Abstract:

The complexity and diversity of the stock market have long intrigued researchers, prompting efforts to develop methods for predicting its future movements. However, the volatility observed across global stock markets makes this task particularly challenging. While statistical methods and modeling techniques are useful, they are often insufficient to tackle the wide range of issues encountered when predicting stock market trends. Traditional approaches have struggled to provide solutions to the intricate problems within the market. Machine learning and artificial intelligence (AI) tools, on the other hand, offer an effective means of handling the complexities of Big Data. This paper proposes the use of six distinct algorithms—Generalized Linear Model, Deep Learning, Decision Tree, Random Forest, Gradient Boosted Trees, and Support Vector Machine—to identify the model that best predicts market movements. These algorithms were tested on BSE index data spanning from April 2015 to March 31, 2020, with the model exhibiting the least relative error being selected. Among the tested models, Gradient Boosted Trees emerged as the most efficient, demonstrating the lowest relative error and standard deviation. Consequently, Gradient Boosted Trees is used to forecast future market trends.

Description:

.