Stock Price Prediction Based on ARIMA-RNN Combined Model

Abstract

In this paper, we proposed a new hybrid ARIMA–RNN model to forecast stock price, the model based on moving average filter. This model can not only overcome the volatility problem of a single model, but also avoid the overfitting problem of neural network. We forecast stock price using ARIMA, RNN and ARIMA-RNN respectively, and we compare the value of MAE, MSE and MAPE of each model. We conclude that the hybrid ARIMA–RNN model has the best forecasting result.