Machine Learning-Based Exchange Rate Forecasting and Performance Evaluation: A Dual CNN-BiLSTM Framework with Explainable AI for the Pakistani Rupee
Contributors
- Walija Naseer Research Scholar, Department of Management Sciences, COMSATS University Islamabad, Abbottabad, Khyber Pakhtunkhwa, Pakistan.
- Dr. Muhammad Naveed Jan Assistant Professor, Department of Sciences, COMSATS University Islamabad, Abbottabad, Khyber Pakhtunkhwa, Pakistan.
- Dr. Muhammad Shariq Assistant Professor, School of Management Sciences, Ghulam Ishaq Khan Institute of Engineering Sciences and Technology, Topi, Khyber Pakhtunkhwa, Pakistan.
Abstract
In this paper, we design and evaluate a novel dual hybrid architecture based on a combination of CNN and BiLSTM network (CNN-BiLSTM), for predicting exchange rates of the Pakistani Rupee (PKR) vis-à-vis the five key currencies: US Dollar (USD), Euro (EUR), British Pound (GBP), Chinese Yuan (CNY), and UAE Dirham (AED). Two different models with same architectural configuration but with independent training sets are designed in this research: one is a short-term model based on 22 technical indicators extracted from OHLC data of daily time horizon to predict tomorrow’s exchange rate, and another is a long-term model based on 26 features derived from six monthly macroeconomic factors: foreign exchange reserves, KIBOR, net foreign assets, inflation, balance of trade, and external debt servicing to predict next month’s exchange rate. With the 2010–2025-time span which includes 2018 devaluation, the COVID-19 pandemic, and the 2022-2023 currency crisis events, both models are validated using walk-forward cross-validation approach and explained by SHapley Additive exPlanations (SHAP). In terms of technical model, MAPE stands at 0.08%-0.36% while the directional accuracy reaches 67.5%-72.8%, while the MAPE of the macroeconomic model stands at 1.23%-2.64%, directional accuracy at 64%-72%. The most important short-term determinants recognized by SHAP are trend-strength (ADX, +DI), while for the long-term inflation rate, net foreign asset value, and KIBOR are identified as the determinants. The results in the simulated trading environment show that profits and positive alpha are generated over buy-and-hold benchmarks for both models and all five currency pairs, showing statistical and economic relevance and interpretability of the CNN-BiLSTM approach.
References
Abouzaid, O., & Boussedra, F. (2025). Artificial intelligence and exchange rate forecasting: Assessing predictive accuracy and macroeconomic sensitivity. Frontiers in Applied Mathematics and Statistics, 11. https://doi.org/10.3389/fams.2025.1654093
Akhtar, S., Ramzan, M., Shah, S., Ahmad, I., Khan, M. I., Ahmad, S., El-Affendi, M. A., & Qureshi, H. (2022). Forecasting exchange rate of Pakistan using time series analysis. Mathematical Problems in Engineering, 2022, 1–11. https://doi.org/10.1155/2022/9108580
Asadullah, M., Bashir, A., & Aleemi, A. R. (2021). Forecasting exchange rates: An empirical application to Pakistani rupee. Journal of Asian Finance, Economics and Business, 8(4), 339–347. https://doi.org/10.13106/jafeb.2021.vol8.no4.0339
Ayitey, M., Appiahene, P., & Appiah, O. (2022). Forex market forecasting with two-layer stacked long short-term memory neural network (LSTM) and correlation analysis. Journal of Electrical Systems and Information Technology, 9(1). https://doi.org/10.1186/s43067-022-00054-1
Biswas, A., Uday, I., Rahat, K., Akter, M., & Mahdy, M. (2023). Forecasting the USD/BDT exchange rate with deep learning models: Inclusion of macroeconomic factors influencing the currency exchange rates. PLOS ONE, 18(2), e0279602. https://doi.org/10.1371/journal.pone.0279602
Chandra, R., Goyal, S., & Gupta, R. (2021). Evaluation of deep learning models for multi-step ahead time series prediction. IEEE Access, 9, 83105–83123. https://doi.org/10.1109/ACCESS.2021.3085085
Chihab, Y., Bousbaa, Z., Chihab, M., Bencharef, O., & Ziti, S. (2019). Algo-trading strategy for intraweek foreign exchange speculation based on random forest and probit regression. Applied Computational Intelligence and Soft Computing, 2019, 1–13. https://doi.org/10.1155/2019/8342461
Corzo, T., Martin-Bujack, K., Portela, J., & Rodríguez-Gallego, A. (2025). Exchange rate regime changes and market efficiency: An event study. Journal of International Financial Markets, Institutions and Money, 100, 102132. https://doi.org/10.1016/j.intfin.2025.102132
D’Angelo, E., & Grimaldi, G. (2017). The effectiveness of the Elliott Waves Theory to forecast financial markets: Evidence from the currency market. International Business Research, 10(6), 1–18. https://doi.org/10.5539/ibr.v10n6p1
Dautel, A., Härdle, W., Lessmann, S., & Seow, H. (2020). Forex exchange rate forecasting using deep recurrent neural networks. Digital Finance, 2(1–2), 69–96. https://doi.org/10.1007/s42521-020-00019-x
Escudero, P., Alcocer, W., & Paredes, J. (2021). Recurrent neural networks and ARIMA models for Euro/Dollar exchange rate forecasting. Applied Sciences, 11(12), 5658. https://doi.org/10.3390/app11125658
Fang, S., Wei, Y., & Wang, S. (2023). 30 years of exchange rate analysis and forecasting: A bibliometric review. Journal of Economic Surveys, 37(4), 1123–1150. https://doi.org/10.1111/joes.12572
Frenkel, J. A., & Razin, A. (1987). The Mundell–Fleming model: A quarter century later (NBER Working Paper No. 2321). National Bureau of Economic Research. https://doi.org/10.3386/w2321
García, F., Guijarro, F., Oliver, J., & Tamošiūnienė, R. (2024). Foreign exchange forecasting models: LSTM and BiLSTM comparison. Engineering Proceedings, 68(1), 19. https://doi.org/10.3390/engproc2024068019
Jabeen, M., Rashid, A., & Ihsan, H. (2022). The news effects on exchange rate returns and volatility: Evidence from Pakistan. International Journal of Finance & Economics, 27(1), 745–769. https://doi.org/10.1002/ijfe.2176
Kim, D., Kim, D., & Choi, S. (2025). Modeling stylized facts in FX markets with FINGAN-BiLSTM: A deep learning approach to financial time series. Entropy, 27(6), 635. https://doi.org/10.3390/e27060635
Korol, T. (2014). A fuzzy logic model for forecasting exchange rates. Knowledge-Based Systems, 67, 49–60. https://doi.org/10.1016/j.knosys.2014.06.009
Liu, S., Huang, Q., Li, M., & Wei, Y. (2024). A new LASSO-BiLSTM-based ensemble learning approach for exchange rate forecasting. Engineering Applications of Artificial Intelligence, 127, 107305. https://doi.org/10.1016/j.engappai.2023.107305
López-Herrera, F., Jiménez, J. G. M., & Santiago, A. R. (2025). Directional forecasting for eight forex pairs against the US dollar using machine learning techniques. Discover Artificial Intelligence, 5, Article 224. https://doi.org/10.1007/s44163-025-00424-4
Mancini Griffoli, T., Meyer, C., Natal, J.-M., & Zanetti, A. (2015). Determinants of the Swiss franc real exchange rate. Swiss Journal of Economics and Statistics, 151(4), 303–328. https://doi.org/10.1007/BF03399419
Meng, S., Chen, A., Wang, C., Zheng, M., Wu, F., Chen, X., Ni, H., & Li, P. (2024). Enhancing exchange rate forecasting with explainable deep learning models. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2410.19241
Neely, C. J., & Weller, P. A. (2011). Technical analysis in the foreign exchange market (Working Paper No. 2011-001B). Federal Reserve Bank of St. Louis. http://research.stlouisfed.org/wp/2011/2011-001.pdf
Nguyen, P. D., Thao, N. N., Kim, T., Nguyen, H.-C., Mach, B.-N., & Nguyen, T. Q. (2024). Deep learning-based predictive models for forex market trends: Practical implementation and performance evaluation. Science Progress, 107(3). https://doi.org/10.1177/00368504241275370
Ni, L., Li, Y., Wang, X., Zhang, J., Yu, J., & Qi, C. (2019). Forecasting of Forex time series data based on deep learning. Procedia Computer Science, 147, 647–652. https://doi.org/10.1016/j.procs.2019.01.189
Nsengiyumva, E., Mung’atu, J., & Ruranga, C. (2025). A comparative study of multivariate CNN, BiLSTM and hybrid CNN–BiLSTM models for forecasting foreign exchange rate using deep learning. Cogent Economics & Finance, 13(1), 2526148. https://doi.org/10.1080/23322039.2025.2526148
Nwanne, T. F. I., & Onyekachi, E. R. (2015). Assessing the effect of external debt servicing and receipt on exchange rate in Nigeria. International Journal of Economics and Finance, 7(9), 278–285. https://doi.org/10.5539/ijef.v7n9p278
Panda, C., & Narasimhan, V. (2007). Forecasting exchange rate better with artificial neural network. Journal of Policy Modeling, 29(2), 227–236. https://doi.org/10.1016/j.jpolmod.2006.01.005
Panopoulou, E., & Souropanis, I. (2019). The role of technical indicators in exchange rate forecasting. Journal of Empirical Finance, 53, 197–221. https://doi.org/10.1016/j.jempfin.2019.07.004
Perla, S., Bisoi, R., & Dash, P. K. (2023). A hybrid neural network and optimization algorithm for forecasting and trend detection of Forex market indices. Decision Analytics Journal, 6, 100193. https://doi.org/10.1016/j.dajour.2023.100193
Perveen, N., Tariq, K., & Ahmad, H. S. (2024). Comparative studies of hybrid ARIMA and artificial neural network (ANN) techniques for predicting exchange rate in Pakistan. Bulletin of Business and Economics, 13(2), 630–636. https://doi.org/10.61506/01.00372
Pornwattanavichai, A., Maneeroj, S., & Boonsiri, S. (2022). BERTFOREX: Cascading model for forex market forecasting using fundamental and technical indicator data based on BERT. IEEE Access, 10, 23425–23437. https://doi.org/10.1109/access.2022.3152152
Qureshi, A. M. (2025). Machine learning prediction of exchange rates using macroeconomic indicators. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5287273
Qureshi, M., Ahmad, N., Ullah, S., & Mustafa, A. R. U. (2023). Forecasting real exchange rate (REER) using artificial intelligence and time series models. Heliyon, 9(5), e16335. https://doi.org/10.1016/j.heliyon.2023.e16335
Rashid, A., & Basit, M. (2021). Empirical determinants of exchange-rate volatility: Evidence from selected Asian economies. Journal of Chinese Economic and Foreign Trade Studies, 15(1), 63–86. https://doi.org/10.1108/JCEFTS-04-2021-0017
Sezer, O. B., Gudelek, M. U., & Ozbayoglu, A. M. (2020). Financial time series forecasting with deep learning : A systematic literature review: 2005–2019. Applied Soft Computing, 90, 106181. https://doi.org/10.1016/j.asoc.2020.106181
Sönmez, F. E., & Birim, Ş. Ö. (2024). Forecasting exchange rate depending on the data volatility: A comparison of deep learning techniques. https://doi.org/10.21203/rs.3.rs-4218174/v1
Wagdi, O., Salman, E., & AlBanna, H. (2023). Integration between technical indicators and artificial neural networks for the prediction of the exchange rate: Evidence from emerging economies. Cogent Economics & Finance, 11(1), 2255049. https://doi.org/10.1080/23322039.2023.2255049
Wang, J., Wang, X., Li, J., & Wang, H. (2021). A prediction model of CNN-TLSTM for USD/CNY exchange rate prediction. IEEE Access, 9, 73346–73354. https://doi.org/10.1109/ACCESS.2021.3080459
Yao, J., & Tan, C. L. (2000). A case study on using neural networks to perform technical forecasting of forex. Neurocomputing, 34(1–4), 79–98. https://doi.org/10.1016/S0925-2312(00)00300-3
Yıldırım, D. C., Toroslu, İ. H., & Fiore, U. (2021). Forecasting directional movement of Forex data using LSTM with technical and macroeconomic indicators. Financial Innovation, 7(1), 1–22. https://doi.org/10.1186/s40854-020-00220-2
Corresponding Author
Dr. Muhammad Naveed Jan
Assistant Professor, Department of Sciences, COMSATS University Islamabad, Abbottabad, Khyber Pakhtunkhwa, Pakistan.
Corresponding Author: [email protected]
