1
$\begingroup$

I am trying to see if and how the news for affects the financial markets. I have a time-series for both of them.

Should I standardise the series? I have a monthly return on prices from the Dow and a monthly score for the sentiment, is that enough? I already tried to get the Standard Deviation for each month by using Excel and highlighting the days of each month (2000 - 2014) but it's taking too long (=STDEV.S(B24:B42)), so I'm wondering if its worth doing and if so, is there a faster way of doing this?

Thanks.

$\endgroup$
  • 1
    $\begingroup$ This is pretty broad, or at least parts of it are. You may or may not get a complete answer & some people may even vote to close. Thus, you may want to think about asking a series of concrete, focused questions. You can ask a new one as each is answered, & you could link between them for context. Note that tutorials & code help are off-topic. $\endgroup$ – gung - Reinstate Monica Mar 31 '15 at 17:43
  • $\begingroup$ @gung Thanks, I have removed the less important part of the questions. Please, see the edit. $\endgroup$ – Adegoke A Mar 31 '15 at 17:47
  • 1
    $\begingroup$ You may want to roll that back. This is now definitely too broad & a good candidate for closure. The part of your initial question that looked best was asking if you needed to standardize the time series somehow. Just asking 'how to build... model' is not a good Q for CV. $\endgroup$ – gung - Reinstate Monica Mar 31 '15 at 17:50
  • 1
    $\begingroup$ That seems like it's probably a reasonable question. You may want to read some of our materials on asking questions here; you can find information in our help center. Since you're new here, you may also want to take our tour. $\endgroup$ – gung - Reinstate Monica Mar 31 '15 at 17:54
  • 1
    $\begingroup$ Forget about this. You need to learn a lot about stock markets before you even start applying statistics. For one, you don't work with prices in these situations, but with returns. $\endgroup$ – Aksakal Mar 31 '15 at 18:33
1
$\begingroup$

You should be aware that it is one thing to compute a correlation coefficient and yet another thing is how to test for independence or significance . Time series data does not qualify for standard textbook solutions which premise independent samples. http://empslocal.ex.ac.uk/people/staff/dbs202/cat/stats/corr.html . Another piece of literature which I wrote to illustrate the difference between cross-sectional data/tools and time series data/tools might help http://www.autobox.com/cms/index.php/afs-university/intro-to-forecasting?start=5 .It is entitled Regression vs Box-Jenkins.

$\endgroup$

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.