Presenting an Innovative Methodology for Predicting Stock Market Trends: A Case Study on <scp>KOSPI</scp> Price Estimation

The stock market is a platform where individuals buy and sell shares of publicly listed corporations, serving as a key indicator of a nation's economic health and business climate. Stock price fluctuations are driven by the balance between supply and demand. While investing in the stock market is unpredictable, its long‐term returns are significant. The main aim of investment in financial markets is to maximize returns, but different factors make the outcome volatile. These factors make predictions about future market outcomes difficult. The purpose of this paper is a hybrid model development for the stock price and market trend precise prediction by using Support Vector Regression (SVR), the Mayfly Algorithm (MA), and the Empirical Mode Decomposition (EMD) accordingly. This research utilized historical data from the KOSPI, gathered from 2 January 2015 to 30 June 2023. In this study, simplifying the dataset is done through EMD, and two optimization techniques—Gravitational Search (GSA) and the MA—enhance the model accuracy. EMD‐MA‐SVR has an efficient performance with the R‐squared value of 0.9891 and directional accuracy of 0.6412 and is hence more precise in stock price prediction compared to the rest of the models. This research compares the hybrid benchmark models with the proposed EMD‐MA‐SVR model concerning performance in capturing intricate patterns for the precision of stock price predictions. To comprehensively validate the robustness and generalizability of the proposed EMD‐MA‐SVR model in multiple splits, a 4‐fold cross‐validation approach has been adopted. Consequently, this model can be considered an efficient tool for investors in terms of accurate stock price prediction and decision‐making. © 2026 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.