Transcript for:
Monte Carlo Simulation in Finance

Monte Carlo simulation is one of the most  famous and widely applied finance techniques.   This is a tool that helps us deal with  uncertainty in complex situations.   It steps on the premise that one of  the best ways to deal with multiple   uncertain variables is to generate a large  number of random observations for each of them   and hence create a joint  distribution regarding the different   possible outcomes resulting from the  combination of the random variables.   In layman terms, a Monte Carlo simulation  consists in the following series of steps:   1st – determine the variables that  represent a source of uncertainty   2nd – assume a distribution for each variable  (please note that the distribution that will be   chosen is up to the analyst’s discretion);  and of course, the final model predictions   will be as good as the analyst’s choice of  distribution for the unknown variables.   Very good. The 3rd step refers to the following:   once we have chosen the different types of  distributions describing the random variables,   we need to carry out iterations with  possible realizations for these variables.   And 4th – by repeating the 3rd step many, many  times (hundreds, but more likely thousands of   times), we would be able to observe a large  number of possible realization paths that   ultimately indicate what could be expected  in terms of the mean of the distribution;   In practice, Monte Carlo simulations  find a large number of applications   in the world of finance. Some examples are: • Options and other derivatives pricing models   (when complex features are involved) • Company valuation   (to cross-check results obtained through a  DCF, use Monte Carlo to triangulate results)   • And Risk management (Monte Carlo is  frequently used for Value at Risk estimation   and volatility modeling) The large number of repetitions involved   in a Monte Carlo simulation is significantly  facilitated by the strong advents of computing   power we have observed in recent years. Running  a Monte Carlo simulation with a very large   number of observations does not represent a  significant issue for most models nowadays.   This technique is very useful whenever  an analyst would like to gain an idea   of the different possible realizations that could  be obtained through a given distribution function.   For example, if a distribution function considers  three possible variables influencing an asset’s   price, then by testing for different values of  these variables, we could obtain a good idea   about the possible prices of the asset (provided  that the distribution function represents well   the development of the asset’s price). Two main drawbacks should come to mind   when performing this technique. First, as already  mentioned the results to be obtained will only be   as good as the assumed model according to which  the random output variable is distributed.   And second, this is a statistical approach and not  an analytical one. It doesn’t allow us to examine   the impact of some of the separate variables  but instead provides aggregate results.