In this paper we consider a class of conditionally Gaussian state space models and discuss how they can provide a flexible and fairly simple tool for modelling financial time series, even in presence of different components in the series, or of stochastic volatility. Estimation can be computed...