Page 1 of 1

Mean Square Error (MSE) for garch

Posted: Sun Dec 18, 2011 12:06 pm
by Nau2306
Hello

I am confused about how to calculate the Mean Square Error for a garch model. The formula for MSE is

MSE = 1/N ( summation i=1 to N (sigma hat - sigma)^2)

I am sorry about the formula. I dont know how to insert a formula on the forum :s, sigma = volatility

What is confusing me is what is the real meaning of sigma hat and sigma??

Is sigma hat the forecasted volatility? In that case what is sigma? If I am producing out of sample forecast, I do get the forecasted volatility, but what about sigma which in theory is the actual volatility, if I am not mistaken? What should I replace sigma with? What value does sigma takes?

Thanks