DEHypGpOls: a genetic programming with evolutionary hyperparameter optimization and its application for stock market trend prediction

Stock markets are a popular kind of financial because the possibility bringing high revenues to their investors. To reduce risk factors for investors, intelligent and automated stock market forecast tools developed by using computational intelligence techniques. This study presents hyperparameter optimal genetic programming-based model generation algorithm a-day-ahead prediction index trends. obtain an from modeling dataset, differential evolution (DE) is employed optimize hyperparameters programming orthogonal least square (GpOls) algorithm. Thus, GpOls agents within search space enables adaptation dataset. evolutionary optimization technique can enhance data-driven performance allow autotuning user-defined parameters. In current study, proposed DE-based hyper-GpOls (DEHypGpOls) used generate forecaster models trend Istanbul Exchange 100 (ISE100) Borsa (BIST100) indexes. this experimental daily data ISE100 BIST100 seven other international models. Experimental studies on 4 different time slots datasets demonstrated that DEHypGpOls could provide 57.87% average accuracy in buy–sell recommendations. The investment simulations with these showed investments indexes according buy or sell signals 4.8% more income compared long-term strategy.