Shahvaroughi Farahani Milad, Razavi Hajiagha Seyed Hossein
Department of Finance, Faculty of Management and Finance, Khatam University, Tehran, Iran.
Department of Management, Faculty of Management and Finance, Khatam University, Tehran, Iran.
Soft comput. 2021;25(13):8483-8513. doi: 10.1007/s00500-021-05775-5. Epub 2021 Apr 25.
Today, stock market has important function and it can be a place as a measure of economic position. People can earn a lot of money and return by investing their money in the stock exchange market. But it is not easy because many factors should be considered. So, there are many ways to predict the movement of share price. The main goal of this article is to predict stock price indices using artificial neural network (ANN) and train it with some new metaheuristic algorithms such as social spider optimization (SSO) and bat algorithm (BA). We used some technical indicators as input variables. Then, we used genetic algorithms (GA) as a heuristic algorithm for feature selection and choosing the best and most related indicators. We used some loss functions such as mean absolute error (MAE) as error evaluation criteria. On the other hand, we used some time series models forecasting like ARMA and ARIMA for prediction of stock price. Finally, we compared the results with each other means ANN-Metaheuristic algorithms and time series models. The statistical population of research have five most important and international indices which were S&P500, DAX, FTSE100, Nasdaq and DJI.
如今，股票市场具有重要作用，它可作为衡量经济状况的一个场所。人们通过在证券交易所市场投资资金能够赚取大量金钱和回报。但这并非易事，因为需要考虑诸多因素。所以，有许多方法来预测股价走势。本文的主要目标是使用人工神经网络（ANN）预测股票价格指数，并使用一些新的元启发式算法（如社会蜘蛛优化算法（SSO）和蝙蝠算法（BA））对其进行训练。我们将一些技术指标用作输入变量。然后，我们使用遗传算法（GA）作为启发式算法进行特征选择并挑选出最佳且最相关的指标。我们使用一些损失函数（如平均绝对误差（MAE））作为误差评估标准。另一方面，我们使用一些时间序列模型（如ARMA和ARIMA）来预测股票价格。最后，我们将结果相互比较，即人工神经网络 - 元启发式算法和时间序列模型的结果。研究的统计总体包括五个最重要的国际指数，即标准普尔500指数、德国DAX指数、英国富时100指数、纳斯达克指数和道琼斯工业平均指数。
