MA(1) process without using built-in functions

Hi everybody, i'm trying to write a MA(1) process without using built-in functions (homework) and with these input parameters

• theta (MA parameter)
• sigma (standard deviation of innovations)
• epsilon0 (starting value)
• T (total length of the time series to simulate)

function y = simulateMA1(theta, sigma, epsilon0, T)
y(1)=epsilon0;                    

for k=2:T                       
    epsilon(k)=(sigma^2)*randn;             
    y(k)=epsilon(k)+theta*epsilon(k-1);     
end                             

y=y'                                       % shifting to column vector

end

I'm using the grader and no error message comes out, but there is something wrong that i'm missing with the model. Can anybody share his tought about this?

Thankss