0
$\begingroup$

I have two questions. The first one is related with the ACF and PACF and the second with the Resulted Plot after using sarima.for I have downloaded the bitcoin price for the last 3 years as an example to learn how to do time-series analysis.

Then as you can see I took the first difference and the lag in order to detrend and remove variance.

enter image description here

How can I interpret the following case?

This is my model:

fit2 <- sarima(Ret.Bit,2,1,2)

with

> $ttable
         Estimate     SE  t.value p.value
ar1       -0.7751 0.0221 -35.1251  0.0000
ar2       -0.9381 0.0175 -53.7412  0.0000
ma1        0.7410 0.0158  46.9190  0.0000
ma2        0.9672 0.0134  72.1050  0.0000
constant   2.4933 1.3152   1.8957  0.0582

Using sarima.for(Ret.Bit,n.ahead = 4,2,1,2 I ve got this strange plot. enter image description here

For me, it is strange both the dates, 1560 etc, and the fact that going out of the plot the line. Please, any help suggestion?

$\endgroup$
  • $\begingroup$ The "strange plot" bit is a question about software and off-topic here, and it is not answerable because you did not provide a reproducible example (including describing which package sarima.for is from, and the type of the variable Ret.Bit, which I'm guessing is wrong). $\endgroup$ – Chris Haug Oct 11 '17 at 17:36
0
$\begingroup$

if the acf of the original series = acf of the residual series your model has self-cancelling structure thus equivalent to a (0,0,0)(0,0,0,) . This is a great example of using VERY BAD software to "identify" a model. Sometimes software is worth what you pay for it. Notice how the ar coefficients are eerily similar to the ma coefficients.

$\endgroup$
  • $\begingroup$ The orders aren't identified by the software, they were supplied in the function call. $\endgroup$ – Chris Haug Oct 11 '17 at 17:31
  • $\begingroup$ @Chris Haug ..Thanks for the correction. My faith in free software has been (partially !) restored. It would be interesting to me if the OP explained how they assumed the model or just saw it in a book/post somewhere and decided to go with it without understanding the consequences. $\endgroup$ – IrishStat Oct 11 '17 at 18:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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