Stock Price Prediction With ARIMA and Deep Learning Models

Financial markets are vital to the capitalist economies and often volatile and hard to predict. This work compares the performance of different time-series models in predicting close price movement for 30 listed stocks from the Dow Johns Industrial Average (DIJA). The mechanisms of autoregressive moving average (ARIMA), artificial neural network (ANN), recurrent neural network (RNN), and long short-term memory (LSTM) are briefly explained in the essay. Comparison results suggest that LSTM and sequence to sequence (Seq2Seq) model with attention estimates fit the price movement pattern well with relatively low latency. Built upon LSTM, Seq2Seq models exhibit good performance in forecasting. The vanilla Seq2Seq model is compared with Seq2Seq with attention in forecasting price several days in the future. Seq2Seq with attention outperforms other models and is capable of generating sequential predictions.