enter image description here enter image description here

I took the residual of a historical stock price $\hat e_t=r_t-\hat \mu_t$, where $r_t$ is the return of a stock and ran ACF and PACF. From the ACF I think that the residual does not follow AR or MA process, and the PACF shows slight MA. But I am new to this and I am not sure if my interpretation is correct.

What do you think?

The Ljung Box test gave Q-statistic of 87.5597

enter image description here

which rejects the null that the autocorrelation coefficients are all zero. I used 40 lags here to be consistent with the ACF and PACF.

Does this contradict or confirm your intuition from visually inspecting ACF and PACF?

  • 1
    $\begingroup$ See + 1 2 3 4 5 $\endgroup$
    – Glen_b
    Feb 22, 2014 at 23:44
  • $\begingroup$ @Glen_b I just ran a Ljung-Box test (with 10 lags, as per the instructor asked) on STATA and the Q-statistic is 44.9364 which is significant, so I reject the null that all the autocorrelation coefficients are zero, and since 44.9364 is very significant, does that mean the residuals exhibit autocorrelation (at least up to 10 lags)? $\endgroup$
    – user95087
    Feb 23, 2014 at 0:22
  • $\begingroup$ Sounds like a different question; comments are not suitable for a series of questions and answers. Modify your question or post a new one. $\endgroup$
    – Glen_b
    Feb 23, 2014 at 0:24
  • $\begingroup$ I deleted my earlier comment; I had forgotten about the big spike in the AR at lag 7 (the plots are so large you can't look at them both at once - smaller plots are easier to look at simultaneously). $\endgroup$
    – Glen_b
    Feb 23, 2014 at 0:28
  • $\begingroup$ @Glen_b question updated, with 40 lags $\endgroup$
    – user95087
    Feb 23, 2014 at 0:29


Your Answer

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

Browse other questions tagged or ask your own question.