Quant - Monte Carlo

one of the major limitations of monte carlo simulation is that it:

1) cannot provide the insight that analytic methods can

2) does not lend itself to performing what if scenarios

3) required that variables be modeled using the normal distribution.

I guessed #2 but the correct answer was #1. The rationale is that MC sims are fairly complex and will provide answers that are no better than the assumptions used and that it cannot provide insights that analytic methods can. I would say that is completely false. I believe that utilizing simulation rather than an OPM or black-scholes is much better, especially with you are dealing at the ends of the spectrum in terms of moneyness of options as BS models break down due to illiquidity.

#2 is, in my opinion, more accurate. As an example, you can run a very easy “what-if” aka sensitivity table in excel for a DCF and sensitize the growth rate and discount rate to determine the PV of CF’s. However, if you wanted to sensitize something in the monte-carlo, you would have to re-run a simulation, and depending on complexity, could take hours to run a single simulation. Any thoughts? 

With computers, MC is run in seconds. Just add /change the various assumptions and boom. But if your assumptions are flawed, so is the result.

125mph that is not entirely accurate. The timing of the MC is based on the complexity of the computation. Our firm has run MC sims that take hours to run a full set of 100,000 trials. I agree with your second statement but that still does not really give a definitive answer yet. 

That being said, in thinking about a sensitivity table with a 5 by 5 interval (i.e. growth changes by 0.5% up 5 intervals and down 5 intervals), it would take the amount of time for 1 simulation multiplied by 11 intervals to run a sensitivity table.