ARCHIVES

Stock market prediction is always challenging due to nonlinearity, volatility, and uncertainty in financial time series data. This paper proposes a novel hybrid soft computing model, termed NEW-HYBRID, which integrates fuzzy logic, artificial neural networks (ANN), genetic algorithms (GA), and support vector regression (SVR) to enhance forecasting accuracy and robustness. The architecture employs fuzzy systems for handling vagueness in market sentiments, ANN for capturing nonlinear patterns, GA for optimal parameter tuning and feature selection, and SVR for precise regression on decomposed signals via empirical mode decomposition (EMD). Data preprocessing includes technical indicators (e.g., RSI, MACD), sentiment analysis from news, and noise reduction. The model predicts daily closing prices and directional trends on benchmarks like S&P 500 and NASDAQ datasets from 2020-2025. Evaluations show superior performance over baselines: RMSE reduced by 22%, MAPE below 1.2%, and accuracy exceeding 92% in trend classification. NEW-HYBRID demonstrates adaptability to real-time data, mitigating overfitting through ensemble mechanisms. This approach advances soft computing paradigms for financial forecasting, offering interpretable and scalable solutions for traders and investors