文章基本信息

标题：EVALUATING THE IMPACT OF MULTIPLE LEADING INDICATOR IN FORECASTING NEXT DAY STOCK PRICE WITH LSTM
本地全文：下载
作者：ADITYO HIDAYAT ; AGUS SIHABUDDIN
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：2
页码：293
出版社：Journal of Theoretical and Applied
摘要：LSTM was the preferred choice for financial time series forecasting, whereas fundamental analysis and technical analysis were among the most favorable feature sets. Earlier studies had several suggestions to improve forecasting performance, by using features known to carry information about the future price movement, and features associated with substantial price movements: the foreign investors' trading volume. Overall trading volume and those volumes from foreign investors have been introduced as a leading indicator. However, empirical literatures which examines price-volume relationship using LSTM had not used foreign investors� trading volume. This study evaluates the use of multiple leading indicators as input, and optimum hyperparameters configurations using LSTM to next day prediction performance. Experiments are evaluated based on 88 stocks in Indonesia stock market, ranging from Jan. 2, 2015 to Dec. 30, 2019. Financial time series forecasting using simple LSTM architecture performs as good as baseline performance with the advantage of fewer computing requirements. Optimum hyperparameters are a single hidden layer, 50 nodes, and ten days of the input window. The highest winning stocks are achieved using feature sets consisting of a lagging indicator (price) and multiple leading indicators (overall trading volume and foreign investors' trading volume). The findings indicate that multiple leading indicators contain predictability factors which can be further explored to improve financial time series forecasting. This study contributes the use of foreign investors� which improves financial time series forecasting with LSTM.
关键词：Financial Time Series Forecasting; Foreign Trading Volume; Stock Market; LSTM