<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>Stock price forecasting is an eye-catching research topic. In previous works, many researchers used a single method or combination of methods to make predictions. However, accurately predicting stock prices is very difficult. To improve the predicting precision, in this study, an innovative prediction approach was proposed by recurrent substitution of forecast error into the historical neural network model through three steps. According to the historical data, the initial predicted value of the next day is obtained through the neural network. Then, the prediction error of the next day is obtained through the neural network according to the historical prediction error. Finally, the initial predicted value and the prediction error are added to obtain the final predicted value of the next day. We use recurrent neural network prediction methods, such as Long Short-Term Memory Network Model and Gated Recurrent Unit, which are popular in the recent neural network study. In the simulations, the past stock prices of China from June 2010 to August 2017 are used as training data, and those from September 2017 to April 2018 are used as test data. The experimental findings demonstrate that the proposed method with forecast error gives a more accurate prediction result for the stock’s high price on the next day, which indicates that the performance of the proposed one is superior to that of the traditional models without forecast error.</p>
</abstract>

Stock price forecasting is an eye-catching research topic. In previous works, many researchers used a single method or combination of methods to make predictions. However, accurately predicting stock prices is very difficult. To improve the predicting precision, in this study, an innovative prediction approach was proposed by recurrent substitution of forecast error into the historical neural network model through three steps. According to the historical data, the initial predicted value of the next day is obtained through the neural network. Then, the prediction error of the next day is obtained through the neural network according to the historical prediction error. Finally, the initial predicted value and the prediction error are added to obtain the final predicted value of the next day. We use recurrent neural network prediction methods, such as Long Short-Term Memory Network Model and Gated Recurrent Unit, which are popular in the recent neural network study. In the simulations, the past stock prices of China from June 2010 to August 2017 are used as training data, and those from September 2017 to April 2018 are used as test data. The experimental findings demonstrate that the proposed method with forecast error gives a more accurate prediction result for the stock’s high price on the next day, which indicates that the performance of the proposed one is superior to that of the traditional models without forecast error.

Predicting stock high price using forecast error with recurrent neural network

School of Computer and Information Engineering

Capital University of Economics and Business

Graduate School of Information, Production and Systems

Applied Mathematics and Nonlinear Sciences

This work is licensed under the Creative Commons Attribution 4.0 International License.

{"article-title":"Predicting stock high price using forecast error with recurrent neural network"}

Stock price forecasting is an eye-catching research topic. In previous works, many researchers used a single method or combination of methods to make...

Models	Time	RMSE	MAPE	MASE	Dstat
GRU	0.090	28.676	0.683	1.281	0.693
GRU,GRU	0.178	21.269	0.449	0.837	0.713
GRU,LSTM	0.193	21.319	0.450	0.840	0.703
GRU,MLP	0.112	23.559	0.517	0.968	0.710

Data set	Training	Test
000001.SH	595.8	106.6
399001.SZ	1773.1	749.3

Models	Time	RMSE	MAPE	MASE	Dstat
GRU	0.090	127.056	1.143	1.048	0.621
GRU,GRU	0.177	115.375	1.021	0.934	0.621
GRU,LSTM	0.109	126.870	1.092	1.007	0.620
GRU,MLP	0.023	120.234	1.050	0.961	0.615

Models	Time	RMSE	MAPE	MASE	Dstat
LSTM	0.102	27.243	0.643	1.204	0.653
LSTM,GRU	0.188	22.195	0.481	0.900	0.673
LSTM,LSTM	0.207	22.225	0.483	0.903	0.667
LSTM,MLP	0.125	23.570	0.526	0.984	0.669

Models	Time	RMSE	MAPE	MASE	Dstat
MLP	0.023	29.953	0.729	1.368	0.640
MLP,GRU	0.114	23.964	0.521	0.973	0.665
MLP,LSTM	0.128	23.328	0.509	0.950	0.658
MLP,MLP	0.045	24.727	0.560	1.047	0.657

Predicting stock high price using forecast error with recurrent neural network

Pubblicato online: 25 mag 2021

Pagine: 283 - 292

Ricevuto: 24 dic 2020

Accettato: 11 apr 2021

DOI: https://doi.org/10.2478/amns.2021.2.00009

Parole chiave
stock price prediction, recurrent neural network, long short-term memory network, gated recurrent unit

© 2021 Zhiguo Bao et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

The experiment results (NN1: GRU, DATA SET: 000001.SH).

The standard deviation of experimental data.

The experiment results (NN1: GRU, DATA SET: 399001.SZ).

The experiment results (NN1: LSTM, DATA SET: 000001.SH).

The experiment results (NN1: MLP, DATA SET: 000001.SH).

Predicting stock high price using forecast error with recurrent neural network

Pubblicato online: 25 mag 2021

Pagine: 283 - 292

Ricevuto: 24 dic 2020

Accettato: 11 apr 2021

DOI: https://doi.org/10.2478/amns.2021.2.00009

Parole chiavestock price prediction, recurrent neural network, long short-term memory network, gated recurrent unit

© 2021 Zhiguo Bao et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

The experiment results (NN1: GRU, DATA SET: 000001.SH).

The standard deviation of experimental data.

The experiment results (NN1: GRU, DATA SET: 399001.SZ).

The experiment results (NN1: LSTM, DATA SET: 000001.SH).

The experiment results (NN1: MLP, DATA SET: 000001.SH).

Parole chiave
stock price prediction, recurrent neural network, long short-term memory network, gated recurrent unit