Stock Market Behaviour Prediction using Long Short-Term Memory Network and Gated Recurrent Unit

Stock Market behaviour prediction has been an area of interest within research for some time, and Recurrent Neural Networks have shown great promise in the way of solving time series based problems. In this study we analyse two Recurrent Neural Network based models. One makes use of Long Short Memory Networks and the other makes use of a variation of the Long Short Memory Networks called the Gated Recurrent Unit for the purpose of Stock Market behaviour prediction. A comparison is made between the two models based on training on the same stock market dataset. Results obtained show a greater accuracy for the Long Short Memory Network model in comparison to the Gated Recurrent Unit based model.