Earnings per Share Forecasting Using Machine Learning Algorithms in the Capital Markets of Iraq and the United Arab Emirates
Keywords:
Earnings per share forecasting, emerging markets, machine learning, deep neural network, ESG, financial data analysis, Iraq, United Arab Emirates, natural language processing, reinforcement learningAbstract
This study aimed to develop and test a hybrid machine learning framework for forecasting earnings per share (EPS) in the emerging capital markets of Iraq and the UAE. Data from 65 Iraqi and 150 Emirati listed companies between 2019 and 2024 were collected, including traditional financial indicators, macroeconomic variables, and ESG metrics. After preprocessing and normalization, several machine learning models were implemented, including artificial neural networks, support vector machines, random forests, gradient boosting machines (GBM), reinforcement learning, federated learning, and NLP-based models using FinBERT and FastText. Model performance was assessed using MAE, RMSE, R², and the Diebold-Mariano test. The GBM model achieved the best performance, with MAE=0.127 and R²=0.78 in Iraq, and MAE=0.110 and R²=0.83 in the UAE. NLP models also performed strongly in the UAE (R²=0.83). Sensitivity analysis revealed that financial variables such as net profit and operating cash flow were the most influential, while oil prices in Iraq and innovation indices in the UAE had the greatest macroeconomic impact. Out-of-sample tests with 2025 data confirmed the superior generalizability of GBM and NLP. Machine learning algorithms, particularly GBM, significantly enhance EPS forecasting accuracy in emerging markets. Structural and institutional differences between Iraq and the UAE highlight the need for context-specific model selection. The proposed hybrid framework offers a practical reference for other developing capital markets.
Downloads
References
Al-Zubaidi, H. S., Abdulhameed, M. A., & Fenjan, F. H. (2025). Evaluating common stock prices through the residual income model: Evidence from Iraq stock exchange. Journal of Economic and Administrative Sciences.
Alqahtani, F., & Mayhew, B. (2022). Macroeconomic uncertainty and earnings predictability in emerging markets. Journal of International Accounting Research, 21(3), 45-68.
Chang, T. Y., Lee, J. H., & Lee, Y. C. (2023). Applying sustainable development goals in financial prediction models: ESG integration for long-term risk assessment. Sustainability.
Chen, H., Wang, J., & Zhou, Y. (2022). Gradient boosting models for financial forecasting: An empirical comparison. Expert Systems with Applications, 193, 116423. https://doi.org/10.1016/j.eswa.2021.116423
Etemadi, H., Ahmadpour, A., & Moshashaei, H. (2015). Earnings per share forecast using extracted rules from decision tree and genetic algorithm. Iranian Journal of Management Studies. https://doi.org/10.1007/s10614-014-9455-6
Hoseinzade, E., Haratizadeh, S., & Ebrahimi, S. (2019). Stock market prediction using machine learning: A hybrid approach. Journal of Forecasting, 38(6), 600-619.
Jamali, A., & Taj Ardekani, A. (2016). Explain the relationship between profit forecasts by management and stock return in accepted companies in Tehran stock exchange. Iranian Journal of Accounting Research.
Kim, S., & Won, H. (2021). Deep learning in financial forecasting: The role of data quality and market stability. Applied Intelligence, 51(5), 3212-3229.
Krauss, C., Do, X. A., & Huck, N. (2020). Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the S&P 500. European Journal of Operational Research, 281(2), 518-532.
Kuryłek, D. (2024). Artificial neural networks and gradient boosting in financial forecasting: Evidence from Poland. Journal of Forecasting.
Mahmud, K., Akter, S., & Ahammed, S. (2024). Value relevance of key accounting information in emerging markets. Asian Journal of Finance & Accounting. https://doi.org/10.54728/JFMG.202310.00072
Pouransari, H., & Chalabi, M. (2014). Event-based stock market prediction using natural language processing. International Journal of Computer Applications.
Raghupathi, W. (1996). A neural network approach to forecasting earnings per share. Information Sciences.
Slavin, H. (2007). Aggregating earnings per share forecasts. International Journal of Forecasting.
Zhang, Y., & Wang, J. (2020). Financial time series forecasting with machine learning: An overview. IEEE Access, 8, 23306-23326.
Zhang, Z., Chen, Y., & Li, J. (2021). Random forest and gradient boosting for stock return prediction: Evidence from emerging markets. Finance Research Letters, 39, 101620. https://doi.org/10.1016/j.frl.2020.101539
Downloads
Published
Submitted
Revised
Accepted
Issue
Section
License
Copyright (c) 2024 Hayder Huodur Radhi Al-Sarray, Meysam Doaei, Ibrahim Abed Mousa Alsabary , Mohammad Alimoradi (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
