I’m trying to implement this paper. It is Nelson Siegel with some extra features, but my question is about Kalman Filter which is used for estimation NS model. On the page 11 there is state equation 14, which contains moving average part. How prediction and update formulas looks like in that case? Didn’t find anything in a literature about that. ARMA models look a like, but they have MA part in observation equation and not in state equation.

I’m trying to implement a Kalman filter on a noisy sine signal, but I’m getting a phase lag on the filtered output which is also higher in amplitude.