0
$\begingroup$

I am looking for a procedure to compute an exponential weighting of returns given a half life parameter.

I ran accross a wikipedia article, can I take it unchanged an assume N(t) is the return at time t ?

$\endgroup$
  • $\begingroup$ I don't think of return as having an exponential half-life. But if you mean return projections then we're in business. This would be more sophisticated than the rules of thumb I model with, but we're trying to capture similar concepts. $\endgroup$ – Nathan S. Mar 8 '15 at 22:38
  • $\begingroup$ Please rephrase your title as a question. $\endgroup$ – SRKX Mar 10 '15 at 7:54
0
$\begingroup$

What you want to do sounds like exponential smoothing of returns. So you want to forecast a return by exponentially weighting recent returns. For exponential smoothing you can look at Hyndman's papers and the e-book or this course on page 34.

I personally don't think that this will give a good forecast! It would be much too easy by the way.

The link that you provide points to exponential decay. We had this in school when radio-active material got less and less. I would not just replace $N(t)$ in the wikipedia article but rather read about statistics e.g. using Hyndman's ressources (I know I cite him a lot ...).

| improve this answer | |
$\endgroup$

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.