0
$\begingroup$

I have economic calendar data since 2007 and currency indexes, I wanted to calculate wich average weight does news have in currency strength/weakness. I was thinking about using PCA, does anyone has a point of view?

$\endgroup$
1
  • 1
    $\begingroup$ Is your currency data daily or intraday? Generally you would define a short time window around each announcement and measure the square (or absolute value of) the price change over this interval. This will tell you which announcement type has the most impact. $\endgroup$
    – noob2
    Aug 25 at 12:26
1
$\begingroup$

This is not an answer, just a long comment.

I believe such general questions need to have a background and more justification. For instance, which methods have you considered and why you think PCA is better than an event analysis? PCA reduces dimensionality (lots of variables to less principal components), so how do you rationalize its use? What is the standard in the literature - is it worth it building on it? Which news will you consider and why? One could say GDP growth rate YoY forecasts (or, say, their difference from actual values) have more explanatory power than Industrial Production YoY, but (s)he needs to justify it. Since this seems like a school essay, the story is very important. Also, intuition says that markets react more to unexpected news, perhaps make a proxy of an "expectation index" to attach to each news data point.

$\endgroup$
4
  • 1
    $\begingroup$ Thank you very much Sir, you gave me an incredible place to start. $\endgroup$ Aug 26 at 9:41
  • $\begingroup$ @user9085964 Wow, I was never expecting this to be accepted as an answer, because it isn't, but on the other hand the question seems too general to me, but I'm not expert on PCA by any means... haha thanks. So since you don't reject my hypothesis that this is a school paper, I have something more to add in case it is indeed. Lots of people including me, often start(ed) a hunting/worship relation with a particular methodology. Writing a school project most of the times just means: $\endgroup$ Aug 27 at 10:34
  • $\begingroup$ 1. Take the reader by the hand with simple concrete step-by-step writing, ("imagine that your paper is address to your fellow colleauges" as the best mentor of my life told me). 2. Emphasize on the story and intuition, ideally finding a literature gap. 3. Do methodologically whatever the literature considers standard. Avoid out-of-place exotic methodologies (i'm not implying PCA is one, i'm not familiar with the literature). Methodologies are just a tool. Most of the paper's creativity, as human-beings, lies in the story set-up! $\endgroup$ Aug 27 at 10:34
  • $\begingroup$ Then, if time permits Explore some of the new questions which will surely be created by your journey. If you really want to try sth exotic go for it. The above is just a personal opinion. $\endgroup$ Aug 27 at 10:34

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.