Predicting Stock Price with LSTM Networks

Stock market strategies are complicated and rely on a vast quantity of data. As a result, stock price forecasting has always been challenging for many experts and investors. As a result of significant research, many machine learning algorithms have been built without being explicitly written to handle complicated computational issues and improve prediction capabilities. This study, looks into a mechanism for forecasting stock price swings. To estimate, The estimation was done with a recurrent neural network (RNN) based on long short-term memory (LSTM) if the S & P500 will rise (up) or fall (down) over the following trading month, based on the volatility of the variable. The three pieces of time sequence data are return, trading volume, and trading volume. The accuracy and the area under the (ROC) curves are used to alter hyper-parameters and assess prediction performance (AUC). It finds out that LSTM models perform comparably to baseline linear classification models.