The complexity of financial time series is an important issue for nonlinear dynamic systems. We propose multiscale power spectral entropy. Based on this method, this paper uses the complex entropy causal plane ( ) to evaluate the complexity of the stock market. Multiscale power spectral entropy takes full advantage of the interrelationships between data in state space and estimates system complexity from different temporal resolutions. Then, we use a complex causal entropy plane to track changes in stock signals. The simulation data are used to test the performance of this method. Finally, we compare the method with the traditional power spectral entropy method. The results show that the method is more sensitive to changes in the stock market and can fully extract the intrinsic dynamics of the stock sequence.