Cryptocurrencies, particularly Bitcoin, as digital assets based on blockchain technology, have increasingly attracted the attention of investors and financial researchers due to features such as decentralization, transaction transparency, and rapid transfer. However, this market is inherently subject to high volatility and is strongly influenced by economic, political, and technological factors. This challenge becomes even more critical in the context of Iran’s economic crisis, caused by international sanctions, currency fluctuations, and high inflation, which highlights the necessity of developing more accurate prediction models. The aim of this study is to assess the feasibility and evaluate the performance of metaheuristic algorithms in optimizing artificial neural networks for forecasting Bitcoin price trends. To this end, the hybrid Fire Hawk Optimizer–Giza Pyramids Construction (FHO–GPC) algorithm was compared with the Moth Ox Optimizer (MO), and the dataset was divided into training (80%) and testing (20%) subsets. The parameters of a multilayer perceptron (MLP) neural network were optimized using these algorithms; specifically, FHO performed the global search, while GPC carried out local optimization. The findings revealed that the MO algorithm, by significantly reducing error metrics (RMSE and MAE) and increasing the coefficient of determination (R²), delivered more accurate results compared to the hybrid model. This superiority was particularly evident under Iran’s volatile and crisis-driven economic conditions. Overall, the results suggest that the MO algorithm can serve as an effective approach for enhancing prediction accuracy and reducing investment risk in emerging and high-risk financial markets