ABSTRACT The growing use of communication networks in radio electric spectrum applications has reached a saturation point. Spectral prediction has gained importance, as several models have been studied to improve the accuracy of predicted traffic compared to real traffic. This research aims to generate a spectrum prediction method within a radio electric space. The obtained traffic is generated with the MFHW algorithm, which creates traces based on parameters given by the user: mean, Hurst parameter and spectrum width. The results revealed marginal errors of 0.04% for the Hurst parameter and 0.085% for the estimation of the multifractal spectrum width. The MFHW algorithm builds traces with self-similarity, long-range dependence (LRD), and multifractal characteristics, inherent to the behavior of the radioelectric spectrum. Hence, the method discussed in this paper can establish a modeling framework for the generation of traces applied to spectrum prediction based on multifractal analysis.