System Dynamics Features and Regime-Adaptive Ensembles for High-Frequency Currency Trading: A Multi-Modal Machine Learning Approach
Keywords:
System Dynamics, Regime-Adaptive Ensembles, High-Frequency Trading, Currency Prediction, Multi-Modal Machine LearningAbstract
This study introduces a novel multi-modal machine learning framework for high-frequency EURUSD trading that combines regime-adaptive ensemble approaches with system dynamics features. We develop seven system dynamics features—RiskIndex, CarryFlow, CapIn, CapOut, FlowPressure, FairValue_px, and Mispricing—that illustrate the small-scale functioning of the market. Our RAGe-ENS (Regime-Adaptive Gradient Ensemble) approach adjusts the weights of Transformer and XGBoost forecasts according on their ability to identify regimes and their agreement with actual results. Utilizing 4-hour EURUSD data from 2012 to 2025 (20,119 observations), we examine many models over three time periods (1, 3, and 6 periods). RAGe-ENS performs exceptionally well, according to the results, with Sharpe ratios of 2.91 (H=1), 1.41 (H=3), and 1.47 (H=6). Compared to the performance of individual models, this is far superior. The Sharpe ratios of H=1 and H=3 increase by 19.4%, 88.6%, and 6%, respectively, depending on the system dynamics aspects. The framework produces alpha in high-frequency currency markets, as evidenced by its high PSR values and significant statistical significance.
References
M. Ayitey Junior, P. Appiahene, O. Appiah, and C. N. Bombie, "Forex market forecasting using machine learning: Systematic Literature Review and meta-analysis," Journal of Big Data, vol. 10, no. 1, 2023, doi: 10.1186/s40537-022-00676-2.
B. Yuan, "Machine learning in foreign exchange," 2025, doi: 10.2139/ssrn.5142754.
L. Li, P. A. Matt, and C. Heumann, "Forecasting foreign exchange rates with regression networks tuned by Bayesian optimization," 2022. [Online]. Available: http://arxiv.org/abs/2204.12914.
S. Ghahremani and U. T. Nguyen, "Prediction of foreign currency exchange rates using an attention-based long short-term memory network," Machine Learning with Applications, vol. 20, p. 100648, 2025, doi: 10.1016/j.mlwa.2025.100648.
T. Fischer, M. Sterling, and S. Lessmann, "Fx-spot predictions with state-of-the-art transformer and time embeddings," Expert Systems with Applications, vol. 249, p. 123538, 2024, doi: 10.1016/j.eswa.2024.123538.
S. T. Mndawe, B. S. Paul, and W. Doorsamy, "Development of a Stock Price Prediction Framework for Intelligent Media and Technical Analysis," Applied Sciences, vol. 12, no. 2, 2022, doi: 10.3390/app12020719.
P. Sevastjanov, K. Kaczmarek, and L. Rutkowski, "A multi-model approach to the development of algorithmic trading systems for the Forex market," Expert Systems with Applications, vol. 236, p. 121310, 2024, doi: 10.1016/j.eswa.2023.121310.
A. Khan, "A System Dynamics Model Of Exchange Rate Determination And Forecasting," SEISENSE Journal of Management, vol. 3, no. 4, pp. 44-55, 2020, doi: 10.33215/sjom.v3i4.367.
J. Bae, "Predictive Powers of Carbon Emissions and Oil Price on Stock and Foreign Exchange (FX) Markets Using Multivariate Recurrent Neural Networks (RNN) Model," Journal of Global Research in Education and Social Science, vol. 18, no. 4, pp. 112-121, 2024, doi: 10.56557/jogress/2024/v18i48943.
M. S. Islam and E. Hossain, "Foreign exchange currency rate prediction using a GRU-LSTM hybrid network," Soft Computing Letters, vol. 3, p. 100009, 2021, doi: 10.1016/j.socl.2020.100009.
G. Jung and S. Y. Choi, "Forecasting foreign exchange volatility using deep learning autoencoder-LSTM techniques," Complexity, vol. 2021, 2021, doi: 10.1155/2021/6647534.
A. A. Baasher and M. W. Fakhr, "Forex trend classification using machine learning techniques," 2011, pp. 41-47. [Online]. Available: https://www.researchgate.net/profile/Mohamed-Fakhr-3/publication/303080330_Forex_trend_classification_using_machine_learning_techniques/links/59f19896a6fdcc1dc7b90bff/Forex-trend-classification-using-machine-learning-techniques.pdf.
S. Enkhbayar and R. Ślepaczuk, "Predictive modeling of foreign exchange trading signals using machine learning techniques," Expert Systems with Applications, vol. 285, p. 127729, 2025, doi: 10.1016/j.eswa.2025.127729.
L. Ni, Y. Li, X. Wang, J. Zhang, J. Yu, and C. Qi, "Forecasting of Forex Time Series Data Based on Deep Learning," 2019, vol. 147, pp. 647-652, doi: 10.1016/j.procs.2019.01.189.
T. E. Clark and K. D. West, "Using out-of-sample mean squared prediction errors to test the martingale difference hypothesis," Journal of Econometrics, vol. 135, no. 1, pp. 155-186, 2006, doi: 10.1016/j.jeconom.2005.07.014.
G. Wang, T. Tao, J. Ma, H. Li, H. Fu, and Y. Chu, "An improved ensemble learning method for exchange rate forecasting based on complementary effect of shallow and deep features," Expert Systems with Applications, vol. 184, p. 115569, 2021, doi: 10.1016/j.eswa.2021.115569.
A. Ismailov and B. Rossi, "Uncertainty and deviations from uncovered interest rate parity," Journal of International Money and Finance, vol. 88, pp. 242-259, 2018, doi: 10.1016/j.jimonfin.2017.07.012.
N. H. Zainuddin et al., "Improvement of time forecasting models using a novel hybridization of bootstrap and double bootstrap artificial neural networks," Applied Soft Computing, vol. 84, p. 105676, 2019, doi: 10.1016/j.asoc.2019.105676.
P. Escudero, W. Alcocer, and J. Paredes, "Recurrent neural networks and ARIMA models for euro/dollar exchange rate forecasting," Applied Sciences (Switzerland), vol. 11, no. 12, 2021, doi: 10.3390/app11125658.
A. Sadeghi, A. Daneshvar, and M. Madanchi Zaj, "Combined ensemble multi-class SVM and fuzzy NSGA-II for trend forecasting and trading in Forex markets," Expert Systems with Applications, vol. 185, p. 115566, 2021, doi: 10.1016/j.eswa.2021.115566.
J. C. Garza Sepúlveda, F. Lopez-Irarragorri, and S. E. Schaeffer, "Forecasting Forex Trend Indicators with Fuzzy Rough Sets," Computational Economics, vol. 62, no. 1, pp. 229-287, 2023, doi: 10.1007/s10614-022-10281-3.
V. Vajda, P. Kisela, and M. Užik, "The Impact of Fundamental Information on EUR Currency - Evidence from the Market," 2015, vol. 32, pp. 87-94, doi: 10.1016/S2212-5671(15)01368-4.
P. Juszczuk and L. Kruś, "Soft multicriteria computing supporting decisions on the Forex market," Applied Soft Computing, vol. 96, p. 106654, 2020, doi: 10.1016/j.asoc.2020.106654.
E. Dockery and I. Todorov, "Further evidence on the returns to technical trading rules: Insights from fourteen currencies," Journal of Multinational Financial Management, vol. 69, p. 100808, 2023, doi: 10.1016/j.mulfin.2023.100808.
C. J. Neely, P. A. Weller, and J. M. Ulrich, "The Adaptive Markets Hypothesis: Evidence from the Foreign Exchange Market," Journal of Financial and Quantitative Analysis, vol. 44, no. 2, pp. 467-488, 2009, doi: 10.1017/S0022109009090103.
T. Ding, D. Li, and R. Sun, "Suboptimal Local Minima Exist for Wide Neural Networks with Smooth Activations," Mathematics of Operations Research, vol. 47, no. 4, pp. 2784-2814, 2022, doi: 10.1287/moor.2021.1228.
H. Ince and T. B. Trafalis, "A hybrid model for exchange rate prediction," Decision Support Systems, vol. 42, no. 2, pp. 1054-1062, 2006, doi: 10.1016/j.dss.2005.09.001.
K. Jazayeri, "Predicting multi-horizon currency exchange rates using a stacked ensemble of random forest and SVR," Intelligent Decision Technologies, vol. 18, no. 1, pp. 297-325, 2024, doi: 10.3233/IDT-230194.
T. K. Ho, "Random decision forests," 1995, vol. 1, pp. 278-282, doi: 10.1109/ICDAR.1995.598994.
T. Chen and C. Guestrin, "XGBoost: A scalable tree boosting system," 2016, pp. 785-794, doi: 10.1145/2939672.2939785.
K. Fataliyev and W. Liu, "MCASP: multi-modal cross attention network for stock market prediction," in Proceedings of the 21st Annual Workshop of the Australasian Language Technology Association, 2023, pp. 67-77. [Online]. Available: https://aclanthology.org/2023.alta-1.7/.
H. N. Bhandari, B. Rimal, N. R. Pokhrel, R. Rimal, K. R. Dahal, and R. K. C. Khatri, "Predicting stock market index using LSTM," Machine Learning with Applications, vol. 9, p. 100320, 2022, doi: 10.1016/j.mlwa.2022.100320.
J. Li et al., "Feature Selection: A Data Perspective," ACM Comput. Surv., vol. 50, no. 6, 2017, doi: 10.1145/3136625.
M. Argoty-Erazo, A. Blázquez-Zaballos, C. A. Argoty-Eraso, L. L. Lorente-Leyva, N. N. Sánchez-Pozo, and D. H. Peluffo-Ordóñez, "A Novel Linear-Model-Based Methodology for Predicting the Directional Movement of the Euro-Dollar Exchange Rate," IEEE Access, vol. 11, pp. 67249-67284, 2023, doi: 10.1109/ACCESS.2023.3285082.
A. Graves and J. Schmidhuber, "Framewise phoneme classification with bidirectional LSTM and other neural network architectures," Neural Networks, vol. 18, no. 5, pp. 602-610, 2005, doi: 10.1016/j.neunet.2005.06.042.
A. Vaswani et al., "Attention Is All You Need," 2017. [Online]. Available: https://proceedings.neurips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html.
A. Pornwattanavichai, S. Maneeroj, and S. Boonsiri, "BERTFOREX: Cascading Model for Forex Market Forecasting Using Fundamental and Technical Indicator Data Based on BERT," IEEE Access, vol. 10, pp. 23425-23437, 2022, doi: 10.1109/ACCESS.2022.3152152.
M. Ulina, R. Purba, and A. Halim, "Foreign Exchange Prediction using CEEMDAN and Improved FA-LSTM," 2020, doi: 10.1109/ICIC50835.2020.9288615.
A. Saghafi, M. Bagherian, and F. Shokoohi, "Forecasting Forex EUR/USD Closing Prices Using a Dual-Input Deep Learning Model with Technical and Fundamental Indicators," Mathematics, vol. 13, no. 9, 2025, doi: 10.3390/math13091472.
L. Qi, M. Khushi, and J. Poon, "Event-Driven LSTM For Forex Price Prediction," 2021, doi: 10.1109/CSDE50874.2020.9411540.
H. Yang, "Multimodal Stock Price Forecasting Using Attention Mechanism Based on Multi-Task Learning," 2024, vol. 14332 LNCS, pp. 454-468, doi: 10.1007/978-981-97-2390-4_31.
R. P. De Silva and H. Pathberiya, "Directional Forecast with Dynamic Volatility and Time Regime Classification: An Evaluation on EUR/USD," 2022, doi: 10.1109/SLAAI-ICAI56923.2022.10002484.
Downloads
Published
Submitted
Revised
Accepted
Issue
Section
License
Copyright (c) 2025 Nima Heidari (Author); Saeed Mirzamohammadi (Corresponding author); Babak Amiri (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
