Estimations of opening and closing stock prices through machine learning methods and deep learning algorithms

Forecasting the stock markets' prices in emerging economies such as Turkey is quite important, however, the existence of all kinds of speculative movements in the markets causes ups and downs to be very high in these stock markets. This situation makes it difficult for investors to predict stock price movements and thus cause losses since they cannot evaluate their investments correctly. For his reason, investors feel the need to use various practices and methods to predict future stock prices. In order to gain these forecasts, machine learning methods and deep learning algorithms have been recently used frequently in the field of finance as well as other fields. In this study, data between January 2010 and January 2019 were used as the data set for the second session opening and closing prices of the stocks of the companies traded on the Borsa İstanbul National-100 Index (BIST 100). In the light of machine learning methods and deep learning algorithms, it is aimed to estimate the opening and closing prices of these stocks. In this context, in order to create a prediction model, machine learning methods such as Multilayer Perceptrons (MLP) and Support Vector Machines (SVMs) as well as deep learning algorithm, Long Short Term Memory (LSTM) method have been used. As a result of the study, it has been observed that MLP and LSTM networks make more consistent estimates than SVMs. This result has been supported by two-tailed t test analysis, which shows whether there is a difference between the groups in the analyzes.