Hybrid Approach for Financial Time Series Prediction: Integrating Extended Kalman Filter with Particle Swarm Optimization

Time series in finance is important for predicting appropriately for decision-making in trading, investment, and risk management. This paper, suggests a new hybrid model that integrates the Extended Kalman Filter with Particle Swarm Optimization to increase prediction accuracy and robustness. EKF captures the nonlinear and dynamic characteristics of financial data, and it will therefore be able to catch the time-varying market patterns. PSO will ensure global convergence and performance improvement by optimizing the model parameters. Experiments conducted on real stock price data and market indices demonstrate that the hybrid EKFPSO model significantly outperforms standalone EKF. Hybridization is a key improvement like MSE (0.1071 vs 0.166), RMSE (0.3273 vs 0.4074), and MAPE (2.19% vs 3.38%), which reduces prediction error and trend capturing. It demonstrates reducing prediction errors and correctly capturing trends. Moreover, the framework demonstrates its efficacy through the lens of disastrous conditions in financial markets, which provides a vital context for the analyses and the implications therein. The study also shows how hybrid methodologies can provide the financial analyst with a highly robust and adaptable tool, capable of competing with specialist methodologies in their domain.