期刊名称：Dokuz Eylül Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi
印刷版ISSN：1302-504X
电子版ISSN：2147-7973
出版年度：2016
卷号：31
期号：2
DOI：10.24988/deuiibf.2016312517
语种：English
出版社：Dokuz Eylül University
摘要：Forecasting time series is an attractive field. Stock prices are also time series and forecasting stock prices has many advantages. In previous studies, it is observed that generally one-day ahead stock prices or trends are forecasted. In this study, different from the literature, 1-day ahead, 2-days ahead and 20-days ahead closing prices are forecasted. The dataset belongs to stocks listed in Borsa Istanbul 30 Index between the period January 2010 and November 2015. Several technical indicators are used as input features. The forecasting model used in this study is artificial neural network models. At the end of the study it is observed that hit rate of 20 days ahead forecasting of stock prices reached 72.88%. The results of the study reveal that multi-step ahead forecasting has important potential that deserve a deep examination.
其他摘要：Zaman serileri tahmini literatürde ilgi gören bir konudur. Hisse senedi fiyatları da zaman serisi oluşturmaktadır ve hisse senedi fiyat tahmininin birçok avantajlı tarafı mevcuttur. Literatürde hisse senedi fiyat tahmini üzerine yapılan çalışmalarda genel
关键词：Multi-step-ahead price forcasting;stock price forecasting;artificial neural networks;Borsa Istanbul
其他关键词：Çok Aşamalı Fiyat Tahmini;Hisse Senedi Fiyat Tahmini;Yapay Sinir Ağları;Borsa İstanbul
