arch matlab程序怎么写

%% coed_MS_ARCH_3S.txt

%-------------------------------------------

% Written by Zhu, Junjun,

% Ph.D candidate in School of Economics

% Fudan University, Shanghai 200433

% Nov. 2009

% please notice, we can not guarantee that these codes are without mistakes.

% You may use these at your own risks.

%-------------------------------------------

clc

clear

load data_200910weekex

y = data_200910weekex

n = size(y,1)

x = ones(n,1)

sn = 3% number of state

nk = 1% regressor number in mean equation

np = 3% p-term in GARCH

y_p = y(np+1:n)

x_p = x(np+1:n)

pi_pri = [18 2 22 28 22 2 16] % prior for pi

bols = inv(x'*x)*x'*y

s2ols = (y-x*bols)'*(y-x*bols)/(n-1)

xsquare=x'*x

indexSwitch =0

switch_1 = 0

switch_2 = 0

switch_3 = 0

P=[0.8 0.1 0.10.1 0.8 0.10.1 0.1 0.8]

%spec = garchset('distribution', 'T','p',0,'q',3)

%[coeff, errors,LLF,innovations,sigmas,summary] = garchfit(spec,y)

%garchdisp(coeff,errors)

% coeff.C = [0.0005]coeff.K = 0.00015coeff.ARCH = [0.197,0.26,0.292]

% default model: y(t) = C +e(t)h(t) = K + GARCH*h(t-1) + ARCH*e(t-1)^2

% beta = [0.00160.00160.0016]

% a_1 = [0.000740.20.260.29]% ARCH coeff. a0, a1, a2, a3 for state 1

% a_2 = [0.000740.20.260.29]

% a_3 = [0.000740.20.260.29]

beta = [-0.00110.02-0.0061]

a_1 = [0.00050.110.150.08]

a_2 = [0.00140.220.170.13]

a_3 = [0.00220.140.100.15]

var_err = (y-x*beta(1)).^2

beta_ = []

a1_ = []

a2_ = []

a3_ = []

P_ = []

nburn=0%number of burnin replications

nrep=30000%number of retained replications

ntotal=nburn+nrep

对garch模型做预测可以用matlab自带的garchfit（）函数，该函数主要用于估计ARMAX / GARCH模型参数。

garchfit（）函数使用格式：

[Coeff,Errors,LLF,Innovations,Sigmas,Summary] = garchfit(Spec,Series,X)

Coeff——输入参数。接受由garchset，garchget，garchsim，garchinfer，和garchpred结构产生的参数。

Errors——系数的估计误差（即标准误差）的结构。

LLF——对于优化目标函数值与参数相关的估计发现Coeff。garchfit执行优化使用优化工具箱fmincon函数。

Innovations——创建（即残差）序列推导的时间序列列向量。

Sigmas——与创建相对应的条件标准偏差向量。

Summary——显示优化过程的摘要信息结构。

Spec——包含条件均值和方差规范的GARCH规范结构。它还包含估计所需的优化参数。通过调用garchset创建这个结构。

Series——观测的时间序列列向量。

X——观测数据的时间序列回归矩阵。

例如：

clc

spec = garchset('C',0,'K',0.0001,'GARCH',0.9,'ARCH',0.05) %指定模型的结构

[e,s,y]= garchsim(spec,1000)

[Coeff,Errors,LLF,Innovations,Sigmas,Summary] = garchfit(spec,y)  %拟合参数

运行后得到的部分结果

需要。

又称“广义ARCH模型（Generalized ARCH）”、“广义自回归条件异方差模型”

自从Engle(1982)提出ARCH模型分析时间序列的异方差性以后，

波勒斯列夫T.Bollerslev(1986)又提出了GARCH模型，GARCH模型是一个专门针对金融数据所量体订做的回归模型，除去和普通回归模型相同的之处，GARCH对误差的方差进行了进一步的建模。

---