On Sep 13, 12:10 pm, Jerry W. Lewis
wrote:
I can't speak to financial convention, since I don't work in that area.

Fair enough. I am in a similar situation.


However, for most things that you would want to do with a standard
deviation, back-transforming the standard deviation itself would not
be useful.

Oh, I agree; that is, we would have to continually convert back and
forth. And I ass-u-me that financial software does the right thing.
I was just asking about the convention for the __input__, if there is
any.

I have used free online software (read: poorly documented) that asks
for "average return" and "volatility" or "standard deviation". I
would expect the two should be expressed in terms of the same "space",
be it the returns or the log returns.

And in the early days of my learning in this area (some years ago),
google searchs turned up lengthly explanations that converted the
average log return to the geometric mean of the return and similarly
for the standard deviation of the log returns.

(And of course, some software that I have used want the beta for
"volatility", which is something else entirely.)

If the world were a consistent place, the software would ask for
statistics about the "returns" or the "log returns" per se. That
would clear things up. But except for some educational web sites, I
have never seen the term "log return" used explicity -- for example,
never in sales literature that includes performance data.

Oh well, I suspect there is no canonical answer to my question. I
really just wanted to raise the visibility of the fact that the
definition "volalitity" as the sd of the log return, per se, is not
necessary so cut-and-dried. At least, that has been my experience.