How to determine Confidence Interval of each Parameter Estimation results of the GARCH model through fminsearch syntax?
6 次查看(过去 30 天)
显示 更早的评论
I doing the estimation of the parameters of the GARCH model through this data and syntax below where the rv is realized volatility and then I want to know the standars error and the P-value for each parameter estimation results but I think I must to know the confidence interval of these result parameter estimation first. I hope and I'm so glad if one can help me to solve this problem. Thank You.
Y65 = xlsread("Y65",1,"A2:A89");
n = length(Y65);
rv = [];
moving_window = 5;
for i = 1:n-moving_window+1
rv(i) = var(Y65(i:i+moving_window-1));
end
Y65_1 = Y65(moving_window:end)';
logL = @(theta) -(sum((log(1./(sqrt(2.*pi.*(theta(1)+theta(2).*...
Y65_1(1:end-1).^2+theta(3).*rv(1:end-1)))))-(1./2).*...
(Y65_1(2:end)./(sqrt(theta(1)+theta(2).*...
Y65_1(1:end-1).^2+theta(3).*rv(1:end-1)))).^2)));
theta_0 = [0.99 0.3 0.3];
[theta_hat NegLL] = fminsearch(logL, theta_0);
beta_0 = theta_hat(1)
beta_1 = theta_hat(2)
beta_2 = theta_hat(3)
-NegLL
0 个评论
回答(1 个)
Vanshika Vaishnav
2023-3-7
The 'fminsearch' function is a black box, unconstrained nonlinear solver. This means that no assumptions about the problem are made and 'fminsearch' is unaware that it is solving a curve-fitting problem. As a result, the solver returns only basic information. This means that confidence intervals will need to be calculated by the user. If you would like to use 'fminsearch' for curve fitting, a couple of resources on how to manually compute confidence intervals can be found here:
0 个评论
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 GARCH Model 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!