Artificial Intelligent based day-ahead stock market profit forecasting

Abstract

This article develops an artificial intelligent based forecasting model for the day ahead stock market profit. The suggested model is constructed based on the long short term memory (LSTM) to let it learn the long term reliance of the stock market samples. As a powerful deep learning approach, LSTM makes it possible to learn the very complex data prediction and classification problems. Due to very complex nature of the stock market profit, an evolutionary algorithm based of shuffled frog leaping algorithm (SFLA) is developed to provide a competitive random search. The proposed SFLA can mimic the frog life style for making a cooperative search in a global space. In order to improve the training process, a correction approach based on the mutation and crossover is developed which can enhance the SFLA performance. Real stock market profit dataset are used to assess the high performance of the proposed model.