Forecast stock index return with neuro fuzzy network structure: BIST 100 case

In this study, Istanbul Stock Exchange 100 index (BIST - 100) tried to predict using gold prices, exchange rates, consumer price index, interest rates on deposits, the Dow Jones Index, the current account deficit and gross domestic product values by fuzzy neural network inference method . The reason for selection stock market index as the subject, as an investment instrument stock market index is an important alternative investment instrument. The dynamic nature of the stock market index is affected by different variables for each period. The timeliness of the data affect the efficacy, produce result. In this respect, 2013 year-end data used in study. Study included time from 2002 to 2013. Data were included Monthly in the study. As a result, the impact on the index ofoverall economic balance were observed. Increased significantly of the success of prediction with the obtained models by varying of parameters values of fuzzy inference system are set forth. Prediction success rate of 93% was achieved with build different fuzzy neural network models. With differentiate parameters estimates success is increased by %10 according to R2, between the best and worst models.