Hi,I'm running an EKF on randomly generated data (using the code in the book "Extended Volatility Arbitrage", A.Javaheri), without the MLE part yet, and constantly have trouble with things such as negative x1 estimates (which lead to -ve square roots), or 0 weighting parameters which lead to div by 0 etc.Does this happen often, is the only solution to tweak the parameters around?I'm using data generated with:mu = 0.6omega = 0.41theta = 15xi = 0.09rho = -0.5With the same parameter values in the filter (and initial guesses x0 = 0.04 and P0 = 10^-6)Dunno if anyone can suggest help ? Thanks

This seems to be a pretty old post...but I am now having exactly the same problem. Intrigued if you got anywhere with this (& if you can still remember)!