• Unbareen Arif University of Education, Pakistan
  • Naveed Khan Abbottabad University of Science and Technology, Pakistan
  • Junaid Rauf Burki MOL Pakistan, OIL and Gas Company. B.v.


Stock returns, forecasting, investment decisions, Financial Markets, Machine Learning. JEL Classification: G17, C53, G11, E44, M15


The Stock Exchange performance is reflector of financial health of economy, while prediction of stock returns considered to be a complex task.  The stock returns are determined by many factors from company specific to the macroeconomic indicators and also dependent on behavioral factors associated with the investors sentiments. The emergence of Artificial Intelligence (AI) and Machine Learning (ML) have revolutionized the traditional statistical approaches to forecast returns. The Current study is an attempt to forecast stock returns using ML Algorithm “prophet” to analyze performance of the prediction model via comparison of forecasted returns with actual returns. The model is implemented and tested in emerging market stock returns where the returns are highly volatile. For the purpose of analysis, data of all the firms listed in Oil and Gas sector in PSX were selected w.e.f. 2012 to 2021.The data distributed in training and testing samples to forecast returns with prophet model using python. The model performance is evaluated with evaluation matrix of MAE, MSE and RMSE. The results of the study indicated that the OGDC stock has reported superior performance of ML algorithm to forecast returns with MAE 0.002, MSE 0.0001 and RMSE 0.0108. The  98% indicates that the Machine learning prophet model has greater ability to predict returns as the algorithm provides flexibility to capture trend, seasonality and holidays effect to forecast results according to the analyst requirements. The findings of study are useful for the fund Managers, investors and researchers to analyze trend and make optimal investment decisions.




How to Cite

Arif, U., Khan, N., & Burki, J. R. . (2023). FORECASTING STOCK RETURNS: APPLICATION OF MACHINE LEARNING ALGORITHM TO PSX. International Journal of Business and Management Sciences, 4(3), 45-57. Retrieved from