I have a somewhat related question, which could also provide an answer to another unanswered question of mine: TimeSeriesModelFit // (G)ARCH estimating methods
Hopefully this thread will be revised by someone after so long time.
Below I have uploaded a notebook where I try to find 103 parameters.
Excel's Solver can do the maximization with less parameters relatively easy (I didn't try with so many), but of course there is a precision concern. MaxBFGS from OxMetrics, on the other hand, does it reasonably fast.
FindMaximum in Mathematica, however, takes years...
Hi mam what about simulation of likelihood eatimates and their variances?