Calibration of Ornstein-Uhlenbeck process

[Maurelius]

Hello EViews forum!

I'm trying to calculate the half-life of mean reversion of daily stockprices by the formula of T1/2 = ln(2)/k, where k is the speed of mean reversion.

But to solve for k I need to do a calibration/estimation of the Ornstein-Uhlenbeck by the Least Sqare method.

I tried following the steps presented here: http://fxpaul.wordpress.com/2011/05/26/ ... k-process/

I just don't know how to do it in practice. Would anybody care to explain thoroughly? I don't know how to solve for the parameters, i.e. where do I get deltaT, a and b?

Can I do a regression first in Excel and then use the parameters of a and b to solve for the other parameters?

Best regards,

Maurelius

[trubador]

The reference you cited is quite explanatory. If you had searched the forum, you would have found the following post on mean reversion: http://forums.eviews.com/viewtopic.php? ... 78&p=10789
Since you are using regularly spaced data, delta t is equal to 1 (or one day to be more specific).

If you need further help for estimating OLS or AR models in EViews, please see EViews Illustrated and manuals in the Help system. Not to mention the search facility of the forum...