Dear  Visit
 
The  results you have obtained are usually caused by regressing the lagged return  instead of the leading return. That is you must regress the return of the period  following the log price against which you regressing:
 
dx =  x(t+dt) - x(t) against x(t)
 
Hope  this helps.
 
Regards
 
Les.
-----Original Message-----
From: visit_thailand@hotmail.com  [mailto:visit_thailand@hotmail.com]
Sent: Friday, November 02, 2001  14:02
To: Vince.J.Kaminski@enron.com;  chris_strickland@compuserve.com; les_clewlow@compuserve.com;  contact@lacimagroup.com
Subject: Please advice on the estimate of  mean reversion rate for electricity price
Dear All,
My mane is Visit Phunnarungsi. I used to e-mail Vince Kaminski about the  advice on his article "The Challenge of Pricing and Risk Managing Electricity  Derivatives" and he had mailed me the copy.
I am now modelling the Queensland electricity spot price using Geometric  Brownian Mean Reverting Jump Diffusion Model and have followed your paper  "Making the most of mean reversion" to estimate the mean reversion speed. I  use Queensland half-hourly price during 13 December, 1998-30 June 2001 giving  about 44,000 price observations. 
However, the result from Ordinary Least Squares was not as expected due to  different sign for both slope & intercept. The coefficient and standard  error are as followed:
Intercept: -0.3931 (0.0076)
Slope: 0.1171 (0.0022)
R Square: 0.0585
Therefore I could not estimate the mean reversion rate as the estimated  slope has the positive sign. I have also tried monthly data and the results  are the same. It would be appreciated if you could advice me on this  matter.
Kindest regards,
Visit

Get your FREE download of MSN Explorer at http://explorer.msn.com