are there some tests which are usually performed in order to ensure that a calibration procedure is sound? That is, say you have a model depending on N parameters and one tries to fit them in order to match M market observables. Are there common "market practice" tests which are usually done in order to ensure the stability of the calibration (besides maybe trivial ones like shifting the market datas or dependence on the inital guess...)?


  • $\begingroup$ Hi Profolio, and welcome to SE QF! Could you please be a little bit more specific as to what you are trying to calibrate / fit? The first thing that comes to my mind with your question, generally speaking, is to make sure that calibrated prices / quotes are within bid/ask spreads. But I have the feeling that this is not what you are looking for, no? $\endgroup$ Apr 17 '20 at 12:28
  • $\begingroup$ If you are asking about statistical models, there is a lot of statistical test. In case of linear regression, for example tests if model parameters are zero or not. There is also goodness of fit test ($\chi^2$) to test quality of distribution fitting. $\endgroup$ Apr 17 '20 at 13:38
  • $\begingroup$ Another way to validate your model is to run out-of-sample tests of your calibrated model to see if the performance is comparable or if you suffer from over-/underfitting. $\endgroup$
    – Andreas
    Apr 17 '20 at 16:00

Your Answer

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

Browse other questions tagged or ask your own question.