@article{article_638575, title={Effectiveness of Turkish Derivatives Market and Forecasting Comparative Prices for the Contracts}, journal={Ege Academic Review}, volume={19}, pages={469–485}, year={2019}, DOI={10.21121/eab.638575}, author={Taş, Taner and Selim, Sibel}, keywords={Derivatives Markets,Efficient Market Hypothesis,Unit Root Test,ARIMA Models,ARCH/ GARCH Models,Artificial Neural Networks}, abstract={Derivative markets developed for eliminating  uncertainty and risk arising from financial markets  can make predictions about the future by using  past price movements in case the market is not  effective. In this context, in this study, firstly, the  effectiveness of the Turkish Derivatives Market was  tested by applying the Augmented Dickey-Fuller  (ADF), Phillips-Perron (PP) and Kwiatkowski et al.  (KPSS) linear unit root tests and Kapetanios et al.  (KSS) nonlinear unit root test. As a result of all unit  root tests, it was concluded that the series did not  show random walk, so that the market was not  effective. Then, the method that shows the highest  performance is tried to be determined when  forecasting the end of day settlement price of the  TL/Dollar and Bist-30 contracts which is traded in the  Derivatives Market. For this purpose, the forecasting  results produced by the time series analysis methods  are compared with the results of the artificial neural  network model which has the best performance by  employing different architectures, layer numbers,  cell numbers in layers, activation functions and  learning methods using the data which is provided  from Borsa Istanbul Inc. and covering the dates  between 04.02.2005 and 31.12.2015.According to  the results of analysis, ARMA (4,4) model performed  better than RBF-1-BL artificial neural network  model and ARCH (1) model for TL/Dollar contract  series. For the Bist-30 contract series, TDNN-1-B-L  artificial neural network model has higher predictive  performance than ARMA (4.5) and ARCH (1) models. }, number={4}, publisher={Ege University}