A Trend-based Stock Index Forecasting Model with Gated Recurrent Neural Network

Prediction of stock index is seen as a challenging task of financial time series prediction. In this study, we proposed a movement trend-based data preparation method, preprocessing the trend indicator in two steps, and modeled stock index using Gated Recurrent Unit (GRU). In the two-stage preproces...

Full description

Saved in:
Bibliographic Details
Published in2018 IEEE International Conference on Progress in Informatics and Computing (PIC) pp. 425 - 429
Main Authors Liu, Yijiao, Liu, Xinghua
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2018
Subjects
data preparation
Feature extraction
Forecasting
Gated Recurrent Unit
Indexes
Predictive models
Recurrent neural networks
stock prediction
Time series analysis
Online AccessGet full text
DOI10.1109/PIC.2018.8706267

Cover

More Information
Summary:Prediction of stock index is seen as a challenging task of financial time series prediction. In this study, we proposed a movement trend-based data preparation method, preprocessing the trend indicator in two steps, and modeled stock index using Gated Recurrent Unit (GRU). In the two-stage preprocessing, the trend indicators were extracted from five aspects and then were discretized based on the dynamic relationship. And three Recurrent Neural Networks were applied to forecast financial time series. Compared with random tri-prediction method, our model improved the stock index movement trend prediction from 33% to 68%.
DOI:10.1109/PIC.2018.8706267