Hey Guys,
Does anybody have a clue, in the chapter about limitations of historical estimates, what it means that long time periods create a temptation to use more frequent data, such as weekly data?
Kind regards
Is this related to Economic forecast challenges?
Yes
Using frequent data usually increases standard deviation. Lenghtening the measurement period is more desireable as it would increase Sharpe Ratio.
edit: nvm. My brain is fried, didnt realize you were talking about economics.
I think this is more related to regime change.
Way I understood is the longer time horizon has regime change issue, i.e. low inflation and hyper inflation.
So in that case longer time horizon data is non-stationary.
How would using weekly data help?
ok thanks for the inputs. I guess with a long time period in the series, you have an incentive to use more frequent data in your performance, because that increases the sharpe ratio (in the long-run). that would make sense to me. With monthly data, the volatility will be higher and the Sharpe ratio lower.
Ignore that sharpe ratio answer, it was in different context.
Back to your question, using daily data will have noise, stocks going in opposite direction, random movement etc. As a result correlation would be lower. Using monthly data however would “smoothen” these movements and higher correlation. Using higher frequency solves this problem of smoothing I guess.
If you talking about Non-repeating data pattern and time-period bias, then you are right. But this is not the limitation of history based estimation.
Correlation should be higher with the noise i thought.
Nope.