Financial sequence prediction based on swarm intelligence algorithms and internet of things

This paper aims to predict the financial time series. Swarm intelligence algorithms usually use metadata to ensure objectivity without the statistical assumptions of the data. This paper proposed a prediction algorithm integrating multiple support vector regression (SVR) models. The algorithm selects different datasets to train these SVR models. This algorithm also adopts reasonable weights to combine the forecasting results of multiple models to reduce the overall prediction error. The weight of each model is dynamically adjusted according to its recent prediction accuracy. Therefore, this algorithm is adaptive and can deal with nonstationary problems. Five international authoritative stock indexes are used to compare the hybrid SVR model with a single SVR model for performance validation from the perspectives of normalized mean squared error, weighted directional symmetry, and root mean squared error. The results demonstrate that the hybrid SVR model has significantly improved the prediction accuracy and generalization ability of the prediction algorithm compared with a single SVR model. It reveals that selecting the appropriate input vector can achieve an excellent prediction effect.