Forecasting common stock returns using time series models and neural networks

An application Study in Iraqi Stock Exchange for the period (2006-2023)

A Dissertation Submitted To The Council of College of Administration and Economics at Karbala University, As Partial Fulfillment of the Requirements for PH.D. Degree Financial and Banking Sciences.

Prepared by

Karrar Hatem Attia Al-Badiri

Under The Supervision

prof. Dr. Ali Ahmed Faris Al-Kaabi

 The study aimed to Improving Common Stock investment decisions through Forecasting the returns of Common Stock through a series of historical data for a sample of companies listed in the Iraq Stock Exchange, And testing the proposed prediction models to reach the best model that can be used by the investor to make the right investment decision, Perhaps time series models and neural network models were and still represent an intellectual debate about the validity and preference of these models in predicting common stock returns after the controversy increased over the preference of one over the other, Therefore, this study came to find out this controversy and try to solve it by testing the above models in the light of the data obtained for the study sample represented in the Iraq Stock Exchange and by (31) companies for the period from (30/6/2006) to (31/1/ 2023), And using many financial, statistical and mathematical methods, The study reached a number of conclusions, Perhaps the most important of them is the superiority of multilayer perceptron network models in reaching accurate prediction values ​​over Box and Jenkins models, The Perceptron network achieved preference at the level of (30) companies out of (31) companies, with a rate of (97%).

The study came out with a number of recommendations, Perhaps the most important of them is the need to adopt multi-layer perceptron network models in predicting the returns of common stock of the study sample companies, with the exception of the National Agricultural Production Company, which recommends using Box and Jenkins models in predicting the returns of its shares, as a result of reaching its preference in forecasting.

Forecasting common stock returns using time series models and neural networks

An application Study in Iraqi Stock Exchange for the period (2006-2023)

A Dissertation Submitted To The Council of College of Administration and Economics at Karbala University, As Partial Fulfillment of the Requirements for PH.D. Degree Financial and Banking Sciences.

Prepared by

Karrar Hatem Attia Al-Badiri

Under The Supervision

prof. Dr. Ali Ahmed Faris Al-Kaabi

 The study aimed to Improving Common Stock investment decisions through Forecasting the returns of Common Stock through a series of historical data for a sample of companies listed in the Iraq Stock Exchange, And testing the proposed prediction models to reach the best model that can be used by the investor to make the right investment decision, Perhaps time series models and neural network models were and still represent an intellectual debate about the validity and preference of these models in predicting common stock returns after the controversy increased over the preference of one over the other, Therefore, this study came to find out this controversy and try to solve it by testing the above models in the light of the data obtained for the study sample represented in the Iraq Stock Exchange and by (31) companies for the period from (30/6/2006) to (31/1/ 2023), And using many financial, statistical and mathematical methods, The study reached a number of conclusions, Perhaps the most important of them is the superiority of multilayer perceptron network models in reaching accurate prediction values ​​over Box and Jenkins models, The Perceptron network achieved preference at the level of (30) companies out of (31) companies, with a rate of (97%).

The study came out with a number of recommendations, Perhaps the most important of them is the need to adopt multi-layer perceptron network models in predicting the returns of common stock of the study sample companies, with the exception of the National Agricultural Production Company, which recommends using Box and Jenkins models in predicting the returns of its shares, as a result of reaching its preference in forecasting.