Predict stock prices using the Generative Adversarial Networks

Abstract

Predicting the price of a stock is very difficult. Due to very volatile prices. Many traders incorrectly predict stock prices, forex or trading commodities. It takes an analysis of each price movement. The purpose of the analysis is to predict price movements. One of them is the use of indicators that seek to help predict prices. Currently the development of Artificial Intelligence (AI) has grown very rapidly. Machine learning which is part of AI is also used to predict prices. Stocks are data that are related to time. Just like the weather. If the stock is analyzed then the suitable method is the time series method. The method used is Deep Learning, namely Recurrent Neural Network (RNN). A recurrent Neural Network is the same as Artificial Neural Network (ANN). ANN performs the processing of sequential data. RNN does not discard past problem data information, but will also enter past information as input. This is what distinguishes RNN and ANN. In the Recurrent Neural Network, there is a Long Short Term Memory Algorithm, Gated Recurrent Unit (GRU). One of the algorithms that can be used to predict stock prices is the Generative Adversarial Network. This algorithm was modified before being used. In the GAN algorithm, there are Generators and Discriminators. Because stock is a process that is carried out in the presence of time or time series, the Generative is modified with Long Short Term Memory and Discriminator uses Long Short Term Memory