Too many input arguments

Hi,

Im using a code given below ab=nd it gives the following error.

Can someone please help me with it?

thanks

ERROR:

Error using gtest
Too many output arguments.

Error in test (line 37)
beta = gtest(Rt);

CODE:

   clc;
  clear;

% This program is for T=1000

T=1000;
B=1000;
R=1000;
ahead=10;

%generation the initial GARCH(1,1) data set.

a0 = 0.05; a1 = 0.1; b1 = 0.85;

epsi=zeros(3000,1);
simsig=zeros(3000,1);

simsig(1) = a0 ;
s=(simsig(1))^0.5;
epsi(1) = trnd(5,1) * s;

   for i = 2:3000

        simsig(i) = a0+ a1*(epsi(i-1))^2+ b1*simsig(i-1);
        s=(simsig(i))^0.5;
        epsi(i) = trnd(5,1) * s;

   end

yt1=epsi.^2;

ytlast=epsi(3000)^2;
htlast=simsig(3000);
Rt=yt1(2001:3000);

beta = gtest(Rt);

    function gtest(Rt)
    %PURPOSE: estimates a GARCH(1,1) under the assumption that residuals are
    %Student-t distributed
    %Make sure that rtSP.mat is in the same location as this code.
    %--------------------------------------------------------------------------
    %USAGE : beta=test_ml5(Rt);
    %--------------------------------------------------------------------------
    %INPUT: Rt - Vector, the time series of returns (follow instruction on line
    %       15 to use your own vector of data)
    %--------------------------------------------------------------------------
    %RETURN: beta - vector, GARCH(1,1) parameters of the maximum likelihood (w,alpha,beta,nu)
    %--------------------------------------------------------------------------

    tic;

    m=mean(Rt);
    Rt=Rt-m;

    % parameters for GARCH: w, alpha, beta, nu
    beta0=[0.01; 0.1; 0.8; 4];
    lb=[0; 0.0001; 0.0001; 2 ]; 
    ub=[1; 0.3; 0.99; 30];
    A=[0 1 1 0]; %Ax<b
    b=1;

    [lb beta0 ub];

    options=optimset('Diagnostics','on','Display','iter');

    [beta,stderr1,vc,logl]=Max_lik(@likGARCH11T,beta0,'Hessian',A,b,[],[],lb,ub,[],options,Rt);
    [beta,stderr2,vc,logl]=Max_lik(@likGARCH11T,beta0,'Sandwich',A,b,[],[],lb,ub,[],options,Rt);

    [stderr1 stderr2];

    toc
    function l=likGARCH11T(b,rt)
    % Computes the likelihood vector of a GARCH11
    % presumes that Rt is a vector of centered returns 

    w     =  b(1); 
    alpha =  b(2); 
    beta  =  b(3);
    nu    =  b(4);

    rt2   = rt.^2;
    [T,K] = size(rt2); 
    ht    = zeros(T,1);
    ht(1) = sum(rt2)/T;

    for i=2:T
        ht(i) = w + alpha*rt2(i-1) + beta * ht(i-1);
    end

    sqrtht  = sqrt(ht);
    x       = rt./sqrtht;

    %*** Plug in Student-t

    np1= (nu+1)/2;
    nh = nu/2;
    l1 = gammaln(np1);
    l2 = gammaln(nh);
    l3 = log(sqrtht);
    l4 = 0.5*log(nu*pi);
    l5 = np1*log(1 + (1/nu) * (x.^2));

    l=l1-l2-l3-l4-l5;

Answer 1

[Walter Roberson]

You have coded

function gtest(Rt)

That declares that gtest() takes one item of input and returns nothing. Your code that calls on gtest() believes that gtest() returns something. You will need to decide which part of the code is correct.

[dav]

Thanks.

I had a little error in generation Rt vector and I corrected it (edited above code).

Now Rt is a vector of values.

So function gtest(Rt) gets that vector and returns "beta"

is it right?

I still get the same error

[Walter Roberson]

You still have

function gtest(Rt)

which is for a routine that does not return anything.

[dav]

I think I understand your point.

Is it okay to have function l=likGARCH11T(b,rt) inside gtest(Rt) or do I need to save it as a different function.

Thanks

[Walter Roberson]

If the only place that calls likGARCH11T() is the gtest() code, then it is fine to include the definition of likGARCH11T() after the definition of gtest() in the the same gtest.m file.

[dav]

thank you very much. Finally, I got this function gtest(Rt) code from an online post there, they had put a ";" right in front of function l=likGARCH11T(b,rt)

(function l=likGARCH11T(b,rt);)

is that necessary?

Again, thank you for all your help!

dav