A supervised fuzzy network analysis for risk assessment in stock markets: An ANFIS approach

In this paper we have used an adaptive neuro-fuzzy inference system (ANFIS) approach to predict the risk of stocks. Previous works just predict the return of stocks and make their portfolio based on the predicted return. But for developing a portfolio both risk and return should be predicted. Our model predicts the risk without needing to experts and just with using available data in the market. To generate the membership functions, we use Fuzzy C-mean clustering algorithm. To test our neuro-fuzzy model we´ve used data on portfolios constituted from the Tehran Stock Exchange. The results show that the error of prediction is so small.