Developing an Investment Efficiency Prediction Model Using Meta-Analysis and Comparing Its Predictive Power with the Model of Biddle et al. (2009)

Abstract

This study aims to develop a predictive model for investment efficiency using a meta-analysis approach and compare its predictive power with the model of Biddle et al. (2009). The study employed a meta-analysis approach to develop the predictive model. 33 articles out of 85 were selected to identify the factors influencing investment efficiency. The baseline model and the developed model were compared using neural network methods to evaluate their predictive accuracy. The Wilcoxon test was also used to assess the significance of the results. The results indicated that the developed model, using 13 input features, achieved higher predictive accuracy than the baseline model, which only used one feature. The developed model explained 75.75% of the variance in the target variable with an R² of 0.7575, while the baseline model explained only 3.7% of the variance. The Wilcoxon test confirmed a significant difference in the predictive accuracy between the two models. The developed model, incorporating diverse and complex variables, provided a more accurate prediction of investment efficiency in companies listed on the Tehran Stock Exchange. It is recommended that this model be used in investment decision-making processes.