When to use Monte-Carlo?

[quotes]

In terms of European Option, worse than an expectation numerical integral.

In terms of Path-dependent exotic, looks good but with too little sample problem.

In terms of American Option, already other methods.

When is the application of MC better than other approaches?

 

[DoubleTrouble]

Well, given that Fubini's theorem doesn't apply and the dimension is greater than 3. Then I know of no other numerical integration technique apart from Monte Carlo that works well. The higher the dimension, the more favourable MC will be.