The complexity of financial time series is an important issue for nonlinear dynamic systems. Generalized power spectrum cumulative Tsallis entropy (PSCTE) is a newly proposed model for measuring dissimilarities between different time series. It solves the problem of traditional Shannon entropy inconsistency. In addition, the power spectrum is used to calculate the probability in the algorithm. In this paper, PSCTE is applied to simulation data sets, and financial time series are used to verify PSCTE reliability. The results show that PSCTE can be worked as an effective tool to measure dissimilarities and help identify signal patterns. Finally, we also obtain the geographical division of the stock market.