behaviour, social media, and financial news. Thus, this study recommends using regression and machine learning algorithms to estimate the price of equities on upcoming days based on investor sentiment. The experiment is conducted on Yahoo Finances and combines Twitter repository data on investor sentiment. In the subsequent step, the concept of sentiment analysis is applied to the monitoring of Twitter user tweets. The tweets are then sorted into positive and negative groups based on the sentiment score. In addition, machine learning algorithms are used to forecast Yahoo Finance stock values. To solve this issue, we propose reducing the complexity of time sequence models by employing regression approaches that integrate a hybridized concept of sentiment analysis and machine learning algorithms, which may result in higher accuracy. The testing results validate the best linear regression prediction accuracy and demonstrate an overall system performance enhancement.