Forecasting of Stock Returns based on the Approach of Bayesian Models Averaging; Quantum Finance and Continuous Wavelet Analysis

Linear models due to the lack of correct extraction of the shape of the conditional distribution of data; Failure to record the dynamic behavior of the conditional distribution of data; the existence of limiting assumptions contrary to reality; They do not have the proper ability to predict returns in today's world. The main goal of the current research is to resolve the ambiguity in determining the appropriate model for forecasting stock returns in Tehran capital market in different time frames.This research is of an applied type. The time domain of the data used in this research is daily data from 2018/9/23 to 2022/09/23. To predict and model stock returns in this research from 8 categories of estimation models 1- Classical or Structural, 2- Non-Structural regressions; 3- Time-varying Parameter Bayesian regressions, 4- Discrete Wavelet transform and Continuous Wavelet transform models, 5- Metaheuristic Approaches, 6- Simple and Deep Artificial Neural Networks approaches, 7- Stochastic differential, 8- Financial quantum were investigated.Based on the results in the short term of 1 day, Bayesian averaging models; In the medium term of 16 days, financial quantum models and in the long term of 32 days, continuous wave models had higher accuracy. Based on the finding of research, it can be acknowledged that in order to predict stock returns, it is necessary to use different models in different time frames, and using the same approach will reduce the accuracy of Predicting stock returns.