Abstract

Theoretical background: Undoubtedly, forecasting share prices is a non-trivial task. However, thanks to the rapid development of science, in recent years, more and more research work has been devoted to the prediction of stock market prices, thanks to which it is becoming more accessible and easier to forecast, among other things, share prices using a variety of methods, including machine learning and deep learning in particular. In the age of big data, deep learning used to predict prices and stock trends has become even more popular than before. At the same time, attempts are made to develop solutions similar to the method of analyzing stock market data by humans, i.e., data analysis presented in the form of charts. Therefore, it is worth taking a closer look at the possibilities of discovering patterns in stock charts by convolutional neural networks.Purpose of the article: The adopted research hypothesis says that convolutional neural networks can potentially analyze stock data. The study aims to use a machine learning method, a convolutional neural network, to analyze stock market charts. For this purpose, a convolutional neural network will be built and programmed, which will be able to recognize relevant information based on previously prepared images constituting simplified stock charts.Research methods: The study used the method of deep machine learning, including convolutional neural networks, which are characterized, among others, by the ability to process and analyze graphical data. In particular, the TensorFlow library was used with Keras functions, which have implemented algorithms of convolutional neural networks.Main findings: The conducted research experiment showed high efficiency (close to 100%) of the adopted algorithm and the proposed convolutional neural network structure, thanks to which it can be assumed that convolutional neural networks contain a high potential for the analysis of stock data represented by charts showing the behavior of share prices or other financial instruments.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call