Neural Networks for Time Series Forecasting to Predict The Return of Stock Index

Generating accurate Predictions of Financial Market movements has always been a key goal for all stakeholders of an economy. The various algorithms used for time series forecasting. It can be categorized into linear model (ARIMA, SARIMA) and non-linear models (neural networks). A linear regression model is not suitable for time series because these model are neglected a problems on time. It adds additional information that makes time series problems even more difficult to handle .The traditional model used to predict time series like autoregressive integrated moving average (ARIMA) and seasonal autoregressive integrated moving average model (SARIMA) but this model is not sufficient for time series .that’s why Neural Networks (NN) is important essential for time series forecasting. Only certain type of Neural Networks Suitable for time series model .Such as Convolutional neural network (CNN) and Long Short-Term Memory (LSTM) neural networks. These machine learning models that enable better predictions by improving the accuracy of time series. In this paper , different time series models are taken and compare them with neural networks for predict the returns for the S&P500 (Standard & Poor’s 500 ) index .This paper also concludes that neural networks is suited for time series forecasting . The experiments were executed in jupyter notebook 6.0.3 .the codes and simulations were implemented using python 3.6.5 with a dependency of tensorflow 1.15.0 and keras 2.3.1