\ FORECASTING ISTANBUL STOCK EXCHANGE

FORECASTING ISTANBUL STOCK EXCHANGE

buğra
Buğra Erkartal
Lisansüstü Bursiyer

Tarih ve Saat

25 Mayıs 2018 - 10:00

Yer

Mühendislik Binası A-511

Title: FORECASTING ISTANBUL STOCK EXCHANGE

Abstract: Machine learning methods such as Artificial Neural Networks (ANN) and Support Vector Machines (SVM) have many applications in the area of finance. The majority of these applications focus on the popular problems of share price or exchange rate movement prediction, however, applications on corporate bankruptcy prediction, corporate bond classification and other areas also arise in the literature. The reason behind using machine learning methods is that they are powerful classifiers that are flexible in terms of their assumptions as compared to statistical methods. The flexibility of machine learning classifiers makes them a useful tool to resort to for share price prediction.

This study proposes two novel models, the second being a restricted version of the first. The models generate 5 days ahead buy/sell signals for GARAN (Garanti Bankasi A.Ş.),  an equity share that is the top traded stock in BIST100, Istanbul Stock Exchange -Turkey. The first model includes global macroeconomic indicators as well as local inputs whereas the second model is focused more on local inputs. The performances of the two models are tested using SVM, Neural Network with Back-Propagation (BPN), and Decision Tree (DT) algorithms. Though BPN and SVM have previously been used to predict BIST100 Index movement, DT has not been utilized before with this purpose. The performance of all proposed models/methods are tested for a time span of about 6 months. A simple trading strategy is implemented based on buy/sell signals to calculate the rate of return on investment during the testing period.  All three algorithms are executed on a rolling horizon basis, that is, they are re-run weekly with updated data to generate daily buy/sell signals for the next week. The results illustrate that DT has about 80% prediction accuracy using both models and it outperforms BPN and SVM that have up to 60% prediction accuracy.

Biography: Reşat Buğra Erkartal is an M.Sc. Candidate at Industrial and Systems Engineering Graduate School of Natural and Applied Sciences at Yeditepe University. After he graduated from Deutsche Schule Istanbul in 2003, he graduated from Systems Engineering Dept., Yeditepe University in 2008. After his completion of MSc., he will continue his career in Yeditepe University as a PhD student starting on October, 2018.

Yeditepe Üniversitesi, Endüstri ve Sistem Mühendisliği
26 Ağustos Yerleşimi, Kayışdağı Cad. 34755 Ataşehir, İstanbul

(216) 578 04 50 info@sye.yeditepe.edu.tr