There is a very strong indication that the model errors are not white. Since you have not built a noise model, this is O

[fyousefi]

There is a very strong indication that the model errors are not white. Since you have not built a noise model, this is OK. However the model estimate might be improved if you also include a model of the noise. (Set 'DisturbanceModel' to 'Estimate' for state-space models, or use BJ or ARMAX for polynomial models.)