July 7th, 2012, 4:15 pm
In my experience, problems with MC's and "sqrt" models are typically due to repeated hits on V=0.So, I would first try to get your scheme working well for relatively short time horizons with the Feller condition(s) satisfied (so nohits in the continuum limit) and forget that 'martingale correction'. The 'full truncation' method should work ok.Then, try cases when the Feller condition(s) are violated but T < 3, say, again without the martingale correction.If your real interest lies in setups more difficult than this, MC may not be the way to go at all, and youshould probably elaborate on what you are trying to accomplish.
Last edited by
Alan on July 6th, 2012, 10:00 pm, edited 1 time in total.