Time Series Analytics of BIST 100 Companies’ Stocks

Hüseyin Akkaş; Prof. Dr. Hamza Erol

Makale detayları

Time Series Analytics of BIST 100 Companies’ Stocks

Yazarlar:

Hüseyin Akkaş | Prof. Dr. Hamza Erol

Yayıncı:

Mersin University

Ana Başlıklar:

Yapay Zeka

Dergi:

Bilgisayar Bilimleri ve Teknolojileri Dergisi Vol.3 Issue 2, pp.65-71

Belge Tipi:

Makale

Yayın Tarihi:

2/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281351

pISSN:

2717-8579

Anahtar kelimeler:

Özet:

 Today, the stock market is a fairly large market which includes tradable instruments, currency, futures etc. Investors can invest in this market by choosing the investment instruments suitable for them. But the stock market is a highly volatile market, dependent on many variables today. Factors such as bank rate, unemployment, inflation rate are just some of the factors that affect the stock market. These movements, which take place in the market, which contains so much data and factors, are also occuring quite rapidly. It is difficult for the investor to catch up with these movements sometimes. Therefore confusions and hesitations may occur. In general, deep learning architectures and machine learning methods give very successful results in big data analytics lately. In this study, time series analytics of stocks traded in the stock market were performed by using Gated Recurrent Units and Recurrent Neural Networks, some of the deep learning architectures, and the results were evaluated. The purpose of the study is not to provide an investment recommendation, but to test the success of these architectures in big data analytics. In this study, the past 7 years data of ODAS, YKBNK and KARSN stocks, which are the 3 most traded stocks in BIST 100, were analyzed and their value movements were tried to be estimated.

 Today, the stock market is a fairly large market which includes tradable instruments, currency, futures etc. Investors can invest in this market by choosing the investment instruments suitable for them. But the stock market is a highly volatile market, dependent on many variables today. Factors such as bank rate, unemployment, inflation rate are just some of the factors that affect the stock market. These movements, which take place in the market, which contains so much data and factors, are also occuring quite rapidly. It is difficult for the investor to catch up with these movements sometimes. Therefore confusions and hesitations may occur. In general, deep learning architectures and machine learning methods give very successful results in big data analytics lately. In this study, time series analytics of stocks traded in the stock market were performed by using Gated Recurrent Units and Recurrent Neural Networks, some of the deep learning architectures, and the results were evaluated. The purpose of the study is not to provide an investment recommendation, but to test the success of these architectures in big data analytics. In this study, the past 7 years data of ODAS, YKBNK and KARSN stocks, which are the 3 most traded stocks in BIST 100, were analyzed and their value movements were tried to be estimated.

 Today, the stock market is a fairly large market which includes tradable instruments, currency, futures etc. Investors can invest in this market by choosing the investment instruments suitable for them. But the stock market is a highly volatile market, dependent on many variables today. Factors such as bank rate, unemployment, inflation rate are just some of the factors that affect the stock market. These movements, which take place in the market, which contains so much data and factors, are also occuring quite rapidly. It is difficult for the investor to catch up with these movements sometimes. Therefore confusions and hesitations may occur. In general, deep learning architectures and machine learning methods give very successful results in big data analytics lately. In this study, time series analytics of stocks traded in the stock market were performed by using Gated Recurrent Units and Recurrent Neural Networks, some of the deep learning architectures, and the results were evaluated. The purpose of the study is not to provide an investment recommendation, but to test the success of these architectures in big data analytics. In this study, the past 7 years data of ODAS, YKBNK and KARSN stocks, which are the 3 most traded stocks in BIST 100, were analyzed and their value movements were tried to be estimated.

 Borsa, günümüzde alınıp satılabilir menkul kuymetler, döviz, vadeli işlemler vb. içeren oldukça büyük bir piyasadır. Yatırımcılar, kendilerine uygun yatırım araçlarını seçerek bu pazara yatırım yapabilmektedirler. Fakat borsa, günümüzde birçok değişkene bağlı, oldukça hareketli bir pazardır. Faiz oranı, işsizlik, enflasyon oranı gibi etkenler borsayı etkileyen faktörlerden sadece bazılarıdır. Bu kadar çok veriyi ve etkeni içeren pazarda gerçekleşen bu hareketler de oldukça hızlı gelişmektedir. Yatırımcının bu hareketlere yetişmesi de zaman zaman zorlaşmaktadır. Dolayısıyla kafa karışıklıkları ve tereddütler meydana gelebilmektedir. Genel olarak derin öğrenme mimarileri ve makine öğrenmesi yöntemleri günümüzde büyük veri analitiği konusunda oldukça başarılı sonuçlar vermektedir. Bu çalışmada da derin öğrenme mimarilerinden Kapılı Tekrarlayan Birimler Gated Recurrent Units ve Tekrarlayan Sinir Ağları Recurrent Neural Networks kullanılarak borsada işlem gören hisse senetlerinin zaman serisi analitiği gerçekleştirilmiş ve sonuçlar değerlendirilmiştir. Çalışmanın amacı bir yatırım tavsiyesi sunmak değil, kullanılan yöntemlerin büyük veri analitiğindeki başarısını test etmektir. Bu çalışmada, BIST 100 içinde en çok işlem gören 3 hisse senetlerinden ODAS, YKBNK ve KARSN hisse senetlerinin geçmiş 7 yıllık verilerinin analitiği gerçekleştirilmiş ve değer hareketleri tahmin edilmeye çalışılmıştır.

Makale detayları

Time Series Analytics of BIST 100 Companies’ Stocks

Yazarlar:

Hüseyin Akkaş | Prof. Dr. Hamza Erol

Yayıncı:

Mersin University

Ana Başlıklar:

Yapay Zeka

Dergi:

Bilgisayar Bilimleri ve Teknolojileri Dergisi Vol.3 Issue 2, pp.65-71

Belge Tipi:

Makale

Yayın Tarihi:

2/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281351

pISSN:

2717-8579

Big Data

Deep Learning

BIST 100

GRU

RNN

Büyük Veri

Derin Öğrenme

BIST 100

GRU

RNN

Sayı

Makale detayları

Yazarlar: Hüseyin Akkaş | Prof. Dr. Hamza Erol

Yayıncı: Mersin University

Ana Başlıklar: Yapay Zeka

Dergi: Bilgisayar Bilimleri ve Teknolojileri Dergisi Vol.3 Issue 2, pp.65-71

Belge Tipi: Makale

Yayın Tarihi: 2/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001281351

pISSN: 2717-8579

Sayı

Yazarlar:

Hüseyin Akkaş | Prof. Dr. Hamza Erol

Yayıncı:

Mersin University

Ana Başlıklar:

Yapay Zeka

Dergi:

Bilgisayar Bilimleri ve Teknolojileri Dergisi Vol.3 Issue 2, pp.65-71

Belge Tipi:

Makale

Yayın Tarihi:

2/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281351

pISSN:

2717-8579