Hi,
I am having trouble modelling a time series for monthly stock data.
The monthly returns have negative autocorrelation for a single lag. The correlogram then decays to zero.
However, the monthly returns also exhibit a positive autocorrelation to the 12 month total return (the total return over the last 12 months, not the 1 month return that happened 12 months ago) and a negative correlation with the 36 month return.
What is the best way to model this scenario? I have tried doing a linear regression, but it does not give a good fit:
Monthly return = A * [Previous Month Return] + B * [Previous 12 month return] + C * [Previous 36 month return] + ERROR TERM
The above model also fails to take into account the fact that the monthly return is also positively correlated to the total return over months 9-12, and negatively correlated to the total return over months 32-50. Though these multi-month returns are obviously highly correlated to the 12 and 36 month returns.
What sort of models should I be looking at to analyse this data more successfully?
Kind regards,
AS.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|