Applying the COPRAS Method for Stock Price Prediction: A Comparative Analysis of Selected Companies

This paper explores the application in advanced algorithms and methodologies Stock price forecasting is extremely important as it has a significant financial impact, influencing investment decisions and market strategies accurate market forecasting. Effective predictive models offer benefits beyond wealth creation, including early identification of failing investments to prevent market disruptions. The study acknowledges the inherent challenges in financial time series analysis—non-stationarity, nonlinearity, and high volatility—which necessitate continued development of more sophisticated analytical approaches. The research examines both internal factors (company profits, liabilities, competition) and external factors (oil prices, exchange rates, political stability) that influence stock prices. Traditional stock trend prediction algorithms rely on historical time series data, while newer approaches incorporate AI and Machine learning methods, including artificial neural networks (ANNs), are important in advanced predictive modeling and have shown better performance than traditional regression methods, though significant challenges remain in optimizing input variables and network architecture. The paper particularly highlights the promising performance The study uses Machine learning and deep learning approaches, particularly the use of Generative Adversarial Networks (GANs), are used to predict final prices. which outperform other methods. In addition, the research introduces a novel approach stock price prediction model, built on the basis of Dragonfly Algorithm (DFA)-optimized Deep Belief Network (DBN), which leverages market Sentiment analysis from social media data. For evaluation, this study uses the Complex Proportionality Assessment (COPRAS) method along with other multi-criteria decision-making methods (MCDM) methods, such as the Analytical Hierarchy Process (AHP), are used in decision-making processes analysis. These methods help assess and rank different prediction models based on multiple criteria, providing a robust framework for addressing complex financial forecasting challenges. The integrated approach combining statistical methods with machine learning techniques based on technical analysis offers valuable insights for investors, particularly those focused on short-term investments. This research contributes to ongoing efforts to develop more accurate forecasting models, though acknowledges that no universal solution has yet been established.