Predictive Analysis of Stock Market Trends Using Support Vector Machines and Neural Networks

Stock market forecast is tough because is a nonlinear and highly fluctuating system. Procedures such as ARIMA and moving averages are challenging in mitigating derived complex market fluctuations. To solve this problem, to introduce a novel model based on SVM and NN. The SVM performs best-turning movements of stock while the NN is apt at its ability to model nonlinear relations in large sample sizes. The model performance was evaluated on several stock datasets and was proven to outperform standalone models by a large margin. The hybrid model has 0.0251 MAE and RMSE of 0.0329, outperforming SVM MAE 00356, and RMSE 0.0498. Hence, the SH model earned an MAE of 0.0347 and RMSE of 0.0455, the DBT model MAE of 0.0402 and RMSE of 0.0564, while the NN model has an MAE of 0.0356 and RMSE of 0.0498. The average daily prediction accuracy was 84.1%, weekly 89.3%, and monthly 93.6%. Overall, these results demonstrate the hybrid model's stock market trend prediction potential it seems to be a potential technique for such predictions.