Let's say I have time series of $n$ (simulated) data from the GARCH(1,1) process: $$X_{t+1,n} = \sigma_{t+1,n}(...)*Z_{t+1,n}$$ where $Z_{t,n}$ is iid student t-distributed with $v$ degree of freedom.
I want to create a distributional forecast for $X_{t+1,n}$. When working with $Z$ normally distributed I simply let $X_{t+1,n}$ have the conditional distribution: $N(0,\sigma_{t,n}^2)$.

Is it different when the distribution of $Z$ is student-t? In general, how do I create distributional forecast for $X_{t+1,n}$ when $(X_{1,n},...X_{t,n},\sigma_{1,n},...,\sigma_{t,n}$) is known?

(my purpose is application so a theoretical derivation is not necessary)

  • $\begingroup$ I assumes $\sigma^2$ is a typo? You mean $X_{t+1}=\sigma_t(..)*Z_t$ $\endgroup$
    – k.dkhk
    Commented Nov 5, 2017 at 13:49
  • $\begingroup$ I believe it is standard notation to have $X_{t+1} = \sigma_{t+1}^2 Z_{t+1}$, where $\sigma_{t+1}^2 = \omega + \alpha \varepsilon_t^2 + \beta \sigma_t^2$. $\endgroup$ Commented Jan 20, 2018 at 17:43
  • $\begingroup$ @JohanStaxJakobsen you mean $X_{t+1} = \sigma_{t+1} Z_{t+1}$ and $\sigma^2_{t+1} = \omega + \alpha X_{t}^2 + \beta \sigma^2_t$ $\endgroup$
    – Taylor
    Commented Jan 20, 2018 at 19:01
  • $\begingroup$ @Taylor Yes - of course. Sorry. $\endgroup$ Commented Jan 20, 2018 at 19:05

1 Answer 1


The distributional forecast of $X_{t+1,n}$ is a non-standardized Student's t-distribution with location parameter $\mu = 0$ and scale parameter $\sigma_{t+1,n}$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.