EUVIMEDEUVIMED entdecken
Crossref

Comparative Analysis of Cryptocurrency Market Behavior in Iran and the United States Using Hybrid Deep Learning Models and Indigenous Trading Strategy Design

Alireza Zamanian

Accounting, Finance and Computational Intelligence · 2026

DOI: 10.61838/jafci.264 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Accounting, Finance and Computational Intelligence, 2026. Verfasst von Alireza Zamanian.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study aims to develop a spatial–deep learning hybrid model to predict cryptocurrency returns and compare the behavior of the Iranian and U.S. crypto markets in order to design localized trading strategies. This applied–developmental quantitative study analyzed daily data of 32 leading cryptocurrencies from 2018 to 2024. The model architecture integrated: (1) spatial econometrics to capture contagion and network effects; (2) advanced deep learning models such as Graph Neural Networks, CNN–LSTM with attention, and Transformers to detect nonlinear temporal patterns; and (3) deep reinforcement learning for trading strategy optimization. Model performance was assessed using the Diebold–Mariano test and benchmarked against eight comparative models. The proposed spatial–deep hybrid model significantly outperformed both traditional econometric and single-component deep learning models. In the U.S. market, on-chain metrics and technical indicators were dominant predictors of price behavior, while in Iran, macroeconomic factors such as unofficial exchange rates and inflation played a key role. The hybrid model achieved an annualized return of 67.89% with a Sharpe ratio of 1.89 in the U.S. and 51.23% with a Sharpe ratio of 1.48 in Iran. By integrating spatial dependencies and nonlinear temporal dynamics, the model enhanced both predictive accuracy and economic interpretability. The clear divergence between the two markets underscores the vital influence of institutional and macroeconomic environments on crypto asset behavior. The results challenge the weak-form Efficient Market Hypothesis while supporting the Adaptive Markets Hypothesis, offering a novel analytical and strategic framework for digital asset investment.

Zitieren

APA

Zamanian, A. (2026). Comparative Analysis of Cryptocurrency Market Behavior in Iran and the United States Using Hybrid Deep Learning Models and Indigenous Trading Strategy Design. Accounting, Finance and Computational Intelligence. https://doi.org/10.61838/jafci.264

Vancouver

Zamanian A. Comparative Analysis of Cryptocurrency Market Behavior in Iran and the United States Using Hybrid Deep Learning Models and Indigenous Trading Strategy Design. Accounting, Finance and Computational Intelligence. 2026. doi:10.61838/jafci.264

BibTeX
@article{Zamanian2026,
  author = {Zamanian, Alireza},
  title = {Comparative Analysis of Cryptocurrency Market Behavior in Iran and the United States Using Hybrid Deep Learning Models and Indigenous Trading Strategy Design},
  journal = {Accounting, Finance and Computational Intelligence},
  year = {2026},
  doi = {10.61838/jafci.264}
}
Kostenlos starten →