Model-based fuzzy time series clustering of conditional higher moments

This paper develops a new time series clustering procedure allowing for heteroskedasticity, non-normality and model's non-linearity. At this aim, we follow fuzzy approach. Specifically, considering Dynamic Conditional Score (DCS) model, propose to cluster according their estimated conditional moments via the Autocorrelation-based C-means (A-FCM) algorithm. The DCS parametric modeling is appealing because of its generality computational feasibility. usefulness proposed illustrated using an experiment with simulated data several empirical applications financial assuming both linear nonlinear models' specification under assumptions about density function.