Designing a prediction model for longterm stock return with nonparametric simulation of debt security return

Predicting stock return rate is always one of important issues of financial markets and also major factor in investment. Predicting excess stock return based on debt-based data, other economic variables and capital market in Iran is the main purpose of this research. The proposed model emphasizes on nonlinear relationship between a set of quantitative variables and its structure basis is parametric and nonparametric models by using local kernel smoothing. First of all effective variables of debt return rate was investigated and identified. After identifying effective variables in first stage, prediction model for debt security return rate has been structured. This prediction is designed and presented by two methods of parametric and nonparametric prediction for debt security. In the following in order to determining effective factors of excess stock return rate prediction model structuring is investigated in the form of linear parametric and nonlinear nonparametric, by using results of first stage in three different steps and by observing R_v^2 criteria. The result of model implementation in all three examined steps and based on R_v^2 criteria show that results of various models of nonparametric approaches work much better than parametric approaches.