Is it possible to generate multiple (e.g. 3) time series, each with predefined volatility, and together with an predefined correlation matrix , plus, one of the time series has a predefined autocorrelation ? If so, how to do that?

  • 1
    $\begingroup$ Maybe you are seeking something along the lines of this post? $\endgroup$ – javlacalle Jan 10 '17 at 7:39

Have a look here. General idea is to generate normal random variables with give covariance matrix.

If Z is a vector of uncorrelated Gaussian random variables, and C is a square root of the given covariance matrix, then target variable will be mean+CZ.

To obtain C You may use Cholesky decomposition.

If You may obtain Gaussian variable with a given covariance matrix, then You may convert it to differently distributed variable using inverse CDF method.

Important to notice, that CDF of any target vector is uniformly distributed, than You just map your CDF to a desired. Finally You have random variable from any distribution wich has given covariance matrix.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.