Estimation of İstanbul Stock Exchange share return via investor sentiment

Estimation of share return in developing countries such as Turkey is among the important subjects in finance studies. Also, this estimation is important for investors. In this study, it was aimed to estimate stock returns through investor sentiment. A detailed literature study was conducted in the theoretical part of the study. In the application part, the BIST 100 Return Index was used as dependent variable, also indirect (trading volume, trading ratio etc) and direct variables (consumer confidence index, VIX fear index), which represented the Investor Sentiment, were employed in the models developed for the estimation. In this context, a total of 158 models, 12 of which were direct (direct) and 146 of which were indirect variables, were analyzed using Fuzzy Logic and Support Vector Machines methods. Monthly data between 2007 and 2018 were used for all analyzed models. 70% of the 144-month data collected for the variables was used as training data and 30% was used to measure the predictive success of mathematical models. In the obtained findings, it was found that the estimations made by using indirect variables with the support vector machine method gave more successful results. It was inferred from the findings that the investor sentiment was successful in predicting stock returns. Among indirect variables, trading volume and trading ratio variables and TÜİK-TGE and VIX Fear Index from the direct variables were able to predict the stock returns significantly. Generally speaking, BIST 100 Return Index values were estimated to be statistically successful with Investor Sensitivity values.