wilmott.com

hey,I implemented a code to price exotic options in a Heston-Stochastic-Volatility, Merton-Jump-Diffusion and Bates model by means of Monte Carlo simulation. Does anybody know a good source to compare my results to? The goal for this exercise is to show that different models can agree more or less on the prices of vanillas but totally disagree on the prices of exotics (e.g. Heston-Stochastic-Volatility vs. Bates model). Obviously, the more complex the payoff of the exotic the more pronounced should this effect be. Therefore, I'm currently considering down-and-out puts. Does anybody have experience with which payoff I can generate the greatest discrepancies?thanks, bernd

Well, just thinking out loud, you could probably make the (percentage) discrepancy arbitrarily large = infinity.Say I invent an up-barrier contract that pays S(T1) - B, where T1 is the first instant that S(t) crosses the barrier B.Any diffusion (say Heston) will have the value zero and any process with up-jumps will have a positive value.

thanks Alan but I was thinking about something that actually exists (and is commonly known).As I see it for some barrier option with a short horizon and a distant barrier I will get the greatest differences between stochastic vol and jump-diffusion models. Any other suggestions?

Various cliquet-type structures (heavily marketed for a while by life insurance companies) are notoriously sensitive tostochastic vol., so maybe those are an example in the other direction -- where it is not the jumps driving the discrepancies.

QuoteOriginally posted by: AlanVarious cliquet-type structures (heavily marketed for a while by life insurance companies) are notoriously sensitive tostochastic vol., so maybe those are an example in the other direction -- where it is not the jumps driving the discrepancies.They still are.....A 2 or 3 asset rainbow option would be a decent one as long as your willing to calibrate 2 or 3 different vol skews

you might want to implement a local vol montecarlo ( and derive your local vol from the respective model)as an additional comparison...that would ensure you agree perfectly with european vol surface, and would expose you to forward skew diffs....as relevant for cliquetsother than that i would agree - looking at short horizon distant barrier... but its probably important to check you are calibrating to suitable short end vols- ie that imply a blow up in your vol of vol? if you don't use jumps

"A Perfect Calibration! Now What?"Wim Schoutens