The Empirical Mode Decomposition (EMD) denoising and Multifractal Detrended Cross-Correlation Analysis (MF-DCCA) have been combined to address the shortcomings of traditional portfolio models,thus a multiple fractal portfolio model under EMD denoising and MF-DCCA is proposed in this paper.The new model,abbreviated as EMD-Mean-MF-DCCA,is applied to a highly speculative cryptocurrency portfolio,which is then verified by out-of-sample test and analysis with rolling window technique.The results show that whether the cryptocurrency price is on upward or downward trend,the proposed EMD-Mean-MF-DCCA is significantly optimized in terms of profitability and Sharpe ratio compared with traditional portfolio models and non-denoised fractal portfolio models.Moreover,the portfolio investment scheme under the new model has better risk resistance capability when the price of cryptocurrency falls sharply.